Cell Culture Bioprocess Engineering

Cell Culture Bioprocess Engineering
Cell Culture
Bioprocess Engineering
Wei-Shou Hu
Department of Chemical Engineering and Material Science
University of Minnesota
Minneapolis, MN
With Contributions From:
Weichang Zhou
Gargi Seth
Sadettin Ozturk
Chun Zhang
Copyright 2012 by Wei-Shou Hu

ISBN: 978-0-9856626-0-8
http://www.cellprocessbook.com/
Preface

For over two decades, we have assembled innovative guest lecturers to share
their research and best-practices at our annual cell culture bioprocessing short course
at the University of Minnesota. This course was created for industrial practitioners of
the production of biologics. This book is the culmination of two decades of accumulated
expertise, practical know-how and insight into future trends.

There have been many books and courses on cell culture technology covering
topics from a technical or business perspective. The goal of this course and this book is to
bring new knowledge from cutting-edge research into the very practical setting of todays
industrial laboratories.

A second goal of this course is to prepare industrial practitioners and students from
different academic disciplines to collaborate in todays cross-disciplinary teams. In the
course of delivering a molecule from a gene sequence in the laboratory to a product in the
manufacturing plant, scientists and engineers must quickly communicate, troubleshoot and
innovate. The fundamental knowledge for practicing industrial cell culture spans from cell
biology and physiology to process engineering principles in stoichiometry, reactor kinetics
and scale up. Thus, we have designed this course for students of diverse backgrounds.

The book is used in the classroom of our annual course. The layout of the book is
thus designed to facilitate the delivery of information. The left panels are graphs, tables,
diagrams, highlights of key points and space for note taking; while the right panels are
descriptive text.

This course has been given around the world: in Europe, East and South Asia, South
America and as an internal course at many corporations. Over three thousand industrial
biotechnologists have taken this course. With the technology of biologics production
spreading to wider regions of the world, this book will meet a timely need of many who
practice the technology but cannot attend the course in Minnesota. The book is published
in an electronic form to allow for more frequent future updates, and for easy distribution to
the parts of the world where the biologics manufacturing is quickly expanding.

Wei-Shou Hu
Acknowledgements

The authoring of this book has been influenced by many who have lectured in the
summer course at the University of Minnesota over the years. Foremost, thanks go to
Anthony J. Sinskey, Michael C. Flickinger, Donald McClure and Fredrick Srienc who started
the course with me originally. Konstantin Konstantinov, James Piret, James N. Thomas,
Randall Kaufman, Florian Wurm, John Aunins, Michael Betenbaugh, Sadettin Ozturk,
Matthew Croughan, Weichang Zhou, Chun Zhang and Gargi Seth all contributed to enrich
the course.

Many former and current members of my research laboratory at the University of
Minnesota contributed to the preparation of course materials. These include Derek Adams,
Marlene Castro, Bhanu Chandra Mulukutla, Anushree Chatterjee, Anna Europa, Patrick Fu,
Chetan Gadgil, Mugdha Gadgil, Anshu Gambhir, Patrick Hossler, Claire Hypolite, Nitya M.
Jacob, Kathryn Johnson, Anne Kantardjieff, Edmund Kao, Anurag Khetan, Rashmi Korke,
Huong Le, Jongchan Lee, Marcela de Leon Gatti, Sarika Mehra, Jason D. Owens, Yonsil Park,
Gargi Seth, Shikha Sharma, Kartik Subramanian, Siguang Sui, Katie Wlaschin, and Kathy
Wong. Gargi Seth, Sadettin Ozturk, Weichang Zhou and Chun Zhang, whose participation in
the course led to the development of new chapters, are noted as contributors.

This book, which began as a set of lecture notes, has gone through many years of
refinement in organization by many skillful hands. Kimberly Durand first took the notes to
digital form in a CD ROM. Ruth Patton, Radha Dalal, Katherine Matthews, Heather Wooten,
Kirsten Keefe, Jessica Raines-Jones, Kimberly Coffee and Kaitlyn Pladson continued to
shape it. At the long last, Erin Fenton and Jenna Novotny took it to current form. Kimberly
Durand also coordinated our final publication efforts.

This book is dedicated to the students, fellows and staff formerly and currently in
my laboratory at the University of Minnesota. It is through working with them that the
materials used in the book were distilled. It was also through their educating me with new
knowledge, new concepts, and new tools that this book took its shape. I must also thank my
dear friend and close colleague, Miranda Yap of Bioprocess Technology Institute, Singapore,
with whom I have had a wonderful and long collaboration.

Finally, I wish for my lovely family, Jenny, Kenny and my wife, Sheau-Ping to share
the joy of the books completion.
Wei-Shou Hu

Contents In Brief
Overview of Cell Culture Technology. . . . . . . . . . . . . . . . . . . . . . . 1
Cell Biology for Bioprocessing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Cell Physiology for Process Engineering. . . . . . . . . . . . . . . . . . . . . 57
Medium Design for Cell Culture Processing. . . . . . . . . . . . . . . . . . 97
Cell Line Development. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
Stoichiometry and Kinetics of Cell Cultivation. . . . . . . . . . . . . . . . 147
Cell Culture Data Analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
Metabolic Flux Analysis in Cell Culture Systems . . . . . . . . . . . . . . 175
Cell Culture Bioreactors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
Oxygen Transfer in Cell Culture Bioreactors . . . . . . . . . . . . . . . . . 213
Fedbatch Culture and Dynamic Nutrient Feeding. . . . . . . . . . . . . 233
Cell Retention and Perfusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
Scaling Up and Scaling Down for Cell Culture Bioreactors. . . . . . 263
Cell Culture Genomics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
285
Index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
309
ACKNOWLEDGEMENTS |VII
Overview of Cell Culture

Technology
Cell Culture Engineering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Cell Culture Products. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Virus Vaccines and Protein Therapeutics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Protein Molecule as Therapeutics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Industrial Cell Lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Biosimilars or Follow-on-Biologics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Manufacturing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Alternative Technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Product Quality and Process Robustness. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Critical Feature of rDNA Proteins from Mammalian Cells . . . . . . . . . . . . . . . . . . 14
Concluding Remarks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Cell Culture Engineering

In the past decade we have seen continuous
growth in mammalian cell culture bioprocessing,
driven primarily by the expansion of therapeutic
antibody production in the pharmaceutical
industry. The range and quantity of products
have both significantly increased over the past ten
years. Also fueling this growth are the increasing
numbers of therapeutic protein candidates in the
drug development pipeline that can potentially
render many more untreatable diseases treatable.
Recombinant therapeutic proteins have yielded major
advances in healthcare. Their societal impacts may
even rival those of antibiotics, whose discovery and
clinical applications transformed much of modern
medicine. Microbial fermentation technology
enabled pharmaceutical industry to make penicillin
widely available between 1950 and 1970. Today, we
see cell culture processing technology enabling this
new class of protein biologics to reach needy patients.
As with any product, manufacturers are under
continual pressure to produce more with less. In
OVERVIEW |1
the case of cell culture bioprocess technology, we

see increasing demand for therapeutic proteins,
coupled with the strain of often prohibitively high
investment costs for new manufacturing facilities.
Thus, we must constantly re-evaluate, streamline
and refine, to increase production without the
luxury of totally new and improved facilities.
Fig. 1.1: Historical trend of penicillin titer and

value
As we look to the future of cell culture processing, it

is useful to look back at the history and development
of penicillin. This specific case highlights lessons that
are almost universally relevant for the manufacturing
of other products. Todays innovations will all travel
through some variation of these phases, from the
moment of discovery, expansion and distribution,
maturation and even demise of the product.
Pencillin is also representative of the many strides
made in the broader field of microbial natural
products that preceded todays protein biologics.
Sir Alexander Flemings discovery of penicillin began

a new chapter in biotechnology. In the twentyfive years following the first clinical applications
(pioneered by Edward Penley Abraham) both
the product titer and the production volume of
penicillin increased almost exponentially. This
rapid expansion in production quantities and
titer was then followed by a period of slower
but steady growth over the next fifty years.
The roughly three orders of magnitude increase
in production volume and product concentration
was the result of relentless effort on the part
of process scientists and engineers. These
engineers looked for hidden opportunities for
strain improvement, media development, and
much more. As a result, we have seen steady
productivity growth due to improvements in oxygen
transfer, heat transfer, and mixing characteristics.
Additional advances in on-line sensing, sterility
control, equipment reliability and process
control all contributed to technological success.
It should be noted that the success of process
technology also eventually drove down the
price. Penicillin G is no longer produced in the
United States; the cost of production is now
OVERVIEW OF CELL CULTURE TECHNOLOGY |2
dramatically lower in other parts of the world.
Now, two decades after the first introduction

of therapeutic biologics, we have seen titers
in large manufacturing processes increase
from tens of milligrams per liter to more than
five grams per liter for many immunoglobulin
products today. Although little published
information is available, the production cost has
also decreased by at least an order of magnitude
since the beginning of cell culture products.
A graph of historical data for cell culture products

plotted one or two decades from now will likely
resemble that of penicillin. Cell culture production
today is likely around the transition from the
exponential growth stage to the steady and
slower growth period. However, it is important
to note that, in terms of both absolute quantity
of product produced and economic value, the
slower and steadier phase is as critical as the
early rapid growth stage for the product life cycle.
Even for penicillin, there were tremendous process

enhancements after the initial rapid growth phase.
Due to these improvements, major medicines
became affordable for the worlds population. The
next question for bioprocess scientists and engineers
is: How can cell culture processing accomplish what
the antibiotics industry has achieved for our society?
Bioprocess scientists and engineers possess
genomics and genome engineering tools that were
not previously available to antibiotic researchers or
even to the early innovators of cell culture processes.
These new genome-wide investigative and
engineering tools will greatly facilitate the designing
and engineering of cells with desired growth and
production characteristics. Process technologists
will need to harness the power of genomics and
genome engineering to enhance productivity and
process robustness. This will also facilitate the
expansion of biosimilars (i.e., Follow-on biologics)
and make many medicines available to needy
patients around the world at an affordable cost.
Much of the process technology employed in cell
culture biologics was developed for antibiotic
Cell Culture Products
production. In transforming cell culture products

from laboratory discovery to clinical reality, many
innovations in the design and engineering of gene
constructs, cells, products, and processes have been
conceived and implemented. These technologies are
also likely to help new technologies move forward.
The next generation technology that will benefit
most from cell culture innovations is stem cell based
therapy. This technology is still in its infancy, but
its significant potential impact on our society will
compel cell culture technologists to push the evelope.
Virus Vaccines and Protein Therapeutics Cell culture processes have been used to produce
viral vaccines for over half a century. Virus
production in animals or in tissues has been in
practice for over two centuries. The most notable
example is the pox vaccine from cow. Most of the
tissue-based production methods have since been
replaced by cell culture processes. A tissue system
that is still in use is the chick egg. This process is
begun by seeding a virus into 10-day-old embryos
in chicken eggs. A few days later, the replicated
virus is then isolated from infected embryos.
Early cell culture processes were an extension
of tissue culture, using primary cells explanted
from various tissues (such as chick embryos
and monkey kidneys) for the virus to infect
and replicate. The primary cells used in virus
production have mostly been replaced by cell
strains or even cell lines, which can be cultivated
over many generations to build up stocks (or a cell
bank) for routine use to ensure consistent quality.
Most viruses used as vaccines have been inactivated

by formalin treatment to render the virus incapable
of infection. However, the treated virus particles
retain a small degree of immunogenicity to elicit
the immune response in vaccine applications.
There are cases in which live attenuated viruses are
used. These attenuated viruses have been adapted,
Table 1. Principal Viral Vaccines Used in Prevention of

Human Virus Diseases
Disease
Source of vaccine
Condition of
virus
Poliomyelitis
Tissue culture (human

diploid cell line, monkey
kidney)
Live
attenuated,
inactivated
Measles
Tissue culture (chicken

embryo)
Live attenuated
Mumps
Tissue culture (chicken

embryo)
Live attenuated
Rubella
Tissue culture (duck

embryo, rabbit, or human Live attenuated
diploid)
Smallpox
(vaccinia)
Lymph from calf or sheep

Live vaccinia
(glycerolated, lyophilized)
Smallpox
(vaccinia)
Chorioallantois, tissue
cultures (lyophilized)
Vaccinia
Yellow fever
Tissue cultures and eggs

(17D strain)
Live attenuated
Influenza
Highly purified subunit

forms of chicken embryo
allantoic fluid
(formalinized UV
irradiated)
Inactivated
Influenza
Cell culture (MDCK, Vero) Attenuated
Rabies
Duck embryo or human

diploid cells
Inactivated
Adenovirus
Human diploid cell

cultures
Live attenuated
Japanese B
encephalitis
Mouse brain
(formalinized), cell
culture
Inactivated
Venezuelan
equine
cephalomyelitis
Guinea pig heart cell

culture
Live attenuated
Eastern equine
Chicken embryo cell

culture
Inactivated
Western equine
Chicken embryo cell

culture
Inactivated
Russian spring summer

encephalitis
Mouse brain
(formalinized)
Inactivated
often by the prolonged cultivation in a non-human

host species so that the adapted strain is no longer
virulent to humans. These viruses are still capable
of replication, which significantly reduces the dose
required for immunization. However, they also carry
a very low, but non-zero, risk of reverting to their
wild type form and causing an infection in the patient.
Vaccine technology predated modern cell culture

for recombinant protein production by over
two decades. Although recombinant therapeutic
proteins propelled the advances in cell culture
technology, proteins derived from tissues, and
even cell culture, were used for therapeutic
purposes even before the arrival of recombinant
DNA technology. These examples include
insulin, urokinase, Factor VIII, and interferon.
The generation of recombinant DNA therapeutic

proteins, such as human growth hormone and
insulin, were first produced in microorganisms.
The next wave were human proteins, which
naturally circulate in human blood and require
post-translational modifications, such as complex
disulfide-bond formation and glycosylation.
These proteins can not be replicated in microbial
systems. For the production of those proteins,
mammalian cells must be, and were, employed.
Initially, hybridoma cells were used. These are

fusion products of the non-antibody-secreting,
but continuously proliferating, myeloma cell
and the antibody-secreting, but non-dividing,
lymphocyte. This soon gave way to recombinant
DNA technology. After the introduction of tissue
plasminogen activation (tPA) by Genentech
in 1987, erythropoietin (EPO) and Factor VIII
also reached the market in following years.
Antibody products and antibody-based fusion
proteins have since blossomed. They make up
the bulk of the protein drugs in clinical use.
Table 2. Therapeutic Protein Biologics Produced in

Non-Mammalian Host
Activity/Use
Granulocyte colonystimulating factor
(Neupogen)
White blood cell growth for

Neutropenia
Insulin (Humulin)
Diabetes
-Interferon (Intron-A)
Anticancer, viral infections
Somatropin [human growth

hormone] (Humatrope)
Growth deficiencies
Somatropin [human growth

hormone] (Protopin/
Nutropin)
Growth deficiencies
Interleukin-2 (Proleukin)
Kidney Cancer
Table 3. Non-Antibody Products Produced in Mammalian Cells

Trade name
Type
Therapeutic Use
Manufacturer
U.S.
approval
year
Host
Aldurazyme
Laronidase
Mucopolysaccharid-eosis I
Genzyme
2006
CHO
Cerezyme
-glucocerebrosidase Gauchers disease
Genzyme
1994
CHO
Myozyme
Fabrazyme
-galactosidase
Pompe disease
Genzyme
2006
CHO
-galactosidase
Fabry disease
Genzyme
2003
CHO
Naglazyme
N-acetylgalactosamie Mucopolysaccharideosis VI
4-sulfatase
BioMarin Pharmaceutical 2005
CHO
Orencia
Ig-CTLA4 fusion
Rheumatoid arthritis
Bristol-Myers Squibb
2005
CHO
Luveris
Luteinizing hormone
Infertility
Serono
2004
CHO
Activase
Tissue plasminogen
activator
Acute myocardial infraction
Genentech
1987
CHO
Epogen/
Procrit
EPO
Anemia
Amgen/Ortho Biotech
1989
CHO
Aranesp
EPO (engineered)
Anemia
Amgen
2001
CHO
Pulmozyme
Deoxyribonuclease I
Cystic fibrosis
Genentech
1993
CHO
Avonex
Interferon-
Relapsing multiple sclerosis
Biogen Idec
1996
CHO
Rebif
Interferon-
Relapsing multiple sclerosis
Serono
2002
CHO
Follistim/
Gonal-F
Follicle stimulating
hormone
Infertility
Serono/NV Organon
1997
CHO
Benefix
Factor IX
Hemophillia A
Wyeth
2000
CHO
Enbrel
TNF receptor fusion
Amgen, Wyeth
1998
CHO
Tenecteplase
Tissue plasminogen
activator
(engineered)
Myocardial infraction
Genentech
2000
CHO
ReFacto
Factor VIII
Hemophilia A
Wyeth
2000
CHO
Advate
Factor VIII
(engineered)
Hemophilia A
Baxter
2003
CHO
Table 4. Therapeutic Antibody Products

Trade name
mAb type
Therapeutic Use
Manufacturer
U.S.
approval
year
Host
Orthoclone
OKT3
Muromomab CD3
Reversal of acute kidney

transplant rejection
Johnson & Johnson
1986 Hybridoma
ReoPro
Anti-Abciximab
Prevention of blood clots
Centocor
1994 SP2/0
Rituxan
Anti-CD20 mAb
Non-Hodgkins lymphoma
Genentech, Biogen
IDEC
1997 CHO
Zenapax
(Daclizumab)
Humanized, anti-subunit T cell IL-2

receptor
Prevention of acute kidney

transplant rejection
Protein Design
Labs
1997 NS0
Simulect
(Basiliximab)
Chimeric, anti-chain T cell IL-2

receptor
Prophylaxis of acute organ

rejection in allogeneic renal
transplantation
Novartis
1998
Synagis
(Palivizumab)
Humanized, anti-A
antigen of RSV
Prophylaxis of lowerrespiratory-tract disease
MedImmune
1998 CHO
Remicade
Anti-TNF- - mAb
Active Crohns disease
Centocor
1998 SP2/0
Herceptin
Anti-HER2 mAb
Metastatic breast cancer
Genentech
1998 CHO
Mylotarg
Anti-CD33
Acute myeloid leukemia
Wyeth
2000 CHO
Campath
Anti-CD52 mAb
Chronic lymphocytic
leukemia
Millennium,
Berlex, Genzyme
2001 CHO
Zevalin
Anti-CD20 murine
mAb
Non-Hodgkins lymphoma
Biogen IDEC
2002 CHO
Humira
Anti-TNF- mAb
Abbott
2002 CHO
Xolair
Humanized, AntiIgE mAb
Moderate/severe asthema
Genentech
2003 CHO
BEXXAR
Anti- CD20 mAb
Follicular non-Hodgkins
lymphoma
GSK
2003 CHO
Raptiva
Anti-CD11a mAb
Chronic psoriasis
Genentech
2003 CHO
Erbitux
Chimeric antibody
raised against
human EGF
receptor
EGF receptorexpressing
metastatic colorectal cancer
Imclone Systems,
Bristol-Myers
Squibb, Merck
2004 CHO
Avastin
Anti-VEGF
Metastatic colorectal cancer

and lung cancer
Genetech
2004 CHO
Soliris
Antibody binding
to C5
Paroxysmal nocturnal
hemoglobinuria
Alexion
2007 NS0
Vectibix
Anti-EGFR mAb
Metastatic colorectal cancer
Amgen
2006 CHO
Protein Molecule as Therapeutics
The early generation of protein therapeutics consisted

of all molecules native to humans. Many antibody
molecules retained part of the sequence of the
immunized species (e.g., mouse or rabbit), although
later generations of antibody molecules were all
humanized or were human antibodies. Some products
are engineered molecules with altered amino acid
sequences that enhance their drug characteristics.
Table 5. Industrial Cell Lines

Major Cell Strains and Lines for Human Biologics Production
Human Vaccines
Primary Cells
Green monkey kidney cells (no

longer used)
Chicken embryo cells
Cell strains
MRC5 (human lung fibroblast)
Cell line
Vero (monkey kidney epithelial

cell), MDCK
Recombinant Proteins
Species cell line derived from
Human
HEK 293, Per C6
Mouse
C-127, NSO, hybridoma cells, SP2/0
Chinese Hamster
CHO
Syrian hamster
BHK
Subsequent products entail fusion proteins, in

which domains (or fragments) of different human
protein molecules are joined. A prominent example
is the fusion molecule of the Fc fragment of IgG and
the TNF binding fragment of the TNF receptor.
This molecule was developed by then Immunex
(now Amgen) for inhibition of TNF to suppress its
inflammatory effect. Another class of product entails
completely foreign proteins, such as recombinant
protein or designer proteins, which have enhanced
potency for eliciting an immune response.
Table 6. Cell Lines Used in the Production of Veterinarian

Vaccines*
Vaccines
Cell line
Bovine viral diarrhea virus
MDBK
Bovine parainfluenza virus type 3
MDBK
Bovine rhinotracheitis virus
MDBK
Bovine respiratory syncytial virus
MDBK
Feline leukemia virus
FL72
Feline panleukopenia virus
CRFK
Feline chlamydia
CRFK
Canine parvovirus
CRFK
Canine distemper
Vero
Canine adenovirus type 2
Vero
Ehrlichia canis
DH82
Rabies
BHK-21
Eastern equine encephalitis virus
Vero
Western equine encephalitis virus
Vero
Equine rotavirus
MA104
Equine rhinopneumonitis virus type 1 and 4 Equine Dermal

Equine influenza virus
MDCK
Foot and mouth disease virus
BHK-21
Swine parvovirus
ST, PK
Swine influenza virus
MDCK
*This table was provided by Terry Ng, 2001. Organisms in

italics are intracellular parasitic bacteria.
Industrial Cell Lines
For the production of traditional viral vaccines,

human diploid cell strains are the primary
production vehicle. Viral products differ from
protein products, in that the viral genome, along with
the entire virus particle, is injected into the patient
to elicit a response. Even though the virus particle is
inactivated by formalin or other treatment, there is
still a potential risk of recombination between the
virus genome and the host cell genome that may
result in the transmission of activated oncogenic
or foreign genetic elements to the patient.
Therefore, the vast majority of virus vaccines
are still produced in normal diploid human cells.
Vero and MDCK cells (along with chick embryos)
are notable exceptions of non-human continuous
cell lines used for human vaccine production.
For veterinary vaccines, the selection of host cells
is vastly wider. Both cell lines and tissue-derived
cell strains with limited life spans are widely used.
For the production of recombinant therapeutic
proteins, the cell lines that are primarily used are of
rodent origin and include mouse, chinese hamster,
and syrian hamster cells. Human cells are only
used for a handful of products. The vast majority is
produced using chinese hamster ovary (CHO) cells.
Biosimilars or Follow-on-Biologics
Two decades after the introduction of mammalian

cell-based therapeutic proteins, many of those
medicines patents have expired. A number of
commercially successful therapeutic proteins will
go off patent between 2013 and 2017, including
the blockbuster drugs Remicade and Humira.
These prospects certainly have helped to draw
in investments to follow-on biologics. Generic
versions of those protein therapeutics have begun
to reach patients throughout the world. The terms
biosimilar or follow-on biologic refer to products
that are marketed after the expiration of patents. They
are expected to have similar properties to existing
biologic products. Sandoz was the first company
to launch a biosimilar-human growth hormone,
Omnitrope, in both Europe and the United States.
Follow-on biologics differ from traditional generic

drugs, in that their biological activity, or the efficacy
of their active ingredient, is not as easily defined as
the traditional chemical and natural product drugs.
Traditional drugs, like penicillin and statins, have very
clearly defined chemical structures that also confer
their biochemical activities. Protein therapeutics, on
the other hand, cannot be entirely characterized by
their chemical composition, or primary sequences.
Therefore, their biological equivalency to their
patented and branded counterparts cannot be
established simply by structural similarity or identity.
Table 7. Approved Biosimilars in the EU

Generic
Name
Product
Launch
Recombinant
human
EPO-
Medice
Arzneimittel
Putter
(Germany)
2007
Binocrit
Recombinant
human
EPO-
Sandoz
(Austria)
2007
Epoetin
alfa Hexal
Recombinant
human
EPO-
Hexal
Biotech
(Germany)
2007
Retacrit
Hospira
Enterprises
2007
Silapo
STADA
Arzneimittel
(Germany)
2007
Somatropin
growth
hormone
Sandoz
(Austria)
2006
Somatropin
growth
hormone
Biopartners
(Germany)
2006
Valtropin
Table 8. Marketed Biosimilars in India

Company
Brand Name
Biosimilar
Launch
Ranbaxy
Ceriton
Epoetin
2003
Dr Reddys
Grastim
G-CSF
2001
Reditux
MabThera
2007
Wosulin
Insulin
2003
Wepox
Epoetin
2001
Biovac-B
Hepatitis B
2000
Insurgen
Insulin
2004
BioMab-EGFR
MabThera
2006
Recosulin
Epoetin
2004
Epofit, Erykine
Epoetin
2005
Neukine
G-CSF
2004
Shanpoietin
Epoetin
2005
Shanferon
IFN 2b
2002
Shankinase
Strptokinase
2004
Shanvac B
Hepatitis B
1997
Wockhardt
Biocon
Intas
Pharmaceuticals
Shantha
Biotechnics
The status of molecular folding, glycan composition,

etc. may affect their activity profoundly. The particular
host cell line that is used, as well as the production
process, may affect subtle aspects of the proteins
properties, thus posing a greater uncertainty
about the quality of the product produced
by manufacturers of those follow-ons. While a
biosimilars approval pathway has been established
in Europe, the U.S. has yet to lay down any guidelines.
Table 9. Marketed Biosimilars in China

Company
Biosimilars
Dragon
Pharmaceuticals
Epoetin, filgrastim
Dongbao
Insulin, G-CSF
Anhui Anke
Biotechnology
HGH, interferon alpha
Amyotop
G-CSF, IL-11
GeneLeuk Biotech
G-CSF, PEG filgrastim,

interferon
HangzhouJiuyan
Gene
G-CSF, IL-11
Manufacturing
Table 10. Dose of Some Antibody Product
Approximate
Formulation
Configuration
Product
Disease
Indication
Company
Amevive
Psoriasis
Biogen
7.5mg / 0.5ml;
15mg / 0.5ml
Enbrel
RA
Amgen
25mg
Heceptin
Breast
Cancer
Genentech
440mg / 30cc
Humira
Rheumatoid
arthritis
Abbott
40mg (1ml
prefilled syringe)
Remicade
Crohns
disease, RA
Johnson &
Johnson,
Centocor
100 mg / 20cc
Rituxan
NHL
Genetech/Idec
100mg / 10cc;
500mg / 50cc
Synagis
Respiratory
syncytial
virus
MedImmune
100mg
Xolair
Allergic
Asthma
Genetech/
Tanox/Novatis
150mg / 5cc
Viral vaccines are administered to patients in

relatively minute quantities because a small amount
of antigen proteins is sufficient to elicit immune
response. Cytokine, growth factors, or enzymetypes of proteins (such as EPO, human growth
hormone or tPA) are also given in small doses, in
terms of protein quantity. Depending on the market
size, the production facility of these products may
be relatively small. The biological effect of antibody
products is largely based on their binding to antigen;
this event requires antibody and antigen molecules to
be in some stoichiometric ratio to elicit downstream
target killing or neutralization. Antibodies are
large molecules, as are many antibody-based
fusion proteins. Thus, many antibody products are
administered in relatively high doses. Thus, the
product vessels, and the size of the manufacturing
plant for antibody products, tend to be larger.
The manufacturing process of protein therapeutics

is rather similar to that for traditional biochemical,
such as antibiotics and E. coli-based recombinant
proteins. A typical process entails a couple of seed
expansion reactor cultures before reaching the
production reactor. The process cycle tends to be
longer. Many cell culture manufacturing processes
are operated in fed-batch modes that last ten to
fifteen days. Some are operated as continuous
perfusion processes and last from two to six months.
The recovery process of cell culture products is
simpler than that for bacterial-based recombinant
Manufacturing Plants

Genentechs Vacaville Facility, California

Started construction in 2004, started operation in
2009. Currently inoperative due to capacity reasons
Investment: $800 million
Eight 25,000-liter bioreactor
Production of Herceptin, Avastin and Rituxan
Bristol Myer Squibb, Devens, Masschusetts
Started construction in 2007, validation in 2011
Investment: $750 million
Six 20,000- liter bioreactors, one purification strain
Production of Orencia and other biologics
Biogen IDEC LSM Facility
245,000 ft2 production
Multi-product facility
Six 15,000L production reactor capacity
proteins. The vast majority of processes now employ

a medium with a relatively low concentration of
proteins, to ease the purification operation. With
the high product concentrations in the range of
5 10 g/L, the product molecule should be the
predominant protein in the medium at the end
of cell culture process. The product isolation
and purification process is substantially simpler
than separating intracellular protein products.
Fig. 2.2: Flow chart of a typical recombinant

antibody production process
Alternative Technologies
Other host cells used for biopharmaceutical production
include E.coli and Sacchromyces cerevisiae. Alternative
production systems include:
Insect cell culture
Yeast ( Pichia )
Transgenic animals
Transgenic plants
Mammalian cells, especially CHO and myeloma cells

such as NS0 and SP1/0, have been the workhorse
for the production of protein therapeutics that
require post-translational modifications (e.g.,
glycosylation, -carboxylation, etc). Although those
post-translational modifications cannot be carried
out in bacterial systems (primarily E. coli), there
are a number of host systems that are capable of
performing N- and O-glycosylation and other posttranslational modifications. They have been explored
as the production vehicles of therapeutic proteins.
Insect Cell Culture

Table 11. Insect Cell Culture
Application
Comments
Basic research
Hundreds of genes have been expressed

using baculovirus.
Bioproduction
Using baculovirus expression systems.
Gene therapy
BV may be used as the gene-delivery vehicle.
Bioreagent
production
A number of bioreagent suppliers use BV to

make target proteins, viral components and
other compounds for the research market.
Yeast
Table 12.
Product
Company
Use
Status
Blood
expander
On the
market in
Japan
Medway
(recombinant
human serum
albumin)
Mitsubishi
Tanabe Pharma
Corporation,
Osaka, Japan
Hepatitis B
vaccine
Shantha
Hepatitis B
Biotechnics Ltd.,
India
On the
market in
India
Interferonalpha
Shantha
Hepatitis C/
Biotechnics Ltd., Cancer
India
On the
market in
India
DX-88
Dyax
Corporation,
Cambridge,
Mass.
Hereditary
angioedema
(HAE), a
debilitating
condition
characterized
by acute
attacks of
inflammation.
BLA
submitted
Recombinant
Human
Insulin
Biocon, India
Diabetes, all
types
On the
market in
India
Recombinant
collagen
Fibrogen Inc.,
South San
Francisco
Medical
research
reagents and
dermal filler
On the
market
Botulism
vaccine
USAMRIID/
DynPort
Botulism
vaccine
product
Phase I
(U.S.)
Antithrombolytic
ThromboGenics
Ltd.
Thrombosis Tx
Phase II
Insect cells were explored as a production vehicle for

therapeutic proteins. The glycoforms of the proteins
produced in insect systems are rather different
from those produced in mammals. Overall, such
efforts have largely subsided. However, for other
applications, such as protein production for toxicity
studies and for veterinary vaccine production, the
insect cell culture remains attractive because the
cultivation is relatively straightforward and the
process development time can be relatively short.
The yeast in the genus Pichia is capable of
synthesizing N-glycans that are not the mannoserich types produced in Saccharomyces. They
have been used in the production of recombinant
proteins, including serum albumin.
Advances
have been made in humanizing the glycosylation
characteristics in Pichia systems for the production
of therapeutic proteins. Glycofi (Merck) has worked
towards a multistep genetic engineering process
where non-human glycosylation enzymes were first
eliminated and human glycosylation reactions were
then introduced. A titer of ~ 1.4 g/L of recombinant
proteins has been reported. With further improved
secretion capacities and glycosylation patterns,
these engineered yeast strains may be capable of
producing proteins with consistent glycosylation
patterns, or even with uniform glycans.
Transgenic organisms for the production of

biotherapeutics have been in development for
two decades. These production systems require
a low initial capital investment and have a
relatively easy purification process for glycosylated
products. However, so far, the FDA has approved
only one product, ATryn, which is produced in
transgenic goats milk by GTC Biotherapeutics.
Transgenic Animals
An advantage of transgenic animal production

is its high titer in milk, on the order of 2 10
g/L. However, over the years, the titer in cell
culture processes has also increased to 5 10
g/L range, thereby diminishing this particular
advantage of transgenic animal production.
Table 13. Transgenic Animal Products Approved or Under Development

Species
Company
Product
Status
Comments
Goat
GTC Biotherapeutics,
MA
ATryn- recombinant human

antithrombin-alpha
Approved
Glycosylation patterns differ slightly ( involves

N-glycolylneuramic acid- not seen in humans),
but was not a regulatory hurdle; Predicted
sales of $6-$10 million in 2009.
Goat
PharmAthene, MD
Protexia- recombinant human

butyrylcholinesterase (BChE)
Development
Rabbit
Pharming,
Netherlands
Rhucin-Recombinant human
C1 esterase inhibitor
Phase 3 trials For the treatment of hereditary angiodema.
mAb
Pilot Studies
Chickens Origen Therapeutics,

(eggs)
Medarex Inc., CA
Functional Mabs produced at 3 mg/egg; some

differences in glycosylation; Half life in mouse
serum half that of natural antibodies (reduced
from 200-100h)
Product Quality and Process Robustness

Critical Feature of rDNA Proteins from
Mammalian Cells
Folding and disulfide bond
Glycosylation
N or O - glycosylation
Sulfation or phosphorylation of glycans
Affect solubility, clearance and biological
activities
Other post-translational modifications
Y-carbonxylation
Lipidation
Phosphorylation
In spite of its dominance as the production vehicle

for therapeutic proteins, the mammalian cell
system does have some shortcomings in its process
characteristics. Compared to microbial systems,
mammalian cell systems have a slow growth rate
and a relatively low achievable cell concentration.
The product titer is also substantially lower than
that of extracellular protein produced using fungal
systems. Finally, the optimal range of growth
environments for mammalian cells is much narrower
than the range for either plant or microbial systems.
After years of research effort, the low productivity
that used to be associated with the low cell and
product concentrations has largely been overcome.
Tissue Plasminogen Activator (tPA)1

Single polypeptide chain (70 kDa) or
proteolytically cleaved at ARG276.
Multiple N-linked carbohydrates: ASN117 (high

mannose), ASN184 (50% complex multiantenary,
50% unoccupied), THR61 (O linked fucose).
Contains 35 cysteine residues, 17 pairs of disulfide
bonds. CYS83 can form a disulfide with other free
thiols depending upon the growth medium and
buffer composition.
May form high molecular weight aggregate
(complexes with protease inhibitors) and
proteolytically cleaved tPA.
Erythropoietin
Contains 40% carbohydrate, only 2 disulfide

bonds.
3 N-linked ASN (24,38,83), 1 O linked (SER126)
glycosylation sites.
O-linked site not essential for in vitro or in vivo
activity.
Sialic acid residues (average 10 moles/mole
Epo) responsible for preserving pharmacokinetic
behavior. Muteins lacking 2 or 3 N-linked sites are
poorly secreted.
N linked glycosylation and sialylation is critical to
optimal secretion, structure, in vivo potency.
Through cell adaptation and media development, the

complex nutritional requirements for mammalian
cell growth have been greatly simplified. Now,
the relatively low tolerance of mammalian cells
to their chemical and physical environment has
not prevented highly stressed conditions from
being used in the final production stage. What has
been lagging is our ability to control the quality,
notably the glycosylation profile, of the product.
The mammalian cell system is chosen for protein

production, almost invariably for its capability of posttranslational modifications on the product (such as
the formation of multiple disulfide bonds of tPA and
the glycosylation of Factor VIII and EPO). Major posttranslational modifications commonly seen in protein
therapeutics, such as disulfide bond formation,
N- and O-glycosylation, and phosphorylation,
all involve extensive enzymatic reactions in the
endoplasmic reticulum or in the Golgi apparatus.
The level of those enzymes, as well as the
supply of precursors and co-factors, affects the
outcome of those reactions. The enzyme levels
vary with cell clone and growth stage, while the
supply of precursors and cofactors change with
the chemical environment. These variations
cause fluctuations in the glycans attached to N(asparagin) sites or to O- (serine or tyosine) sites.
For a given glycoprotein, regardless of whether it

is produced in culture or present in circulation,
the glycans attached to different molecules are not
identical. Rather, they are a mixture of different, but
related, forms. In fact, most glycoproteins in blood
circulation also have hetergeneous glycans. The
structure of glycan and the extent of glycosylation on
the protein molecule affect the blood circulation halflife of the protein. In some cases, the glycan structure
even affects the proteins biological functions. Thus,
confining glycan distribution to an acceptable range
is important for the quality control of the product.
The glycosylation pathway is long and complex,
and takes place in multiple compartments in
the cell. Producing a glycoprotein product with
a defined range of glycan structures throughout
Table 14. In Process Structural Alternations to Mammalian

Protein Biologics
Glycoform
Site occupancy
Altered sialic acid content
Uncapped galactosyl residue, High
mannose
Possibly caused
by stochasticity of
glycosylation process
Glycan distribution out of range

Amino acid alterations in protein
Error rate of amino acid
incorporation during
translation (1/1000)
Mis-incorporation (codon
misreading)
Deamidation (Asparagine)
Loss of terminal amino acid
lysine in C-terminus
of heavy chain IgG,
enzymatic cleavage
cyclization of N terminus
glutamine
Most likely occurred

in culture fluid, may
be affected by process
conditions, or even
product titer
Glycation (addition of reducing

sugar (glucose) to amino acids)
Protein aggregation
May be caused by folding

in ER or agglomeration
in culture or in down
stream processing
a products life cycle is still
a challenge.
Glycosylation may affect the folding of the protein

molecule, but it does not affect its structure. Other
post-translational modifications may affect protein
structure. Failure to form a disulfide bond or the mispairing of a disulfide bond both give rise to an altered
protein structure or the improper cross-linking
(multimer formation) among different molecules.
A lack of -carboxylation or phosphorylation
also drastically changes a proteins properties.
Errors in protein synthesis caused by amino acid misincorporation have been reported. Many production
cell lines have multiple copies of the product gene;
a non-silent mutation (i.e., a mutation causing a
change of an amino acid in the protein) in one of
those genes will inevitably result in the presence
of some fraction of mutated protein molecules.
An alteration of the amino acid structure may
also result from chemical modifications after
being secreted into the medium. After being
secreted into culture medium, the product protein
molecules are also subjected to modification by
enzymes released by cells, which are either actively
secreted or released from lysed cells. Extracellular
proteolytic cleavage can give rise to degradation of
Factor VIII, or can alter the ratio of single chain/
double chain molecules of tPA and Protein C. Also,
the sialic acid moiety in glycans may be cleaved
by sialidase released from lysed cells. Table 14
summarizes some more commonly-seen alterations
in protein molecules in cell culture processes.
In the past decade we have seen the productivity

of recombinant cells reaching or even exceeding
the production rate of professional secretors in our
body (such as liver cells or antibody- and insulinsecreting cells). We have also seen the product titer
in the bioreactor approaching the concentrations
of antibody in ascites fluid. As the productivity and
product concentration of cell culture processes
approaching its natural biological counter
parts, we must also be cautious and ask ourselves
whther we are pushing cells protein folding
and processing machinery to operate at its limit.
The unprecedented high productivity is achieved by

operating the reactor at conditions that are neither
optimal for growth nor the natural homeostatic
state. Rather, they are often highly stressed to favor
producing the product of our design. In todays
production cultures, both the intracellular and the
extracellular environments are extremely harsh.
While protein synthesis and secretion is boosted to
a nearly unprecedented level, the cellular machinery
for protein quality control may not be operated at the
same level of stringency as it is for optimal growth.
Process technologists must bear in mind that quality
consistency and process robustness must be the
highest priority when pushing productivity higher.
Concluding Remarks
Over half a century, cell culture processes have
evolved from tissue and small-scale cell culture for
vaccine production to large scale manufacturing
process for protein production.
Therapeutic
proteins, especially antibody and antibodybased proteins, are the dominant products. The
continuing pressure to meet increasing demand for
products has led to many process innovations and
refinements over the past two decades. The cell
and product concentrations in todays process are
nearly two orders of magnitude higher than they
were at the dawn of the recombinant protein era.
The success of this technology has also shifted the
focus from production quantity to product quality.
Cell culture processes now aim to provide optimal

growth conditions for cell expansion, while often
employing highly-stressed conditions for the final
production stage. All must be accomplished without
compromising the quality of product produced.
Achieving those aims through process innovation will
be critical in the next phase of the technology, wherein
follow-ons or biosimilars will have an increasing
presence. Cell culture engineering efforts in the
past quarter century have transformed bioprocess
technology. The advances made in cell culture
technology will greatly facilitate the development
of the emerging stem cell and other cell therapy.
Cell Biology for Bioprocessing

Cells: Source, Composition and Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Cell Source. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Cell Composition and Chemical Environment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Cell Membrane. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
Cytoplasm and Organelles. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Transport Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Major Mechanism of Transport. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Extracellular Matrices and Cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Movement. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Growth, Death and Senescence. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Cell Cycle and Growth Control. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Apoptosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Senescence and Telomeres. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Cells: Source, Composition and Structure

Cell Source
Table 1. Cells Commonly Used in Bioprocessing
Species
W1 - 38
MRC - 5
FS - 4
HEK 293
Vero
MDCK
NS/SP2/0
CHO
BHK
Human
Human
Human
Human
Monkey
Dog
Mouse
Chinese
Hamster
Syrian
Hamster
Fibroblast
Fibroblast
Fibroblast
Epithelial
Epithelial
Epithelial
Lymphoid
Tissue
Isolated
Lung
Lung
Foreskin
Kidney
Kidney
Kidney
Myeloma
Epithelial
Ovary
Epithelial
Kidney
Cell Type
The cells commonly used for the production of

biologics are derived from different tissues of
different species. Thus, they can vary widely at
the genomic level. Their differences are even
visible microscopically, with various numbers of
chromosomes. However, at a physiological and
transcriptome level, cells from the same tissue of
different species are strikingly similar. Their similarity
is much greater than different cell types from the same
animal. For example, chicken embryo fibroblasts look
morphologically very similar to human fibroblasts
from the lung or foreskin, while the epithelial MDCK
cells look rather different from dog fibroblasts even
though they are both derived from the same species.
CELL BIOLOGY |19
CELL ENGINEERING|19
Among the ~200 different types of cells, fibroblasts,

epithelial cells and myeloma cells are most frequently
used cell types in biologics production
Cells in culture bare closer physiological and
morphological characteristics of the tissue they were
derived from than the species
Cells in vivo may be in a quiescent state or in a
proliferative state, but are all adapted to rapid
proliferation in culture
Even though there are about two hundred types of

cells in a vertebrate animal, most cells that are used
for the production of biologics are either epithelial
or fibroblast in nature. These two cell types are
more amenable to isolation from tissues and to
in vitro culture, as demonstrated during the early
explorations on tissue cell isolation more than half a
century ago. NSO and CHO are the two prominent host
cell lines used for therapeutic recombinant protein
production. They exhibit different behaviors and were
derived from two different tissues and two different
species. CHO cells were isolated from the ovary of
a Chinese hamster; NSO cells were isolated from a
mouse myeloma. Cells used for recombinant protein
production are primarily epithelial and lymphatic.
Both fibroblasts and epithelial cells are frequently

used for viral vaccine production. These cells differ in
both their functions and tissue locations. Epithelial
cells line the boundary of tissues, while fibroblasts
make up a larger part of the connective tissue.
Epithelial cells form tightly connected sheets, which
often get damaged, die, and are replenished by new
ones. Thus, many of them are constantly growing in
vivo. Conversely, fibroblasts are mostly quiescent.
They migrate into wounds and begin to grow only
when they are stimulated by various cues. Lymphatic
cells, especially the terminally-differentiated plasma
cells (from which myeloma cells are derived), are
needed to secrete antibodies against a particular
antigen, but only for a limited period of time after
the hosts exposure to the antigen. They undergo
apoptosis days after their differentiation into active
antibody-secreting cells, so that the host does not
continue to have unnecessary or maybe even harmful
antibody molecules in circulation. Such native
characteristics are often still evident in culture.
CELL
BIOLOGY
|20
20 |CELL
BIOLOGY
Cell Composition and Chemical

Environment
Table 2. Typical Composition of a Cell
E. coli
Mammalian Cell
pg / Cell
Wet weight
Dry weight
Protein
Carbohydrate
Lipid
DNA
RNA
Water
Range
3,000
3,000 8,000
600
250
150
120
10
25
300 1,200
200 300
40 200
100 200
8 17
20 40
Volume
4 x 109
cm3
Diameter
18 m
1020
15
12
0.3
0.7
80 85
15
2
2
1
6
70
0.5-2 m
Table 3. Cellular and Extracellular Fluids Ion

Concentration
Plasma
(mmole / l) Interstitial (mmole / l)
Intracellular osmolality
(mmole / l)
140
14
K+
140
Ca++
10-4
Mg++
0.8
0.7
20
Cl
110
110
50
Na+
>10 fold concentation difference for K+ and Na+ across

the plasma membrane
Opposite direction of concentration gradient for K+ and
Na+
Extremely low concentration of Mg++ in intracellular
fluid
Total osmolality 280 mOsm
Most cells in culture have a diameter of about 12 18

m. Some types of stem cells are rather small and have
only a small amount of cytoplasm. In contrast, liver
cells (i.e., hepatocytes) in some species are rather large,
with an average cellular diameter of 20 m. A typical
cell has nearly 80% of its mass as water. Proteins make
up the next largest portion of cell mass, after water.
Other than water and proteins, the other cellular

constituents are present in much smaller amounts
and rarely exceed 10% of the total dry mass. Lipids
make up various membranes of the cell, including the
cytoplasmic membrane and the membrane enclosing
all organelles. Lipids, thus, constitute a significant
portion (about 5-8% of total dry mass) of cell mass.
Carbohydrates (such as glycogen) are used to store

energy in some cells. However, not all cells have a
large amount of free carbohydrates. Carbohydrate
molecules that serve as energy sources are quickly
metabolized to become intermediates in energy
metabolism.
Most carbohydrate moieties that
remain in their carbohydrate forms exist as part of
nucleotides or are conjugated to proteins or lipids.
The size of a haploid genome in a typical mammalian
cell is about 3 Gbp. That equates to about 5 pg of
DNA for a diploid cell. However, DNA is not the
most abundant nucleic acid in the cell. RNAs are
far more abundant than DNA in a cell and include
messenger RNA (mRNA), ribosomal RNA (rRNA),
and others. Ribosomal RNA, which is a major
constituent of the cells protein synthesis machinery,
constitutes over 90% of all RNA in the cell.
Since water constitutes the largest fraction of all cell

materials, the chemical species that are present at a
high concentration in the cytosol are also major cellular
constituents. Combined, all minerals contribute
a significant (~5%) proportion of the dry mass.
The concentrations of some ions are vastly different

inside the cell versus outside the cell. Maintaining these
concentration gradients is critical for cell functions.
The concentration ratio between intracellular and
extracellular K+ and Na+ is in the range of 15 to
CELL BIOLOGY |21
CELL ENGINEERING|21
30. Conversely, their direction of concentration

gradient is opposite: the concentration of K+ and
Na+ should be far higher inside and outside the cell,
respectively. The solutes in a solution exert osmotic
pressure, which is typically quantified by osmolality.
The osmolality of cellular fluid is about 280 mM (or
mOsm). A typical medium has its osmolality at the
same level as in cells, to avoid incurring osmotic stress.
Cell Membrane
Lipid Bilayer Composition
Phospholipids
Constitute the majority (35-70%)
Glycolipids
Neutral glycolipids (e.g. galactocerebroside)
Gangliosides
Have sialic acids
Four types of phospholipids
Three have glycerol as backbone,Phosphotidyl
ethanol amine, Phosphotidyl serine and
Phosphotidyl choline

Serine as backbone
Cultured mammalian cells have long been thought

as being extremely fragile to mechanical stresses
because their cellular materials are surrounded only
by cytoplasmic membrane; the only thing preventing
the cellular content from dissolving into the aqueous
environment is that lipid bilayer. Yet in a modern
manufacturing plant, these tiny cells thrive in
bioreactors of tens of cubic meters in volume under
such highly turbulent conditions. The membrane
surrounding a cell is not merely a double-layer of
lipids, and the integrity of a cell is not merely dictated
by its membrane wrapping.
The lipids which make up the lipid bilayer are
amphipathic. They have a hydrophilic head group,
and a hydrophobic tail group made of fatty acids.
When suspended in an aqueous solution, amphipathic
molecules can form micelles. In such micelles, the
hydrophilic.
Fig.2.1: A phospholipid molecule with glycerol as backbone,

with an ethanol amine, a saturated and an unsaturatted
fatty acid.
CELL
BIOLOGY
|22
22 |CELL
BIOLOGY
Lipid Bilayer
Characteristics of a Lipid Bilayer

The lipid bilayer is a fluid
As temperature decreases, the bilayer transitions
from a fluid state to a gel state
The degree of fatty acid unsaturation affects
the transition temperature of membrane from a fluid
state to a gel state
The magnitude of diffusion of various solutes in the
cell membrane resembles that of a liquid
A lipid bilayer membrane behaves like a fluid. If the

lipid molecules in a specific location are labeled with
a fluorescent dye, the fluorescence disperses shortly
thereafter due to molecular diffusion (instead of
staying in the same place as in a solid). The lateral
diffusion coefficient of a phospholipid molecule in a
bilayer membrane is about 10-8 cm2/s. A lipid molecule
does not flip-flop (or change its side of a lipid bilayer)
without the aid of membrane-bound phospholipid
translocator. Gas species diffuse about equally fast in
a lipid bilayer as they do in water. Even large protein
molecules diffuse in a lipid bilayer membrane.
Fatty acids make up the hydrophobic tail. At very

mild temperatures these acids undergo a phase
transition from a fluid to an ordered structure. Thus,
lipid bilayers also undergo phase transition to form a
liquid crystal at a relatively moderate temperature.
This tightly-packed, ordered structure acts as a very
good barrier to keep most molecules from freely
passing in or out of the cell. The permeability of
most biological molecules across a lipid bilayer
membrane is rather low.
Even the smallest
nutrient, such as glucose and simple amino acids,
cannot pass by fast enough to support cell growth.
Fig. 2.2: Lipid bilayer membrane at a crystaline state and

fluid state
All major biological macromolecules (e.g., DNA,

proteins, and polysaccharides) are biopolymers
made of covalently-bonded monomers. A lipid
bilayer membrane is a not a polymer, rather, it is
an assembly of phospholipids. The non-covalent
nature of phospholipids within the cell membrane
allows it to be very dynamic: expanding, shrinking,
breaking, and fusing rapidly. The lipid bilayer also
envelops various organelles to compartmentalize
regions in the cell for specialized functions. Many of
those organelles are in a constant dynamic process
of membrane budding and fusion. For example,
in trafficking between organelles and in protein
secretion, the cargo is carried inside membrane
vesicles while transiting from one organelle to
another. This process occurs without the need to break
up and re-form a larger number of covalent bonds.
Three types of lipids make up a lipid bilayer
membranes in cells and organelles: phospholipids,
CELL BIOLOGY |23
CELL ENGINEERING|23
Characteristics:
One saturated, one cis-unsaturated (C14-C24)
typically constitute the tail of the phospholipid.
Fatty Acids (the tail group) on the lipid affect the

packing of lipids in bilayer membrane. Saturated
fatty acids allow more dense packing; double
bonds in unsaturated fatty acids creates kinks,
reduce packing, increase fluidity.
Cholesterol has a small head polar group linked to
a rigid planar region of steroid rings followed by
a more flexible non-polar tail. They interact with
phospholipids to stabilize the region closer to the
head group as well as to make the lipid bilayer
less inclined to become crystalline. Overall, they
increase the membrane permeability to small
compounds, and make the membrane less fluid.
Depending on temperature and the degree of
hydration, lipid bilayer is in gel state or in liquid
crystalline state. The temperature of bilayer
phase transition from the crystalline lipid bilayers
to fluid bilayers is affected by fatty acid and
cholesterol composition.
Cholesterol in a Lipid Bilayer
Fig. 2.3: Schematic drawing of a cholesterol molecule

interacting with two phospholipid molecules in one leaflet of
a lipid bilayer
glycolipids, and gangliosides (phospholipids being

the most common). There are also different types
of phospholipids, with either glycerol or serine
as the backbone, with the former being the most
abundant type. Glycerol has three hydroxyl groups
attached to its three carbons and one of them has
a phosphate group, to which an ethanolamine or
serine is attached. The phosphate moiety has a
strong negative charge, thus making this end of
the molecule the highly hydrophilic head group.
The other two hydroxyl groups of glycerol are linked

to two fatty acids through an ester bond. Typically,
one of those two fatty acids is saturated and the other
is unsaturated, with a cis double bond in-between
C14 and C24. The degree of unsaturation affects
the packing of the lipid bilayer. Saturated fatty acids
allow more dense packing, while the double bonds
in the unsaturated fatty acids create kinks, which
reduce packing and increase the membrane fluidity.
A lipid bilayer can be in a gel state or in a liquid

crystalline state depending on the temperature
and degree of hydration. A lipid bilayers phase
transition temperature is affected by its composition
of fatty acid and cholesterol. As temperature
decreases, the lipid bilayer changes from a liquidcrystalline state to crystalline (or gel) state. A
higher content of shorter, unsaturated fatty
acids increases the fluidity of the lipid bilayer
and decreases its phase transition temperature.
Another molecule playing a key role in the membrane
properties of animal cells is cholesterol. Cholesterol
has a small polar head group linked to a rigid planar
region of steroid rings that are further linked to a
more flexible non-polar tail. Cholesterol interacts
with phospholipids to stabilize the region closer to the
head group and to make the lipid bilayer less inclined
to become crystalline. Overall, cholesterol increases
the membrane permeability to small compounds
and makes the membrane less fluid. Cholesterol
content varies in different lipid bilayer membranes.
Its level in the cytoplasmic membrane is higher, but
CELL
BIOLOGY
|24
24 |CELL
BIOLOGY
Membrane Proteins
in the membrane of many organelles it is very low.
A typical biological membrane has ~50% proteins by

mass; in terms of molecules, lipid:protein = 50:1
Metabolically active mitochondrion has 75% protein in
its membrane.
Na+/K+ ATPase acts as a pump, using ATP to pump 3Na+
out and 2K+ into the cell.
The electric protential across the plasma membrane is
about -80mV.
Table 4. Biochemical Composition of Hepatocyte Plasma

Membrane
Total
Lipids
Total
Protein
Protein/
Lipid
mass
ratio
Cholesterol/
Phospholipid
molar ratio
Cholesterol
in total lipids
30-40%
50 -60%
1-2
0.4 - 0.8
12 - 20%
(by
(by mass)
mass)
Adapted from The Liver: Biology & Pathology, 4th Ed., p. 78 (2001)
Phospholipids in
total lipids
50 - 70%
A typical biological membrane has ~50% lipids and

~50% proteins, by mass. In terms of molecules,
however, the lipid:protein ratio is actually about
50:1, since proteins have much higher molecular
weights than lipids. The protein content of a
membrane is greatly affected by the tissue of origin
and by the membranes function in the cell. The
mitochondrial membrane, through which many
molecules (e.g., amino acids, pyruvate, various ions
and many other proteins) pass at a high flux, has a
high protein content of about 75%, by mass. On the
other hand, the myelin membrane, which serves as
a protective sheath between the nerve cell and its
surroundings, has a low protein content of about 25%.
Lipid bilayer membranes separate cellular
content from their surroundings and divide the
organelles from the cytosol. Not only do they
create a barrier for the physical retention of a
cells contents, but they create a rather different
chemical environment across membranes. For
example, cells maintain about an 80 mV electric
potential across the plasma membrane and about
140 mV across the mitochondrial membrane. The
ER membrane separates an oxidative environment
(inside the ER) from a reduced one (in the cytosol).
The maintenance of various chemical, electrical, and

redox potentials across a membrane is accomplished
by various membrane proteins. Rat small intestinal
enterocyte has about 150,000 Na+ pumps per cell,
which collectively allow each cell to transport about
4.5 billion Na+ ions out of the cell, each minute.
The sodium and potassium membrane gradients
generated by those pumps, as well as the electric
potential across cytoplasmic and mitochondrial
membranes, are fundamental to cellular bioenergetics.
CELL BIOLOGY |25
CELL ENGINEERING|25
Membrane Dynamics
Cellular membrane is in a dynamic state

contributed by:
Lipid turn-over
Inter-organelle shift of membrane vesicles
Secretion, endocytosis
Homeostasis of cellular membranes

Professional secretory cells in the body can add
0.5% per minute of their plasma membrane due
to the fusion of secretory vesicles with the plasma
membrane; they must be recycled to maintain a
balance
Phospholipids in the membrane are subject to turnover
The cellular membrane is in a dynamic state;

membrane constituents are continuously being
added and removed. This is not only for membrane
expansion and cell growth, but also for turnover and
for vesicle trafficking. Like other cellular components,
the turnover of the cellular membrane is necessary
to replace lipid molecules that have been oxidized or
damaged, or to allow cells to change their membrane
composition to adapt to new environments.
The turnover rate of a cell membrane varies widely.
Phospholipids are said to have a half-life of three
hours, while the half-life of cholesterol is about
two hours. Cellular membrane proteins are also
turned over. Their half-life ranges from a couple of
minutes to a couple of days, whereas macrophage
membrane proteins are turned extremely rapidly.
Inter-organelle trafficking and the secretion of proteins

into the extracellular environment also contribute to
a membranes dynamic state. Protein molecules that
are destined for export are carried from organelles to
the cytoplasmic membrane by vesicles. Upon reaching
the inner surface of the cytoplasmic membrane,
those vesicles fuse with the cytoplasmic membrane
and release their contents outside of the cell.
In the liver, each hepatocyte synthesizes ~120 x 103

albumin molecules per min (translating to about 15
pg/cell/day). All of those molecules are wrapped in
280 400 nm of vesicles and delivered to the basal
plasma membrane of the cell. The infusion of those
membrane vesicles would cause the membrane
surface to expand at a rate of 0.5%/min. However,
since hepatocytes are typically in a G0 state (i.e., not
dividing), the size of their cytoplasmic membrane
does not need to increase to accommodate cell growth.
Therefore, the lipid molecules that are added to the
cytoplasmic membrane must be recycled back into
the intracellular organelle (Golgi bodies) to maintain
the cytoplasmic membrane in a homeostatic state.
Similarly, cells active in endocytosis can internalize

up to 0.8%/min of a plasma membrane. The loss of
lipids from membrane caused by endocytosis must be
replenished to maintain the size of cells outer envelope.
CELL
BIOLOGY
|26
26 |CELL
BIOLOGY
Cytoplasm and Organelles

Total protein concentration in
cytoplasm
150 g / L (~4 M)
Total protein concentration in

plasma
90 g / L (1.2 M)
Albumin (MW 69,000)
45 g / L (0.65 M)
Globulins (MW 140,000)
25 g / L (0.18 M)
Fibrinogen (MW 400,00)
103 g / L (0.0075 M)
The cytoplasm and nucleus are both enclosed by

cytoplasmic membrane in the cell. The cytoplasm
can be largely divided into two groups: various
organelles and the highly-viscous cytosol. The
cytosol has a very high concentration of proteins (100
300 mg/mL). For comparison, the protein content
in blood plasma is only 90 mg/mL. The cytosol also
contains the inorganic solutes, building blocks, and
intermediates and metabolites of metabolic reactions.
The cytosol is not only full of soluble components.

It also contains large assemblies (or aggregates) of
particles. The ribosome is the main machinery for
making proteins; it is a complex particle consisting
of many ribosomal proteins and ribosomal RNAs
(rRNA). Each cell contains thousands of ribosomes
of ~30 nm in size. Many ribosomes are located on
the cytosolic surface of the endoplasmic membrane
and appear as a black spot, when viewed under an
electron microscope. Some enzymes also form large
complexes that can be seen under electron microscope,
such as pyruvate dehydrogenase complexes.
Cytoplasm is not a simple solution

Some protein complexes (like pyruvate dehydrogenase
and ribosomes) are aggregated
Cytoskeletal network is interspersed in cytosol
Also rich in the cytosol are the fiber-like structures of

the cytoskeleton. These large protein particles, enzyme
complexes, cytoskeletal proteins, and organelles
make the cytoplasm of a cell very crowded and render
its solution phase very dense in mass. Under light
microscopy, an animal cell appears to be primarily
cytoplasm, wrapped in a membrane, with a nucleus
sitting near the center spanning over half of the cells
diameter. Other than the nucleus, various organelles
include the mitochondria, the endoplasmic reticulum,
the Golgi apparatus, peroxisomes, endosomes,
etc., and are visible only by electron microscopy.
CELL BIOLOGY |27
CELL ENGINEERING|27
Nucleus
endocytosis
endosome
lysosome
rough
endoplasmic
reticulum
golgi
apparatus
secretory
vessel
smooth
endoplasmic
reticulum
chromatin
plasma
membrane
nuclear
envelope
(nuclear
membrane)
nucleus
nucleolus
mitochondria
Fig. 2.4: Organelles in an animal cell
In bacterial cells, DNA molecules are roughly

localized at the center of the cell. Both DNA and
RNA synthesis occur in the nucleoid region that is
adjacent to the chromosome formed by the large DNA
molecule. The nucleoid occupies a distinctive part
of the cytoplasm and the DNA is tightly coiled and is
bound by many proteins. If completely extended, a
DNA molecule of an E. coli cell is nearly 1-mm long.
In contrast, a eukaryotic cells genome is separated
into a number of DNA molecules, which each
form a chromosome. Then, the DNA molecules
are segregated into nuclear compartments. The
average genome of a mammalian cell is about three
orders of magnitude larger than that of E. coli. If
stretched, it extends to about 1-m in length. This
large amount of DNA is packed into a small space
by forming DNA-protein (histone) complexes.
DNA/RNA synthesis and ribosome assembly occur

in the nuclear compartment and are segregated
from the metabolic processes and protein synthesis
in the cytoplasm. Ribosomes are assembled in
nucleoli and are subsequently exported into the
cytoplasm to participate in protein synthesis. The
complex tasks of sorting out which segments of DNA,
or which genes, are to be transcribed into RNA at a
given moment occur in the nucleus. A large array
of transcription factors and other transcription
regulators are synthesized in the cytoplasm and then
imported into the nucleus where they bind to specific
genetic loci to perform their role in transcription.
Thus, there is a large volume of material trafficking

between the nucleus and the cytoplasm. Components
of the ribosome, nucleotides/deoxynucleotides,
nuclear structural proteins, and transcription
factors need to be imported into the nucleus. The
RNAs (mRNA, tRNA, and some non-coding RNA)
are exported into the cytosol for protein synthesis.
A double-layered membrane separates the
cytosol and the nucleoplasm. The nucleus and the
mitochondrion are two organelles in the cell that
have double membranes, instead of only a lipidbilayer membrane. Much of the trafficking occurs
CELL
BIOLOGY
|28
28 |CELL
BIOLOGY
through nuclear pores on the surface of the nuclear

membrane (also known as nuclear envelope).
Mitochondria
Mitochondria are....
The most abundant organelle in a cell
(about 1,700 per cell)
Take up to 20% of cell volume
In the catabolism of glucose to carbon
dioxide, the oxygen atom in CO2 is
contributed from water molecules. The
oxygen reacts with H in NADH, FADH2, to
form water in mitochondria
Active mitochondrion has a negative
140 mV electric potential across its inner
membrane, and 1.0 units of pH gradient
(inside mitochondria pH is higher [H+
concentration is lower] and pH is pumped
against concentration gradient)
The membrane potential cannot be charged
up too much. Therfore the homeostasis of
mitochondria is critical.
Cells meet long-term energetic needs by
biogenesis of mitochondria.
The mitochondrian is the most common organelle

in a cell. With about 1,700 per cell, they make
up 20% of the cells volume. Mitochondria are
about the size of bacteria and are thought to
have originated from bacteria-like structures
that were acquired by primitive eukaryotes.
Mitochondria serve as the cells power plants. The
most reactive reactions in the cell (e.g., oxidizing
nutrients and generating energy through electron
transfer and oxidative phosphorylation) take
place in the mitochondria. Cells with different
energy needs have different numbers of these
power plants. In a high-energy demanding cell,
there can be as many as 3,000 mitochondria.
The main ATP-generating process occurs via electron

transfer, across the inner mitochondrial membrane.
The total surface area of all mitochondrial inner
membranes in a cell is greater than that of the
cytoplasmic membrane. At the mitochondrial inner
membrane, reactive electrons in electron transfer
react with oxygen to form H2O. Mitochondria are thus
rich in potentially damaging free radical species. By
confining these reactions to the mitochondria, the cell
can potentially reduce unintended cellular damage.
The mitochondrion resembles a bacterium, not

only in size but also by having its own genome
in the form of a circular DNA molecule. Each
mammalian mitochondrion contains one or
more mitochondrial genomes of about 18 kbp.
The control of mitochondrial DNA replication
is separate from the regulation of genomic DNA
replication. The biogenesis (i.e., the replication)
of mitochondria is independent of cell division.
An active respiring mitochondrion has a negative

140 mV electric potential and pH of 1.0 across its
inner membrane. The pH inside a mitochondrion is
higher, as the H+ ion concentration is lower inside,
so pH is pumped against the concentration gradient.
The pH gradient and the electric potential are critical
CELL BIOLOGY |29
CELL ENGINEERING|29
for cells. The electric potential and pH gradient are

created by pumping protons out of mitochondria.
This occurs while transferring electrons at a high
energetic state in NADH to a low energetic state
that can be received by oxygen to form water. In
other words, the chemical potential energy in the
high energetic electron is transformed and stored
in the electric potential and proton gradient.
The membrane potential cannot become too high; it

can burst the organelle. It is important to maintain
a homeostatic condition in mitochondria. When
the energetic need of a cell is high over a long
period, cells respond by increasing their number
of mitochondria. The flux of energy (primarily
pyruvate) into mitochondria is tightly controlled.
Fig. 2.5: Proton and electric potential (charge) gradient
across mitochondrial inner membrane. The direction of
fluxes of major species are indicated by and arrow. Note
NADH oxidation coupled electron transfer pumps protons
against proton and charge gradient, while the movement
of proton to drive ATP synthesis is in the direction of
proton and charge gradient.
Mitochondria (along with other organelles) cannot

be generated merely from the genetic content in the
nucleus. A cell must have mitochondria at its origin in
order to make more mitochondria as it proliferates.
The mitochondrial genome encodes a number of
mitochondrial proteins and RNA molecules, while other
mitochondrial components are encoded by cellular
genomic DNA and imported into the mitochondria.
CELL
BIOLOGY
|30
30 |CELL
BIOLOGY
Endoplasmic Reticulum
Smooth ER
Function varies with tissue, in liver cells it detoxifies; in
ovary and testes it makes hormones
Rough ER
Proteins for some organelles, integral membrane
proteins and secreted proteins are folded in ER
Professional secretors in the body, such as pancreatic
beta cell, hepatocyte and antibody secreting plasma
cell, all have abundant ER. As B cells differentiate to
become plasma cell, ER and Golgi apparatus expand
drastically, at least by 15 fold
Some characteristics of ER
In hepatocyte, surface of ER is about 63,000 m2 per
cell, or about 40 times of plasma membrane
ER lumen very viscous, gel-like. Diffusion coefficient of
fluorescent probe is 9 - 18 times lower than in water
ER has much higher oxidative environment than
cytoplasm, appropriate for disulfide bond formation
High free Ca2+ environment
Many proteins are present at very high concentrations
(PDI, GRP94, GRP74)
A major site of protein folding, other post-translational
processing, Ca++ homeostasis, cholesterol synthesis
Protein processing occurs in ER

Cleavage of signal peptide
Addition of high mannose core oligosaccharide to Asnx-Ser / Thr N-linked glycosylation site
Trimming of terminal glucose and mannose residues
from initial glycan
Fatty acid addition
The endoplasmic reticulum (ER) is largely classified

into the smooth ER and the rough ER, based on
morphology. The smooth ER is rich in enzymes
involved in chemical transformation reactions.
In liver cells, the smooth ER takes on the role of
detoxification; in the ovaries and testes, it makes
hormones. The rough ER is the site of folding and
processing of proteins destined for some organelles,
integral membranes, and secretion. It gains its name
through the attachment of a large number of ribosomes
to its cytosolic domain surface, thus appearing to be
rough under the transmission electron microscope.
Professional secretors in the body have abundant
ER, such as the pancreatic beta cells that secrete
insulin and the antibody-secreting plasma cells.
As B cells (non-antibody secreting) differentiate
to become plasma cells, the ER and Golgi
apparatus expand drastically, more than 15 fold.
Hepatocytes secrete many proteins, including albumin,

which are coagulation factors that constitute many of
the bloods protein components. In the liver, some
hepatocytes specialize in protein secretion, while
others play major roles in oxidative detoxification.
These hepatocytes have distinctive ERs. Those
involved in protein secretion have an abundance of
rough ER, while those more specialized in xenobiotic
metabolism have an abundance of smooth ER.
ER is also a major site of protein post-translational

modifications, and is involved in Ca+2 homeostasis
and cholesterol synthesis. The ER lumen is rich in
proteins that facilitate protein folding and catalyze
the formation of intermolecular disulfide bonds.
Disulfide bond formation
CELL BIOLOGY |31
CELL ENGINEERING|31
The classical view of the Golgi apparatus is as a

stack of flattened sacks. More recently, the Golgi
apparatus is viewed as a dynamic region where many
key reactions occur. Many proteins are modified
in Golgi bodies after they are folded in the ER.
Golgi Apparatus and Protein PostTranslational Modification
Protein Processing Occurs in Golgi

Addition glycoform modification
Sulfation of tyrosines or carbohydrates
glycosylation
Peptide proteolytic cleavage
Gamma-carboxylation of glutamic acid
Beta-hydroxylation of aspartic acid
Protein Secretion Through ER and Golgi

Apparatus
Some Characteristics of Golgi

Compartmentalized into functionally distinct regions:
Golgi stack (consisting of cis, medial and trans
cisternae), and trans Golgi Network (TGN)
Proteins, lipids are sorted in Golgi for delivery to
different cellular locations
From here proteins go to secretion (exocytosis) or
other organelles
During mitosis the Golgi apparatus breaks down and
reassembles after mitosis
Different molecules of the same secretory protein
spend different amounts of time inside the cell, i.e.
there is a distribution of holding time in the cell
Translation of a protein molecules takes only seconds,
but the secretion process takes tens of minutes to
hours
The Golgi apparatus is loosely divided into four

compartments: cis, medial, trans, and the transGolgi network (TGN). The protein cargo from the
ER is transferred to the cis Golgi and then to other
Golgi compartments through membrane vesicles.
The enzymes in the four Golgi compartments
are not identical. Thus, different reactions
may take place in different compartments.
For a high-producing industrial cell line, the secreted

recombinant product constitutes a very large fraction
of all of the total protein synthesized. Cells devote a
large portion of their protein processing capacity
to the secreted protein product. It is therefore
useful to review this process of protein secretion.
In a professional secretor, approximately 30% of
all cellular proteins are destined for organelles,
membranes, and secretion and are processed through
the ER. Although some proteins are translocated
into the ER post-translationally, most (including
typical recombinant DNA protein molecules)
are translocated as nascent protein molecules.
Proteins destined for secretion have a leader

sequence at the amino terminus that serves as
the signal peptide. After translation initiation, the
signal peptide of the nascent protein is recognized
by signal recognition particles (SRPs). This halts
translation and docks the nascent protein (which
has only the beginning segment of the entire
sequence) to a receptor on the ER membrane. It thus
prevents the protein molecule from being elongated
in the cytosol. The nascent polypeptide is then
transferred to a translocon on the ER membrane.
Subsequently, translation elongation resumes
and the elongating polypeptide passes through
the channel of the translocon into the ER lumen.
Folding of the polypeptide starts immediately upon
CELL
BIOLOGY
|32
32 |CELL
BIOLOGY
Fig. 2.6: Physiological changes incurred during the transition

from B cell to plasma cell
The Becoming of Plasma Cells -
A Hint to the Creation of a Super Secretor
Overloading of secretory protein molecules induces

unfolded protein response (UPR), which triggers the
differentiation process of B cells to become nondividing plasma cells.
B cells differentiate into plasma cells (or memory cells)
upon antigen stimulation, along with helper T cells,
increasing their size significantly and their ER by at
least 15 fold (in 4 days). They also increase metabolic
machinery significantly.
Xbp1 codes for XBP-1 (55KDa). XBP-1 is posttranscriptionally modified upon UPR, and activates
transcription of many ER proteins. Increase in XBP-1
coincides with an increase in antibody production (day
4 ), but lags in ER expansion.
ER appears to expand by increasing its abundance, not
merely selectively increasing some ER proteins. The
expansion starts before mass antibody production.
ER is a very oxidized environment (unlike cytoplasm
that is highly reduced). Disulfide bridges are formed
in ER and catalyzed by protein disulfide isomerase
(PDI). PDI and four oxidoreductase level increases
in ER as the B cells differentiate. Many of the redox
balance enzymes in cytosol and mitochondria are also
upregulated. There is also evidence to show that Golgi
increases along with ER.
Note: the drastically increased antibody secretion is
through biogenesis of ER and other protein secretion
machinery, NOT merely by faster throughput.
translocation into the ER lumen. The signal peptide on

the elongating polypeptide in the ER lumen is cleaved
upon entry into the ER. The protein concentration in
the ER is estimated to be 100 mg/mL, a concentration
at which proteins would otherwise aggregate and fall
out of solution. A class of ER chaperones and other
proteins that facilitate protein folding act on the
nascent protein molecules to prevent aggregation
and assist in folding. Their actions require cellular
energy (ATP). An important member of the ER luminal
chaperones, BiP, is also a component of the translocon
complex. In addition to BiP (also known as GRP78)
major ER luminal chaperones include calnexin,
calreticulin, and protein disulfide isomerase (PDI).
As will be discussed later, extensive glycolyation occurs

along the secretion process, starting in the ER and
continuing into Golgi bodies. In the ER, glycosylation
also serves as an indicator of correct protein folding.
Protein molecules that have completed the folding
process are exported from the ER by inclusion in
membrane vesicles. Vesicle fusion, fission, and
trafficking are the main forms of molecular transfer
from the ER to different organelles. Secretory proteins
in the vesicles are taken from the ER to the cisGolgi, which, along with the trans-Golgi, comprises
an array of tubules and vesicles on the opposite
side of the medial-Golgi. The medial-Golgi, typically
containing three to seven stacks of cisternae, is the
main site of glycan elongation for glycoproteins.
There are two different views on how protein
cargos are transported outward towards the TGN
and eventually to other organelles or secreted out
of the cell. The vesicle diffusion model hypothesis
states that the cargo from an earlier compartment
is transported to the next compartment by the
membrane vesicles. The cisternae maturation model
views the cargo as stationary inside the stack,
once they enter the compartment. The cisternae
(including the cargo) then moves outward, with its
enzyme constituents changing along the way, and
the cargo protein molecules becoming mature.
Eventually, the cargo at TGN is transported through

the vesicles to its final destination, be it the plasma
membrane (for secretion) or to other organelles. As
CELL BIOLOGY |33
CELL ENGINEERING|33
the contents of the early compartment translocate

to the later compartment, they need to be recycled
after the cargo is delivered. Thus, there are vesicles
for retrograde transfer, in addition to anterograde
transfer. The ER, Golgi, lysosomes, and endosomes are
all part of the secretory network. They communicate
through the dynamic trafficking of membrane vesicles.
An estimated 100 to 200 glycosyltransferases,

transporters
of
various
nucleotidesugars,
are
membrane
proteins
that
constitute the majority of Golgi enzymes.
Fig. 2.7: Secretion time of IgG heavy chain. Intracellular

proteins were completely labeled with C14N15 - arginine,
then switched to unlabeled medium. ~50% heavy chain is
secreted in two hours.
Table 5. Secretion time of liver proteins
Transferrin
Ceruloplasmin
Anti-trypsin
IgG
Half-life in ER Half-life in Golgi

(min)
(min)
110
45
80
30
30 - 40
10
Total ~120
Secretory proteins spend different amounts

of time in the ER and Golgi apparatus and in
different proportions
After translation, it takes a finite amount of time to

process protein molecules before they are excreted.
For an average protein of about 350 amino acids in
length, the translation takes only tens of seconds.
However, the time required for synthesized
proteins to be secreted depends on the nature of
the protein, and can thus range from 30 minutes to
a few hours. For example, the 1-protease inhibitor
is among the fastest secreted proteins, with a halflife of about 28 min. Transferrin, in contrast, takes
around two hours to be secreted. Even for the
same protein, the secretion time is not uniform
for all molecules. Rather, we observe distribution
between shorter and longer holding times.
CELL
BIOLOGY
|34
34 |CELL
BIOLOGY
Retrograde
transport
TGN
Trans Golgi
AAAA
A
AA
AA
AA
A
A
AA
Anterograde
transport
Medial Golgi
Endosome
Cis Golgi
Signal
peptide
Endocytosis
SRP
Bip
Endoplasmic
reticulum
Translocon
SRP
AAAAA
Nucleus
Fig. 2.8: Synthesis and secretion of proteins to an extracellular

environment
Protein Secretion
Nascent protein molecules destined for ER have a special ER signal sequence being synthesized in
organized polysome. They are recognized by SRP (signal recognition particle), a ribonucleoprotein.
SRP binding transiently arrests elongation, directing the ribosome/nascent polypeptide complex (RNC)to
the receptor on ER membrane and transfer the growing polypeptide to translocon.
SRP is released from the ribosome/nascent polypeptide complex.
The nascent polypeptide begins to pass through translocon and elongate into ER lumen.
Signal peptide on the elongating polypeptide is cleaved.
Protein folding and post-translation modification begins as polypeptide continues to elongate.
Major ER luminal chaperons: BiP, calnexin, calreticulin and PDI.
Ribosome is released once the translation is complete.
Folded protein (with inner core of glycan if it is a glycoprotein) concentrate at exit site of ER and is thought
to bud into vesicles and translocate to Golgi as pre-Golgi intermediates.
Golgi apparatus is in a dynamic state. There is also retrograde transport (its own proteins need to be
recycled) and anterograde transport.
After reaching trans Golgi network (TGN), secretory proteins are packaged into post Golgi vesicles and
move along cytoskeletal network through cytoplasm to fuse with plasma membrane and be secreted.
Different molecules of the same secretory protein spend different amounts of time inside the cell, i.e.
there is a distribution of holding time in the cell.
Translation of a protein molecules takes only seconds, but the secretion process takes tens of minutes to
hours.
CELL BIOLOGY |35
CELL ENGINEERING|35
Other Organelles
In addition to the nucleus, the mitochondria,

the ER and the Golgi apparatus, a number of
other organelles are also in the cytoplasm.
Lysosome:
Low pH
Site of degradation of cellular materials destined for
degradation and endocytosed material
Part of secretory pathway
Peroxisome:
Site of fatty acid oxidation
Rich in oxidative enzymes
Endosome:
In endocytosis the invaginated plasma membrane
forms small organelles
They move inward along the microtubule network
There is extensive cargo distribution and sorting
Some material sent to lysome
Some recycle to plasma membrane
A lysosome is an organelle with a low pH in its interior.

It is the site of degradation of ingested materials or
cellular materials that are no longer needed by the
cell. Most cellular materials have a useful life span,
regardless of whether they are catalyzing chemical
reactions or playing structural or mechanical roles.
Occasionally, a catalyzing enzyme can be improperly

locked up in its transition state, resulting in an amino
acid being modified to lose its catalytic capability.
Even in the cellular environment, some amino acids
in the protein may get oxidized. The accumulation
of such damages may render a protein nonfunctional. Thus, most proteins have finite life span.
Proteins that need to be turned over are tagged by
ubiquitin and sent to the proteosome for degradation.
Proteosomes are a complex of protealytic
enzymes that are capable of degrading proteins.
Through a process known as endocytosis, eukaryotic

cells can take up external particles by wrapping
the particles within cellular membranes and taking
them up as vesicles enclosed in lipid bilayers. This
process is not seen in prokaryotes. Some cells in
higher organisms are specialized scavengers that
engulf foreign particles or dead cells. Lysosomes are
the sites within those cells where engulfed particles
are degraded. Lysosomes contain a large number of
digestive enzymes. They have proton pumps in their
membrane to maintain a low interior pH (pH = 5.0).
In fat cells, the catabolism of lipid occurs in peroxisomes.

The reactions involved in such metabolism generate
large amounts of reactive oxygen and, thus, need to
be contained within these specialized organelles.
CELL
BIOLOGY
|36
36 |CELL
BIOLOGY
As discussed earlier in this chapter, cells are not

merely a droplet-like structure with its constituents
enclosed by a lipid bilayer. They form various shapes
and sustain mechanical forces by transmitting and
responding to mechanical force stimuli within, and
also between, cells. A class of proteins makes up
the cytoskeleton, accounting for their shape and
their ability to transmit and exert force. The three
major components of cytoskeleton are microtubules,
actin
fibers,
and
intermediate
filaments.
Cytoskeleton
Microtubules
Long, hollow tubes of polymerized subunit tubulin
(MW 50 kD), about 25 nm diameter, more rigid than
actin filaments .
Typically long and straight, many have one end (-)
attached to a single microtubule organizing center
(centrosome).
Form and break down rapidly.
- and -tubuline (GTP) form heterodimers; the dimer
assembles in a head-to-tail fashion into chains called
protofilaments, 13 of which make up the microtubule
wall.
Microtubules also play a key role in intracellular
organelles and small vesicle transport.
For secretory protein, the movement of postGolgi vesicle to plasma membrane is mediated by
microtubules.
A microtubule is an assemblage of hollow tube

structures formed by polymerized - and -tubulin
molecules. The polymerization and de-polymerization
occurs at the ends of the microtubules, allowing
them to extend and shrink their length rapidly.
The hollow organization enables cells to use
a smaller amount of material to give a longer
protrusion with a high structural rigidity, like the
hollow legs of aluminum ladders sold in hardware
stores. If the same amount of material were made
into solid legs, the legs would be rather thin and
would easily deform when subjected to stress.
Being tubes, microtubules are relatively straight
and do not bend in sharp angles or high curvatures.
Microtubules can be extended to protrude from some

region of the cell, and can rapidly shrink to retract a
part of the cell. They are also used as a train track in
the cytoplasm to transport cargos, such organelles
and membrane vesicles, to different parts of the cell.
When two ends of a microtubule are attached to

different objects, they can also pull them together
or push them apart. For example, during mitosis,
multiple microtubules work in coordination to
separate a pair of chromosomes, thus allowing
the daughter cells to each receive one copy.
Fig. 2.9: Structure of a microtubule molecule and its cellular

organization
CELL BIOLOGY |37
CELL ENGINEERING|37
Intermediate Filaments
Play a structural role, stable, transmit mechanical force

The subunit is not a globular protein, but fibril,
different from the tublin and actin
10 nm diameter, has a head, a tail and a -helical rod;
can be made of a wide variety of proteins in various
tissues.
The main role of intermediate filaments is to transmit

mechanical force, like the cable holding a suspended
bridge in place. Each fiber is made of subunit proteins
oriented in the same direction. The tail end of a subunit
is locked into the head of the next subunit. Each
intermediate filament fiber is made of multiple fibrils,
which are in turn made of a series of subunit proteins.
To increase the structural integrity, the tail-head lock

positions of different fibrils are at different locations in
the multi-fibril filament. These intermediate filament
fibers are flexible, capable of absorbing energy
exerted by external force and transmitting it to other
regions of the cell. In a tissue or in interconnected
cells in culture, intermediate filaments also help to
transmit forces between cells. Their deformable
nature allows them to act like a shock absorber
and to reduce the deformation of cells upon stress.
Fig. 2.10: Structure of an intermediate filament

molecule and its cellular organization
Actin Filaments
Actins are two- or three-stranded filaments that

often form web-like bundles underneath the
plasma membrane. Like ropes that are woven into
a net, actin fibers are twisted strings of filaments.
This geometry enhances their structural integrity
while maintaining a high degree of flexibility.
The actin-rich region immediately underneath

the cells plasma membrane often has a high
concentration of actin, where it forms a gel-like
structure, called the cortex. They are like a mesh of
thin nets underneath the lipid bilayer membrane.
They act as the first absorber of external mechanical
perturbations and give the lipid bilayer local shape.
Fig. 2.11: Structure of an actin filament molecule and its

cellular organization
Cells extend their body and spread flat on a surface,

both in tissues and in culture. The edge of an adherent
cell has an irregular shape, much like a fried egg. In
the protruded regions, actin fibers localize in the
lamellipodia and filopodia. In stationary cells, the
CELL
BIOLOGY
|38
38 |CELL
BIOLOGY
Two stranded filaments (F actin) of helical polymer of

actin (G actin)(50 kD)
5-9 nm flexible structures, organized into linear
bundles, 2-D networks and 3-D gels
Distributed all over the cell, but concentrated in the
cortex beneath the plasma membrane
The subunit, G actin, is a globular protein
Projections from cells, like microvilli, lamellipodia,
microspikes and filopodia are maintained by rigid
bundles of actin filaments
In non-motile cells, actin filaments form bundles called
stress fibers, loose meshwork of filaments underlies
cell membrane.
actin fibers form stress fibers throughout the cell,

but in moving cells, these visible fiber structures
tend to localize at the moving ends of the cell.
Actin filaments, along with the other two

major
components
of
the
cytoskeleton,
require many other component proteins to be
present for their polymerization, dissociation,
cargo
translocation,
and
other
functions.
In actively moving cells, stress fibers disappear and

actin filaments concentrate at the leading edge.
There are many actin-related-proteins which affect the
polymerization and motor functions of actin filaments
Actin is involved in motor functions. Best example is
muscle contraction.
Transport Mechanisms
The lipid bilayer membrane separates cells from
their environment. It presents a barrier to keep
most compounds outside the cell and to prevent
those inside from leaking out. It has a very low
permeability for large molecules, like proteins
and polysaccharides. Even polar or charged
small molecules cannot pass through easily.
Fig. 2.12: Order of magnitude estimation of the
permeability of various molecules across lipid membrane
Among the nutrients and metabolites, only oxygen,

fatty acid, and ethanol pass through the membrane
at a fast enough rate to meet growth requirements.
Specialized transport mechanisms mediate the
movement of the vast majority of nutrients and the
excretion of metabolites. Cells have a large number
of transporters (sometimes called permeases)
that allow molecules to cross the cytoplasmic
membrane and membranes of various organelles.
Such transporter-mediated transport is used to
pass small molecular weight compounds (up to
about 1 kDa) across the cell membrane (e.g., sugar,
oligosaccharides, amino acids, oligopeptides,
nucleotides, cholesterol, ions, organic acids, etc.).
Macromolecules are transported across membranes
by membrane fusion (in the secretion process
through the ER and Golgi apparatus), pinocytosis,
CELL BIOLOGY |39
CELL ENGINEERING|39
or exocytosis. Specific receptors, such LDL and

transferrin receptors, may be involved in pinocytosis.
Major Mechanism of Transport

The cellular transport of solutes (as opposed to
macromolecules) is grossly divided into two categories:
transport along or against the concentration gradient
of the solute. The former is thermodynamically
favorable, whereas the latter (called active transport)
requires energy input to make the process possible.
Three Classes of Transport Processes

Channel-mediated diffusion:
Molecules or ion specific; once channel is open, very
fast flux.
Facilitated difussion:
Provides molecule specific opening in the membrane;
barrier for molecular diffusion.
Active transport:
Moves molecules up against a concentration gradient.
Requires ATP or ion gradients of ion (H+, Na+) as an
energy source to drive the transport.
The energy source of active transport can be

derived from coupling to a chemical reaction,
such as the hydrolysis of ATP. Active transport of
a species may also be coupled to the diffusion of
another solute along its concentration gradient.
Thus, the driving force to transfer the second
solute along its gradient is used to push the first
solute to move up against concentration gradient.
Transport along the concentration gradient
can be mediated by carrier (transporter)
proteins
or
by
channel
proteins.
Channel proteins open into a duct-like structure across

the membrane that is specific for a specific solute,
such as water, Na+ or K+. Channel proteins exist either
in an open state or a closed state. Once the channel
is open, the transfer is very fast in the direction of
the concentration gradient. The flux is affected by the
number of channel protein molecules on the membrane
and by the time period that the channel is open.
Carrier-mediated transport is also called facilitated
diffusion. It entails, first, the diffusion of solute
from the high concentration side of the membrane
into the transporter, followed by the translocation
of the solute to the low concentration side of
the membrane. Once on the low concentration
side, the solute is free to diffuse away.
Under normal culture conditions, amino acids and

glucose are transported by facilitated diffusion. A
ubiquitous transporter for glucose is the glucose
transporter 1 (GLUT1). The rate of transport by a
transporter is dependent on the concentration of the
solute. More precisely, it depends on the concentration
CELL
BIOLOGY
|40
40 |CELL
BIOLOGY
difference of the solute across the membrane.

Its mechanism is similar to a typical enzymecatalyzed conversion of a substrate to a product.
The dependence of transport rate to solute
concentration can be described by MichaelisMenton enzyme kinetics. At low concentrations,
the transport rate is in proportion to the solute
concentration, while at high concentrations, the rate
is constant as the transporter becomes saturated.
Fig. 2.13: Three types of transport processes across
the cell membrane
The half-saturation constant (km) for GLUT1 is about

0.1 mM. In the 0.01 0.1 mM range, the glucose
import rate of the cell increases with increasing
glucose concentration. In the range typically used in
cell culture media (1 10 g/L, or 5.5 mM to 55 mM),
the rate is not affected by glucose concentration at all.
Transporters for facilitated diffusion and

active transport can also be categorized
according to the number of solutes each
carries and the direction of solute flow.
General Types of Transporters

Uniporter: Transfers a single molecule (e.g. glucose,
fructose), usually uncharged.
Bispecies-transporter (co-transporters):
Requires stoichiometric exchange of two species
simultaneously; important in change balance.
Symporter:Two species transported in the same
direction.
Antiporter:Two species transported in the
opposite direction.
Uniporters transfer a single solute from a high

concentration side to a low concentration
side, for example the GLUT1 transporter for
glucose and the GLUT5 transporter for fructose.
Symporters and antiporters transfer two

solutes, simultaneously. If the two solutes
move in the same direction, the transporter
is called symporter. Conversely, antiporters
transfer two solutes in opposite directions.
Collectively, symporters and antiporters are called
co-transporters. Co-transporters are often used to
transport charged organic molecules. Dissociable
solutes exit with a counter ion to maintain electric
charge neutrality. When a charged solute moves from
one side of the membrane to the other, the charge
neutrality must be maintained. Otherwise, the net
charge will accumulate across the membrane and
create impudence for further transfer of the solute.
For example, lactic acid exists as lactate in an
aqueous solution. If it is removed from the cell,
a negative charge will be moved along with it.
After a while, the cell membrane will be negatively
charged outside, creating a negative voltage. The
negatively charged outside will prevent further
CELL BIOLOGY |41
CELL ENGINEERING|41
excretion
of
the
negatively
charged
lactate.
In order to prevent the buildup of a charge across

the membrane, charged solutes are transported by
co-transporters. Two mechanisms are commonly
seen: 1) co-transport with a counter ion (such as
H+ for lactate) by a symporter, and 2) co-transport
with an ion of the same positive or negative charge,
but in the opposite direction (such as Cl- for HCO3-).
In co-transporter mediated transfer, the transport rate

is not only affected by the concentration difference of
the solute, but also by the concentration gradient of
the co-transported ion. Thus, the transport of lactate
by the monocarboxylate transporter (MCT) is not only
affected by the concentration of lactate, but also by pH.
Fig. 2.14: Three types of transporters categorized

by the solute being transported
Co-transporters may also be involved in active

transport. One such case involves an ion species
being transported along its concentration
gradient. The tendency of the ion to push
across the transporter is used to drive the
transport of a solute against its gradient.
For example, the Na+/glucose transporter in the

epithelium of the intestine can take up glucose from
the digestive track, even when glucose is lower than
in the cell. This is accomplished by using the Na+/
glucose transporter to transport two Na+ atoms from
the lumen of the intestine into the cell, where the Na+
level is very low. As the sodium ion is transported, a
glucose molecule also binds to the transporter and is
transported simultaneously. The propensity of Na to
move across the membrane is so high that it can drive
glucose to move against a large concentration gradient.
CELL
BIOLOGY
|42
42 |CELL
BIOLOGY
Transport of Nutrients
There are 12 different facilitative glucose transporters

(GluT) in animal cells. Different transporters are
expressed in different tissues.
GluT 1: the major transporter in all cells
Amino acid transporters have overlapping amino acid
specificity. Some amino acids compete for the same
transporter. Many have alternative transporters.
MCTs transport lactate, pyruvate together with H+
Major nutrients like glucose, other sugars, amino

acids and oligopeptides, are taken up by cells
through facilitated transport. The excretion of
lactate, ammonium, and some non-essential amino
acids are also transported by the same mechanism.
A large number of glucose and amino acid transporters

are present in the mammalian genome. Different
glucose transporters are expressed in different tissues
for different cellular needs, but GLUT1 is present in all
cells. GLUT4 is responsive to insulin and is expressed
only in some tissues. Upon insulin stimulation, the
intracellular GLUT4 molecules are translocated
to the plasma membrane to take up glucose.
The number of amino acid transporters in a

mammalian genome is also large. Some transport
entire classes of amino acids that share a
common property, such as the neutral amino acid
transporter for uncharged amino acids. Others are
specific for one or a small number of amino acids.
The monocarboxylic acid transporter (MCT),

the transporter for lactate, also transports
pyruvate. A number of MCTs are expressed
in different tissues. In a cell, different MCTs
are located at the cytoplasmic membrane and
others at the membranes of some organelles.
Another class of transporters for active transport is
the ATP-binding cassette (ABC) transporter, which
transports some hydrophobic compounds by utilizing
ATP. After prolonged exposure to methotrexate, some
cancer cells develop drug resistance by pumping
the chemical out of cells using ABC transporters.
Ion Transport
Bulk ion species (H+, Na+, K+, PO4-3, Cl-) are present
at very different concentrations across the cell
membrane. For instance, Na+ and K+ have opposite
directions in their concentration gradients across the
plasma membrane. The intracellular concentration
of K+ is 20 50 times higher inside than outside the
cell, while the Na+ concentration is 10 15 times
higher outside the cell versus inside the cell. Along
with the concentration gradients of major ions,
CELL BIOLOGY |43
CELL ENGINEERING|43
cells also maintain an electric potential gradient

of -80 mV across their plasma membrane. This
electric potential is fundamental to the transport
of many compounds across the membrane.
Because of this concentration gradient across the

membrane, sodium ions have a natural tendency to
flow into the cell, wherever they are allowed to pass.
The -80 mV membrane potential further enhances the
propensity of Na+ to influx, as the negative charge in
the inner surface of the membrane draws Na+ to move
across the membrane. The combined concentration
and electric potential gradients are used as a driving
force in the Na+-dependent glucose transporter,
which will transfer glucose from the lumen of
the digestive track into intestinal epithelial cells,
moving against a glucose concentration gradient.
The Na+/K+ ATPase transporter, which is present in

the cytoplasmic membranes of all animal cells, is
important for establishing sodium and potassium
gradients across the plasma membrane. ATPase
is an integral cell membrane protein that has
multiple subunits. It simultaneously transports two
K+ ions into the cell and three Na+ out of the cell.
Fig. 2.15: Transporters involving the transfer of ions

and ABC transporter
Although the cytoplasmic membrane is relatively

impermeable to ions, it does allow a small but finite
diffusion of Na+ and K+ along their concentration
gradient (i.e., a net influx of Na+ and a net outflux of K+).
It should be noted that the membrane permeability
for K+ is higher than Na+. Intracellular K+ also leaks
out through potassium channels. Overall, through
diffusion across the membrane and transport by
channel proteins, there is a net movement of ions
into the environment from the cytosol. This balance
is maintained by the action of the Na+/K+ ATPase.
Na+/K+ ATPase has three binding sites for Na+ and

two for K+. The Na+ binding sites on the cytosolic side
of the ATPase have a Km for Na+ in the range of submillimolar concentrations. Because the cytosolic Na+
concentration is about 10 mM, virtually all three Na+
binding sites exposed to cytosol will be occupied.
Na+/K+ ATPase also has a binding site for ATP.
Hydrolysis of ATP results in the phosphorylation of
the protein subunit and release of ADP. This allows
two K+ ions to bind to the protein on the extracellular
CELL
BIOLOGY
|44
44 |CELL
BIOLOGY
The concentrations of other major ion species (H+, Na+,

K+, Ca2+, PO4-3, Cl-) across membrane are polarized. Na+
and K+ have opposite direction in their concentration
gradient across the plasma membrane, which is
about ten fold difference. For Ca2+ the intracellular
concentration is so low that the gradient is extremely
steep.
Na+-K+ ATPases play a major role in maintaining Na+ and
K+ gradients. ATPase utilizes the hydrolysis of ATP as
the energy source.
Iron is extremely reactive and participates in many
redox reactions. In biological systems, it exists as
bound form. In its free form, it catalyses the
formation of peroxide and peroxidizes unsaturated
fatty acids. In cell culture, it is supplied as transferrinbound or bound by other chelators and taken up via
transferrin receptors.
Another class of ions are transported into the cells via
binding to proteins and internalized through specific
transportors. One example is iron transport by
transferrin. Iron is extremely reactive, participates in
many redox reactions. In biological systems, it exists
as bound form. In its free form, it catalyses the
generation of peroxide and peroxidize unsaturated
fatty acids. In cell culture, it is supplied as transferrin
bound or bound by other chelators and taken up via
transferrin receptor.
side, while simultaneously exposing the Na+ binding

sites to the extra-cellular solution. At the external
side of the enzyme, the Km of Na+ is at a higher
value. As a result, Na+ is released to the outside
environment. After the release of Na+, the phosphate
is released from the protein and K+ releases to the
intracellular environment. The Na+/K+ ATPase
then resets, ready for another round of reactions.
The net result is that ATPase uses ATP to pump three

Na+ ions out and two K+ ions into the cell. With its
3:2 stoichiometric ratio of sodium to potassium, a
prolonged operation of ATPase without any balancing
action will inevitably generate a large electric
potential across the membrane. To counter this, cells
also have chloride ion pumps to pump negatively
charged Cl- out of the cell. In some cells, Na+ and Clchannel proteins also facilitate the maintenance of
the membrane electric potential in the correct range.
Extracellular Matrices and Cell

Movement
ECM
The vast majority of cells in the body are embedded

in tissues in an acelluar tissue structure. When
cultured in vitro to plastic or glass surfaces, they
secrete materials onto the surface after adhering.
Those excreted materials to which cells attach are
collectively called the extracellular matrix (ECM).
The ECM is made of proteins, proteoglycans,

and glucosaminoglycans. Different types of cells
excrete somewhat different ECM components.
Among the prominent members of ECM
proteins are collagens, laminins, and fibronectin.
The role of ECM is not merely to provide a surface
appropriate for cell adhesion. It is also important for
cell-surface signaling and growth control. Receptors
on the cell membrane establish cell adhesion
complexes with ECM components and a tension force
is then transmitted through cytoskeletal fibers and
CELL BIOLOGY |45
CELL ENGINEERING|45
ECM Proteins

Collagen
Laminin
Firbronectin
the cells internal signaling pathway to allow cell cycle

to proceed. The cell adhesion receptors exhibit a wide
range of diversity. For example, integrin receptors
have a variety of and components and form a large
number of combinations of integrin complexes that
have different affinity for different ECM components.
As a result, different cell types often have different
ECM requirements for adhesion and growth.
Proteoglycan

Chondroitin sulfate
Glucosaminoglycan

Heparin
Hyaluronic acid
The extracellular matrix is rich in electrocharges,

allowing growth factors, cytokines etc. to be stored
inside them
The extracellular matrices are the substrate for
adhesion of many cell types, and provide cues for
growth, differentiation and development
Cell Movement
Many ECM components are highly negatively

charged.
This allows many protein growth
factors to be adsorbed to the ECM and released
to surrounding cells, perhaps even serving as
chemoattractants. Therefore, they also play a role in
providing cues for cell migration and differentiation.
The vast majority of cells are capable of

movement on surfaces.
In general, cell
movement can be the result of an attraction to
chemicals, or unidirectional random movement.
The filopodia extend as a result of microtubules

growing at the cell front, and establish a grip on the
surface
The subsequent dissociation of cell-substrate
contact at the rear of the cell allows the cells center
of mass to move forward
Cell movement is a coordinated series not

unlike walking. It involves a restructuring of the
cytoskeleton, a protrusion of the membrane, the
establishment of surface adhesions on one side of
the cell, and the detachment of the cell membrane
from adhesion complexes in the rear end of the cell.
The actin fibers and plasma membrane of moving

cells extend the cell to become more elongated in one
direction. The extended regions form lamellipodia,
which may also contain microspikes, or filopodia,
which are active even within a few minutes.
Cells moving in an open surface (i.e., not

crowded) move randomly. They exhibit locomotion
contact inhibition, meaning when two moving
cells encounter each other, both will move away
in opposite directions. Furthermore, the two
daughter cells of a dividing mother cell move away
from each other when cell division is complete.
Cell migration is a regulated by many factors,
including growth factors. For example, epithelial cells
CELL
BIOLOGY
|46
46 |CELL
BIOLOGY
respond to hepatocyte growth factor by moving away

from each other and become more scattered, instead
of forming cell clusters typical of epithelial cells.
Fig. 2.16: A cell moving on its substrate. Lamellipodia

extend in the leading edge on the right side. A few filopodia
are extruding out.
Cell migration is not intentionally controlled

or manipulated in cell bioprocessing. However,
in some cases, when seeding cells into threedimensional matrices for tissue engineering
applications, efficient cell migration into the interior
of matrices is important for subsequent growth.
Growth, Death and Senescence

Cell Cycle and Growth Control
Positive and Negative Cues
Cell Cycle
G1, S, G2 and M constitute four consecutive phases
of cell cycle. S stands for DNA synthesis and M for
mitosis. They are separated by two gaps.
The duration of S and M phases is relatively constant.
Cells grow at different rates and spend differnt
amounts of time in G1 and G2. For mammalian cells,
the duration of S phase is 5-8 hours and that of M
phase is 1 hour.
The progression from G1 to S, and from G2 to M phase
is tightly controlled. Only when a cell is ready, will it
proceed to the next phase.
Before M phase, chromosomes and all organelles,
ribosomes and other cellular contents are all
duplicated from time 0 (immediately after cell
division).
Cell growth is the manifestation of a delicate

balance between positive and negative regulations
that respond to signals both outside and inside
the cell. Positive signals stimulate cell growth
and proliferation and suppress the cell death
mechanism, while negative signals suppress those
events and promote cell death. External signals
from the environment tell cells the availability or
absence of nutrients necessary for DNA replication
and biomass synthesis. External signals from other
part of the body allow cells to coordinate their
response to the need of the organism. The internal
signals modulate cellular programs to increase
cellular component content, to divide, or to die.
Eukaryotic cells progress through four stages in their

procession to grow in cell number: G1, S, G2, and M
phases. G1 and G2 refer to the gap phase. S and M
phases derive their designation from DNA synthesis
and mitosis, respectively. The four stages constitute a
cell cycle and this cycle is repeated every time a single
cell becomes two daughter cells. Cells that are in a long
period of quiescence, such as terminally differentiated
cells, divert from G1 and enter G0 stage indefinitely.
Checkpoints are present between the different stages

of cell cycle control. After mitosis, cells increase in
size and mass. They only enter the S phase from G1 if
cellular conditions are right. Similarly, they only enter
mitosis from G2 if cellular components are ready.
CELL BIOLOGY |47
CELL ENGINEERING|47
Additionally, the decision to enter the S phase is

subjected to the regulation of external positive
mitogenic factors, such as insulin, insulin-like
growth factors, and fibroblast growth factors.
Anchorage-dependent cells also receive growth
stimuli by establishing contacts between the
surface receptors and the ECM, and thereby
maintaining tension in the cytoskeletal network.
Countering the actions of the mitogenic factors are

those factors that provide signals to cause growth
arrest. Cell-cell contact, for instance after reaching
a confluent state, causes growth to cease. Such
contact inhibition of growth has been noted for more
than five decades. Only recently have researchers
found that it is the interference of the adherent
junctions between cells that disrupts a cells
internal signaling networks to cause growth arrest.
Fig. 2.17: Phases of cell cycle and

approximate duration of each phase
Cyclins and CDKs
Whether cells divide and grow or self-destruct is the

outcome of a balancing act of a network of external and
internal positive and negative factors. Loosening the
controls may lead to unscheduled proliferation and
transformation of cells to their malignant derivatives.
The progression through each of the four phases

of the cell cycle (G1, S, G2 and M) is positively
regulated by cyclins and cyclin-dependent kinases
(CDKs) and negatively controlled by CDK inhibitors
(CDI), which deactivate cyclin-CDK complexes.
Each of these regulatory proteins displays

a characteristic periodic dynamic profile
throughout the cell cycle. Each proteins profile
is the result of interactions of the other cell cycle
regulatory components with their expression,
activation,
inactivation,
or
degradation
corresponding to a specific time frame.
An important cell-cycle checkpoint occurs along the
transition from the G1 to S phase. The pivotal player
in the G1/S phase transition is the CDK4/6-cyclin D
complex. The activated CDK4/6-cyclin D complex can
phosphorylate the regulatory protein retinoblastoma
CELL
BIOLOGY
|48
48 |CELL
BIOLOGY
Growth Control
Growth, the increase in biomass and its contents (i.e.
organelles and cytosol) as well as the increase in cell
number of almost all cells (except those which have
been adapted in vitro) is regulated by growth factors,
cytokines.
The regulation balances positive factor (mitogenic) and
negative factors. which prevent cell death or apoptosis.
The most common growth factors for cells of
bioprocess interest are insulin or insulin-like growth
factors (IGF).
IGF and insulin have different receptors on cell surface.
Insulin regulates glucose metabolism and has a
mitogenic effect. IGF, which is used in much lower
concentrations and also has a mitogenic effect.
(pRB). pRB, in its unphosphorylated state, binds

to and inhibits the transcription factor E2F. Upon
phosphorylation, pRB dissociates from E2F, leading
to the activation of cyclin E transcription by E2F.
E2F activation positively regulates the transcription
of genes involved in cell cycle progression.
Inputs from growth factor signaling and cell

adhesion-mediated signaling are prerequisites to the
G1 phase. These two pathways are not independent
of each other. On the contrary, they have rather
extensive crosstalk. In normal untransformed
cells, all the important growth factor signal
transduction cascades are regulated by integrinmediated cell adhesion. As a result, adherent
cells rely on attachment to the ECM for growth.
Except for vaccine production, where normal

diploid human fibroblasts are employed, virtually
all cells used for protein production are continuous
cell lines, including CHO, BHK, HEK 293, and
mouse myeloma cells, such as NS0 and Sp2/0. All
of these cell lines have lost their normal growth
control. Their cell cycle checkpoint controls have
been compromised and their entry into a quiescent
state in the absence of mitogen has been relaxed.
CELL BIOLOGY |49
CELL ENGINEERING|49
Fig. 2.18: Schematic representation of the interaction between cell cycle and apoptosis pathways. CDK Cell cycle-dependent
kinase, IRF-1 interferon-regulatory factor 1, pRb phosphorylated retinoblastoma, ERK extracellular signal regulated kinase, FADD
Fas associated death domain protein, FLIP FLICE-inhibitory protein, Cdc42 cell division cycle 41, EIF4E eukaryotic translation initiation factor 4E, Cyc Cyclin, XIAP cross-linkied inhibitor of apoptosis proteins, Apaf-1 apoptotic peptidase activating factor 1,
BAX Bcl-2 associated X protein, BAK Bcl-2 homologous antagonist/killer, Ub ubiquitination, cyt C cytochrome C
CELL
BIOLOGY
|50
50 |CELL
BIOLOGY
Apoptosis
Cells, under some conditions, commit suicide, undergo
programmed cell death; this is different from cell death
caused by injuries (necrosis).
Necrosis may entail cell swelling, rupture and leakage
of cellular materials. In vivo, necrosis may cause
inflammation of surrounding tissues after encountering
cell debris.
Apoptosis is the process of regulated cell

death in response to developmental cues or
to accumulating non-lethal stresses, such as
nutrient depletion, growth factor deprivation,
virus infection, and/or metabolite accumulation.
This process of programmed death is marked

by specific cell morphological changes: DNA
condensation, chromatin shrinkage, and membrane
bulging (also called blebbing). The final intracellular
event involves a series of cascades leading to cellular
Apoptosis can be caused by the lack of positive signal/
growth factors (e.g. withdrawal of IGF) or the imposition destruction. These final acts of self-destruction
of negative signals.
are similar in all apoptosis mechanisms; however,
Two general pathways for apoptosis: one, induced by
the initiating signal can differ. The two major
specific signals, occurs during cell development; another,
apoptosis signaling pathways are the death
induced by a variety of stress conditions.
receptor pathway and the mitochondrial pathway.
Apoptosis entails cell shrinkage, mitochondria breakage

and release of cytochrome C, DNA fragmentation, and
release of phosphatidylserine from phospholipids which
causes phagocytotic cells to engulf the cell fragments.
Death Receptor Pathway
In many developmental events, individual cells

serve their function only for a finite period of time.
Beyond this period, their existence may interfere
with, or even imperil, the well being of the organism.
In those cases, cells are built to die after their
functional duration by endowing an individual
cells survival to depend on the presence of positive
factors or the absence of negative effectors.
In the event that a cell survival signal is absent

or a cell death signal is present, a cell undergoes
self-destruction and ceases to serve its function.
Developmentally related apoptosis is largely regulated
by death receptors on the cell surface. The death
receptor pathway is mediated by binding, or a lack of
binding, of ligands to death receptors. For example,
immature neurons die in large numbers during early
brain development because neuronal cells require
positive survival signals. The lack of such positive
survival signals leads to neurodegenerative disorders.
Ligand binding to death receptors initiates the

recruitment of an adaptor molecule, Fas-associated
death domain (FADD), to the cytoplasmic end
of the receptors. The presence of FADD causes
caspase 8 or 10 to associate with the receptor,
forming a death inducing signaling complex. The
caspase is then proteolytically activated, triggering
CELL BIOLOGY |51
CELL ENGINEERING|51
the activation of a series of downstream effector

caspases (3, 6 and 7). The activation of these effector
caspases leads to the final stages of cell destruction.
Mitochondrial Pathway
Mitochondrial Apoptotic Pathway

Non-Apoptotic State:
The balance of anti-apoptotic and pro-apoptotic factors
At non-apoptotic state anti-apoptotic factors
hold pro-apoptotic factors in check
Apoptogenic factors are held in inter membrane space
of mitochondria
Apoptotic State:
Upon insult of apoptotic inducing agents or environment
factors
Anti-apoptotic factor(s) had conformational change
(exposing BH3)
Membrane disruption releases cytochrome C
and other pro-apoptotic and apoptogenic factors
The releases of procaspase3 form apoptosome
with the adaptor (Apaf -1), becomes capase-3.
Caspase-3 converts procaspase-9 to caspase-9
which becomes the executioner caspase and starts
the apoptotic cellular event
In addition to its role in energy metabolism,

mitochondria also sequester pro-apoptotic proteins in
the space between their outer and inner membranes.
These pro-apoptotic factors are released in a controlled
manner in stressed cells to initiate apoptosis.
Mitochondrial cytochrome C, a major component
of electron transfer, is also a signal for apoptosis.
As in the control of cell growth, the components of

the mitochondrial apoptosis pathway also involve
positive proapoptotic and negative anti-apoptotic
factors. The Bcl-2 family that consists of over 20
pro- or anti-apoptotic proteins is a major player
in the mitochondrial apoptosis pathway. The proapoptotic subfamily includes Bax, Bak, and Bok,
which all contain BH1, 2, and 3 homology domains.
Upon exposure to death signals, Bax undergoes

conformational changes and translocates to
the mitochondria, where it inserts into the
outer mitochondrial membrane and forms
channels. These channels allow the leakage of
cytochrome C and other pro-apoptotic molecules.
Cytochrome C proceeds to form a complex with
Apf-1, pro-caspase 9, and dATP, known collectively
as the apoptosome. In the apoptosome, the inactive
pro-caspase 9 is activated and the active enzyme
subsequently activates downstream caspases.
Two anti-apoptotic proteins, Bcl-2 and Bcl-xL,

counter the actions of the pro-apoptotic components.
Bcl-2 is localized on the mitochondrial membrane
and inhibits the release of pro-apoptotic molecules
from the mitochondria by maintaining membrane
integrity. Bcl-xL is localized in the cytoplasm and
binds to pro-apoptosis members of the Bcl-2 family.
The involvement of multiple protagonist and
antagonist factors ensures tight control of apoptotic
event. This scheme also provides an amplification
CELL
BIOLOGY
|52
52 |CELL
BIOLOGY
of the desired signal. An excellent example of this

is the cascade of caspases. When cell destruction
is needed, the signal is greatly amplified through
a series of steps where one caspase activates
another caspase in an exponential fashion.
Fig. 2.19: Mitochondrial pathway mediated apoptosis
Senescence and Telomeres
Fig. 2.20: life of normal cells from isolation to

senescence and occurrence of cell line
The vast majority of animal cells isolated from tissues

require surface adhesion in order to multiply, since
they are anchorage-dependent cells. Cells are typically
isolated from tissues by an enzymatic dissociation of
the tissue. After dissociation and the removal of large
chunks of undissociated debris, cells are plated on a
compatible surface overlaid with media. Under the
microscope, cells in media suspension can be seen to
attach to the surface or out-grow from the remaining
tissue chunks. Subsequently, they extend their
body length, spread and begin to multiply. Those
cells derived from normal tissues generally possess
two sets of chromosomes and are diploid cells.
Eukaryotic cells enter an exponential growth phase,

similar to microbial cells. The growth rate slows as
they begin to cover the entire surface area to form a
monolayer. Upon reaching confluence, they stop
dividing. While the cell bodies of neighboring cells
may cross each other, their nuclei never overlap.
This is called contact inhibition of cell growth.
Cells can be dissociated from a surface after being
treated with trypsin (i.e., trypsinized) or with other
proteases. They can then be plated on a larger surface
area for continued growth. This process can then
be repeated to expand the population. Each round
of detachment and expansion is called a passage.
Non-immortalized cells cannot be grown in culture
indefinitely by simple repeated passage in culture.
Normal diploid cells from animals (except stem
CELL BIOLOGY |53
CELL ENGINEERING|53
tissue dissociation
Plating in nutrient medium
In a historical experiment carried out half a century

ago, the continued passaging of mouse fibroblast
cells beyond crisis gave rise to a small fraction of
survivors. These cells eventually grew, expanded,
and could be cultured continuously in vitro without
a limited life span. They were given the name 3T3
because the cells were passaged every three days by
expanding the surface area three times more. These
cells appear normal and are subjected to contact
inhibition of growth under typical culture conditions.
Growth
Replating to larger surface area
cells) have a limited life span in culture. Fibroblasts

(a cell type from connective tissue) isolated from
a mouse embryo can be cultured in vitro for about
60 doublings. As that limit approaches, the cells
begin to fail to reach confluence. Eventually, high
passage cells will cease to grow. This is referred
to as having reached crisis. Such a limit in
the proliferating potential is called Hayflicks
phenomenon. It is a common phenomenon for
all normal diploid cells obtained from vertebrates.
Contact inhibition
Cell Dissociation
Fig. 2.21: Anchorage dependency and contact

inhibition of cultured normal diploid cells
However, although the parental mouse fibroblast

cells had diploid set of chromosomes before
reaching crisis, 3T3 cells have abnormal number
of chromosomes. Cells that succumb to Hayflicks
constraint (e.g., those that are diploid and have a
limited life span) are called cell strains. The cells
that reestablish after crisis and can grow in culture
indefinitely, but are aneuploid (do not have a normal
set of chromosomes), are referred to as cell lines.
Cells derived from cancer also give rise to cell lines.

Cell lines can also be established by immortalization
through viral or oncogene transformation. These
transformed cells are also often aneuploid. They
are capable of growing beyond the monolayer, since
they are not subject to contact inhibition of growth.
Fig. 22: Telomere at the end of a chromosome
Normal diploid cells, thus, appear to count their

number of doublings. They achieve this by using
their telomeres. Telomeres are special repetitive
sequences at the end of chromosomes. They are
not replicated by DNA polymerase during DNA
replication, but are synthesized by telomerase. The
reaction is not precise for accurately reproducing
CELL
BIOLOGY
|54
54 |CELL
BIOLOGY
the number of tandem repeats of the sequence, thus

there can be much variation in telomere length among
cells. As the number of passages increases, telomeres
decrease their length unless they are repaired by
telomerase. For instance, in stem cells, telomerase
activity is high to maintain the telomere length. Unlike
cell strains, stem cells do not exhibit senescence.
Concluding Remarks
In this long chapter we provided a condensed
overview of knowledge in cell biology that is
essential for biotechnologists to practice cell culture
bioprocessess.
The structure and make-up of
cells that gives them their functional versatility,
also constrains their capability.
In practicing
biotechnology, while exploiting their biological
versatility, we also must understand cells structural
and functional constraints and biological limits. In
the meantime, we must also keep in mind that our
objectives are frequently different from scientists
studying the biology of the cell. To fully harness a
cells biological potential, we do not necessarily need
to be bound by the nature of cells; rather we should

employ means to adapt them to serve our goals
better. For example, most cells used for biologics
production were anchorage-dependent originally
and are now cultivated not only in suspension, but
also in highly turbulent flow conditions. Most of the
adaptation processes in the past two decades were
conducted empirically. At a molecular level, what
causes those cells to have the adapted behavior is
poorly understood. By equipping ourselves with a
better knowledge of cells capability and limits, we will
be able to push the technological boundary further.
CELL BIOLOGY |55
CELL ENGINEERING|55
CELL
BIOLOGY
|56
56 |CELL
BIOLOGY
Cell Physiology for Process

Engineering
Overview of Central Metabolism of Cultured Cells. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Glucose and Energy Metabolism. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Oxidation of Glucose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Pentose Phosphate Pathway. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Lactate Formation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Carbon Flow and Reaction Intermediates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
NADH Balance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Role of Transport and Transporters in Metabolism. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Glucose Transporters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Lactate Transport. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
Transport Across Mitochondria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
Regulation of Glucose Metabolism. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
Isozymes and Differential Allosteric Regulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
Signaling Pathways and Regulation by Growth Control . . . . . . . . . . . . . . . . . . . . . 77
Metabolic Homeostasis and Lactate Consumption. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Glutamine and Its Relation to Energy Metabolism . . . . . . . . . . . . . . . . . . . . . . . . . 80
Amino Acid Metabolism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Lipid Metabolism. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Lipid Transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
Fatty Acid Metabolism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Acetyl CoA shuttle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
Cholesterol & Its Biosynthesis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
Glycan Biosynthesis and Protein Glycosylation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
Importance of Glycan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
Protein Folding and Glycosylation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
Glycan Extension in Golgi Apparatus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
Glycan Types and Microheterogeneity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
Synthesis and Transport of Nucleotide Sugar Precursor. . . . . . . . . . . . . . . . . . . . . 93
Glycan Diversity Among Species . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
CELL PHYSIOLOGY FOR PROCESS ENGINEERING |57
Overview of Central Metabolism of Cultured Cells

Glucose Oxidation
C H O + 6O
6
12
6CO + 6H O
Glucose Anaerobic Metabolism

2CH6 $ CHOH $ COOH
CH O
6
12
Glutamine Oxidation
CO (NH ) CH CH CH (NH ) COOH + 4.5O
2
2NH + 5CO + 2H O
3
Cells in culture take up sugar, amino acids, lipids,

and nucleosides from their growth media. They
metabolize these components to derive energy and
use them as building blocks to generate more cell
mass, create more cells, and produce products. The
processes of making more cell mass and protein
product are very energy intensive. Proteins constitute
over 50% of the dry mass in a typical cell; they are
essentially amino acids connected by peptide bonds.
The synthesis of each peptide bond costs at least 3
ATP, which is nearly 1/10 of the amount of energy that
can be obtained by oxidizing one glucose molecule.
One high-producing recombinant cell produces over
40 pg per day of IgG protein. Since an average cell
has about 400 pg of cell mass (or about 200 pg of
cellular proteins) it is easy to see that producing the
protein product is a major energetic load for cells.
A classical cell culture medium contains 1 5 g/L

of glucose, and somewhat lower levels of amino
acids (about 0.8 mM, or 1 g/L). The sum of the
nutrients, together, typically generates only about
1 3 x 109 cells/L, or approximately 0.1 0.3 g/L
of cell dry mass. The efficiency of producing cell
mass from glucose and other nutrients is rather low.
Fig. 3.1: Lactate and ammonium profiles in manufacturing runs

with high product titers (blue) and low product titers (red).
Lactate profile correlates to productivity, but not ammonium.
Glucose is the most important source of energy for

most cells. Even when another sugar such as galactose
or fructose is used as the sole carbohydrate source,
it still enters the metabolic pathway that has evolved
for glucose (called glycolysis) to get catabolized.
The complete oxidation of one glucose molecule

consumes six O2 and generates six H2O and six CO2. For
cells in culture, however, the majority of consumed
glucose is not completely oxidized; it is converted to
lactate and excreted. By converting to lactate instead
of completely oxidizing to CO2, much less energy is
derived from each mole of glucose. This is the root
cause of the low efficiency in conversion from glucose
to cell mass. For some cells, especially transformed
cell lines, each mole of consumed glucose produces
almost two moles of lactate, which is the theoretical
maximum of the glucose to lactate conversion.
Glucose, Glutamine Major Carbon Source
Both glucose and glutamine consumed in excess to

what are need to grow biomass
Most glucose consumed, converts to lactate
Excess glutamine consumption, results in ammonium
excretion
Lactate Acid and Fedbatch Culture Productivity
Lactate is produced in exponential growth phase

In stationary phase, some cultures continue to produce
lactate, others switch to lactate consumption
Lactate consumption correlate to sustained higher
viability and high productivity
This type of wasteful metabolism is common

to almost all vertebrate cells in culture. For
bioprocessing, the accumulation of these byproducts
inhibits cell growth and impedes productivity.
Cells invariably produce lactate from glucose when

growing rapidly. However, under some conditions,
such as the stationary phase of fed-batch culture,
lactate may also be consumed. It is not unusual
that under the same operating conditions, different
culture runs have different metabolic outcomes. In
some runs, lactate production in the rapid growth
phase continues into the stationary phase. In
others, the transition from lactate production to
lactate consumption occurs in the stationary phase.
When production data from a manufacturing
plant were analyzed, it was found that the top
productivity runs switched from lactate production
to lactate consumption, while low productivity runs
remained in lactate production mode throughout the
culture. This is an indication that cell metabolism
plays a key role in determining productivity.
Glucose and Energy Metabolism
Glucose Oxidation
Main metabolic pathways in energy metabolism:
Glycolysis
TCA cycle (tricarboxylic acid cycle)(Kreb cycle)
Pentose phosphate pathway (PPP)
Oxidation of Glucose
Glucose is mainly catabolized through three

pathways: glycolysis, the pentose phosphate
pathway (PPP), and the tricarboxylic acid cycle
(TCA) cycle. In glycolysis, one mole of glucose is
converted to two moles of pyruvate. In this segment
of catabolism, only a small fraction of the chemical
potential energy of glucose is converted to the
usable form of chemical potential energy in the
cell, i.e., ATP. Two moles of ATP are generated
per one mole of glucose. Pyruvate may enter the
TCA cycle for further oxidation, or it may become
a shunted product as lactic acid (at a neutral pH it
exists as lactate). Through the TCA cycle, the carbon
skeleton of glucose is finally broken down to CO2 and
H2O. The PPP is a shunt from glycolysis. It generates
five-carbon sugars for nucleoside synthesis
and supplies NADPH for many biosynthesis
reactions and to maintain a redox state in the cell.
In eukaryotic cells, glycolysis and PPP take

place in the cytosol, while the further oxidation
of pyruvate to CO2 occurs in the mitochondria.
It is in the mitochondria that the majority of
the chemical potential energy of glucose is
converted to ATP for use in cellular synthesis
and other energy-dependent cellular processes.
In glycolysis, glucose is broken down to two

pyruvates and generates two ATP and two
NADH. Pyruvate may then be further converted
to lactate as a final product, instead of entering
the TCA cycle. Thus, when glucose metabolism is
terminated at glycolysis, it produces two lactate
molecules and two ATP, but no additional NADH.
Although overall glycolysis generates high-energy

compounds (e.g., ATP and NADH), its first segment
actually consumes ATP. Two ATP are used to add a
phosphate group to each end of the glucose molecule.
The two phosphate groups pull their surrounding
electron clouds toward the two ends of the
molecule, thereby making the carbon-carbon bond
Glycolysis - Each Mole of Glucose (6 Carbons)
Consumes two moles ATP (to activate to fructose, 1,

6-bisphosphate)
Produces two moles NADH, 4 moles ATP
Net:
Becomes 2 Pyruvate
Produces 2ATP, 2NADH
TCA cycle
Pyruvate enters mitochondrion

Pyruvate loses CO2, becomes acetyl CoA (2 carbons)
Acetyl CoA enters TCA cycle,
Becomes 2 CO2
Produces NADH, FADH2, which stores energy
Never reacts with oxygen directly
NADH, FADH2 enter electron transfer pathway

passes its high energy electron down, also its proton
As electron passes on energetic ladder, it pumps
protons out of mitochondrion
create a high pH inside mitochondrion
also a negative charge of ~120 mV across
mitochondrion inner membrane
The electron and proton, at the bottom of energetic
ladder, react with oxygen to form water.
Oxidative Phosphorylation Pathway
The higher concentration of proton in cytosol and the

negative charge inside mitochondrion drives proton to
move into mitochondria
Those protons moves into mitochondrion by
passing through ATP synthase; as they pass through
ATP synthase, ADP is converted to ATP inside
mitochondrion
There are a lot of fluxes across mitochondrial
membrane, inclusind
going in: pyruvate, ADP, phosphate, H+,
going out: ATP, CO2
in the middle susceptible to enzymatic cleavage.
After cleavage, the six-carbon skeleton becomes

two three-carbon compounds: glyceraldehyde-3phosphate (G3P) and dihydroxyacetone-phosphate
(DHAP). These two compounds are interconvertable
through a reversible reaction. The continued
reaction of glyceraldehyde-3-phosphate (G3P)
effectively draws DHAP toward G3P and moves it
further downstream in glycolysis. The conversion
of two G3P to the end product of two pyruvates also
converts two NAD+ and four ADP to two NADH and
four ATP. The net energetic consequence of the
conversion of glucose to two pyruvate in glycolysis
is the generation of two ATP (because two ATP
are consumed to activate glucose) and two NADH.
The further oxidation of pyruvate takes place in the

mitochondria, where it is first converted to acetyl
CoA, releasing one CO2 and generating one NADH.
Acetyl CoA is then fed into the TCA cycle where it is
broken down to two CO2. The pathway is cyclic, with
four- to six-carbon skeletons cycling in a loop. At the
beginning of the cycle, the four-carbon oxaloacetate
(OAA) takes in acetyl CoA to become citrate. Citrate
has three carboxylic acid groups; hence the name
tricarboxylic acid cycle. The TCA cycle is also
known as the citric acid cycle and the Krebs cycle.
As noted before, molecular oxygen does not react

with the carbon compounds in the reactions from the
TCA cycle. CO2 is released, through decarboxylation
reactions, from the carbon skeleton without the
participation of molecular oxygen. In two of the
reactions catalyzed by pyruvate dehydrogenase
and -ketoglutarate dehydrogenase, the energy
from the breakup of the C-C bond is preserved in
the high-energy compounds acyl CoA (acetyl CoA
and succinyl CoA, respectively) and NADH. In the
other case, one of the three carboxylic acid groups
in citrate is released and one NADH is generated.
If the carbon-carbon bond is broken by directly

reacting with oxygen, such as the case for
combustion, a very high temperature is necessary
to provide the activation energy. As we all know,
it takes a fire to burn wood. Furthermore, the
Figure 3.2. Major pathways in energy metabolism. Glucose and glutamine uptake, glycolysis, lactate excretion,
TCA cycle, and oxidative phosphorylation.
Symbols of metabolites in energy metabolism:
Glc: Glucose; g6p:Glucose 6-phosphate; f6p:Fructose 6-phosphate; f16bp (f16p2): Fructose 1,6-bisphosphate; f26bp (f26p2): Fructose
2,6-bisphosphate; gap: Glyceraldehyde 3-phosphate; Dhap: Dihydroxyacetone phosphate; 1,3bpg: 1,3-bisphosphoglycerate; 3pg:
3-phosphoglycerate; 2pg: 2-phosphoglycerate; pep: Phosphoenolpyruvate; pyr: Pyruvate; lac: Lactate; NADH: Nicotinamide adenine dinucleotide
(reduced); NAD: Nicotinamide adenine dinucleotide (oxidized); NADPH: Nicotinamide adenine dinucleotide phosphate (reduced); NAPD:
Nicotinamide adenine dinucleotide phosphate (oxidized); Gln: glutamine; Glu: glutamate; Asp: aspartate; ala: alanine; Mal: malate; KG:
-ketoglutarate; OAA: oxaloacetate; SucCoA: succinyl CoA; 6pg: 6-phosphogluconate; ru5p: ribulose 5-phosphate; r5p: ribose 5-phosphate; xyl5p:
xylulose 5-phosphate; e4p: erythrose 4-phosphate; s7p:sedoheptulose 7-phosphate
Symbols of enzymes and transporters in energy metabolism:
GLUT: Glucose transporter; HK: Hexokinase; GPI: Glucose phosphate isomerase; PFK: Phosphofructokinase; PFKFB: 6-phosphofructo-2-kinase/
fructose-2,6 bisphosphatase; ALDO: Aldolase; TPI; Triosephosphate isomerase; GAPD: Glyceraldehyde 3-phosphate dehydrogenase; PGK:
Phosphoglycerate kinase; PGM: Phosphoglycerate mutase; ENO: Enolase; PK: Pyruvate kinase; LDH: Lactate dehydrogenase; PYRH: Pyruvate
mitochondrial transporter; G6PD: Glucose 6-phosphate dehydrogenase; 6PGD: 6-Phosphogluconate dehydrogenase; RPE: Ribulose phosphate
epimerase; RPI: Ribose phosphate isomerase; TK: Transketolase
TA: Transaldolase; PRPPS: Phosphoribosylpyrophosphate synthetase; PYRH: Pyruvate mitochondrial transporter; MCT: monocarboxylate transporter
chemical potential energy would have been released

as heat. Cells utilize decarboxylation reactions
to form CO2 and to preserve energy in NADH.
Oxygen is then used to extract chemical potential

energy from NADH and FADH2, in order to
generate ATP that can be used in cellular work.
The participation of oxygen in the oxidation of
NADH/FADH2 generates the six O2 molecules
required to oxidize one glucose, as shown in the
stoichiometric equation of glucose oxidation.
Fig. 3.3: Structure of key compounds in glycolysis.
Fig. 3.4: Structure of key compounds in TCA cycle

and glutamine metabolism.
Extraction of the chemical potential energy of

NADH and FADH2 takes place through an electron
transfer chain residing in the mitochondrial inner
membrane. The high-energy electrons of NADH
and FADH2 enters the electron transfer chain
down the energy ladder, mediated by electron
carriers such cytochrome C. The energy released
is then used to trigger a proton pump to drive
H+ out of the mitochondrial inner membrane. In
the last step of the electron transfer chain, the
electron reacts with oxygen and H+ to form water.
The export of H+ from the mitochondria creates a

single unit pH difference across the membrane, as
well as about -120 mV of electric potential. Because
of the higher pH (lower proton concentration) and
excess negative charge inside the mitochondrial
membrane, there is a propensity for the proton ions
outside the mitochondria to cross the mitochondrial
membrane. They enter the mitochondria through
an ATP synthase embedded in the inner membrane
of mitochondria. Through the process, the act of
a proton passing through ATP synthase brings an
ADP and a phosphate together to synthesize ATP.
The electron transfer and the generation of ATP are
often referred to as oxidative phosphorylation.
The amount of ATP generated per mole of glucose

varies somewhat among different species because
of their variable expression of ATP synthase.
However, in general, the number of ATP generated
per mole of glucose is about 30 32 for mammals.
The older literature tends to list the number as
36 moles of ATP / mole of glucose. Under some
physiological conditions, the electron transfer chain
and oxidative phosphorylation are uncoupled.

Instead of generating ATP, the energy from NADH
is released as heat to maintain body temperature.
The amount of energy, two ATP and two NADH (or

the equivalent of six ATP, since one NADH in the
cytosol can be roughly considered to be two ATP),
produced from splitting glucose into two moles of
pyruvate is only about 1/6 of what can be generated
from completely oxidizing glucose to CO2 and H2O. In
glucose oxidation, the majority of energy conversion in
glucose metabolism therefore occurs in the hundreds
of mitochondria in the cell and not in the cytosol.
Pentose Phosphate Pathway
PPP
Two segments:
Oxidative: remove 1 CO2 from glucose-6-

phosphate,
Generate 5 carbon sugar phosphate for
synthesis of nucleotides and other compounds
produce 2 NADPH
Molecular transformation
Interconverts 5 carbon sugar phosphate to 3
carbon and 6 carbon
To allow NADPH and 5 carbon sugar to be
produced at different ratios
NADPH is important in biosynthesis and in
neutralizing ROS
PPP is an important shunt from glycolysis that

supplies five carbon sugars and NADPH. PPP is
divided into two segments: an oxidative segment
and a monosaccharide transformation pool.
In the first segment, glucose-6-phosphate from

glycolysis is oxidized and then decarboxylated to
form the five-carbon ribulose-5-phosphate and two
NADPH. The five-carbon sugar phosphate is used
in nucleotide (such as ATP and dATP) synthesis
to supply the building units for RNA and DNA.
Cells use two different nicotinamide-adenine

dinucleotides as reductive chemical potential
energy carriers: NADH and NADPH. NADH is used
to store chemical potential energy in glycolysis,
the TCA cycle, and lipid catabolism. Eventually,
NADH is used to derive ATP in the mitochondria.
NADPH, on the other hand, carries a chemical potential
that is used in biosynthetic reactions, such as in the
synthesis of lipids, nucleotides, etc. NADPH is also
used to reduce oxidized glutathione and to regenerate
it. The reduced form of glutathione is important in
maintaining the cells reductive environment and in
the suppression of reactive oxygen species (ROS).
The second segment of PPP is a molecular conversion

pool that allows a two-carbon aldehyde unit or
three-carbon keto units to be translocated among
a number of three-carbon to five-carbon aldoses.
This allows the interconversion of carbohydrate

molecules that are three to seven carbons in length.
This mixing pool enables five-carbon sugars
from the first segment of PPP to be connected to
glycolysis through six-carbon fructose-6-phosphate
or
three-carbon
glyceraldehyde-3-phosphate.
Lactate Formation
Aerobic Glycolysis
Cultured cells and cancer cells undergo glycolysis
and produce lactate even at high oxygen
concentration
This propensity toward lactate production is not
for lack of oxygen (anaerobic glycolysis)
At high glycolysis flux, not all NADH can be
oxidized by electron transfer in mitochondria
Lactate production serves to regenerate NAD
continuous supply of NAD is crucial for glycolysis
to proceed
The first segment of PPP generates five-carbon

ribulose and NADPH at a molecular ratio of 1:2.
However, cells do not always need those two
compounds at a 1:2 proportion. The molecular
conversion in the second segment allows the net flux
from glycolysis to PPP to vary to meet the cellular
demand of ribose and NADPH at different proportions.
Under anaerobic conditions, some bacteria and

yeast produce ethanol or lactate. In the absence of
oxygen, the electron transfer chain does not operate
because no oxygen is available to receive the
electron from NADH. When NAD is not regenerated
by NADH oxidation, the TCA cycle ceases to
operate. The accumulated pyruvate from glycolysis
is then excreted as lactate or ethanol. (Note
that we denote NAD+ as NAD in text without the
superscript + associated with its positive charge).
Mammalian cells in culture take only a small portion

of pyruvate generated in glycolysis into their
mitochondria, for further oxidation to CO2. They
appear to have a limited capacity to translocate
pyruvate into the mitochondria. The rest of glucose
is converted to lactate. This occurs in spite of the
presence of sufficient oxygen. The phenomenon is,
thus, different from anaerobic fermentation in bacteria
or yeast, and is referred to as aerobic glycolysis.
Not all cells in our body convert a large portion
of the glucose they take up into lactate. The vast
majority of cells in our body are in a quiescent (nonproliferating) state. They consume less glucose than
proliferating cells. The contrast in cellular glucose
metabolism, known as the Warburg effect, was first
observed between normal tissues and cancer cells.
While normal cells have a lower glucose flux, cancer
and proliferating cells consume a larger amount of
glucose and convert much of the glucose to lactate.
Energetic Yield of Aerobic Glycolysis

1. Oxidation
Glu cos e + 2ADP + 2Pi + 2NAD+

+ 2ATP + 2NADH
2Pyruvate
2. Reduction [oxidizing pyruvate to lactate (or ethanol as

in yeast)]
2Pyruvate + 2NADH
2Lactate + 2NAD +
Net Reaction:
Glu cos e + 2ADP + 2Pi
2 Lactate + 2ATP
Lactate synthesis is catalyzed by lactate

dehydrogenase.
This reversible reaction
takes
one
pyruvate
and
one
NADH
to become one lactate and one NAD.
In glycolysis, two ATP and two NADH are

generated, along with two pyruvate. Continued
glucose metabolism through glycolysis requires
continued supplies of both ADP and NAD as
reactants. ATP is used by cells to perform many
tasks, such as synthesis, maintaining osmotic
balance, etc. It is continually being consumed in
various cellular reactions and converted back
to ADP to resupply the reactant for glycolysis.
NADH is converted back to NAD through the electron

transfer chain in mitochondria. To be regenerated in
the electron transfer chain, cytosolic NADH must first
enter the mitochondria and the regenerated NAD
must be exported out of the mitochondria. A reaction
with lactate dehydrogenase allows NAD regeneration
from NADH to be carried out into the cytosol,
thereby enabling glycolysis to continue at a high flux.
Cells in culture convert about 90% of their glucose

to lactate. Most of the other 10% of glucose is
converted to CO2. At the completion of glycolysis,
two ATP are generated while about 30 ATP are
generated, following complete oxidation. The
90% of glucose converted to lactate generate
1.8 ATP (2 ATP x 0.9), while the other 10%
generate 3 ATP (30 ATP x 0.1). Aerobic glycolysis
of proliferating cells generates a significant
amount of total energy to allow for proliferation.
Carbon Flow and Reaction

Intermediates
Pyruvate is a controlling node

Generation rate of pyruvate is balanced by entry into
the mitochondrion and conversion to lactate
Reduction of pyruvate to lactate recycles NAD+ for
glycolysis to continue
Lactate dehydrogenase (LDH) is reversible
Transport of lactate by monocarboxylate transporter
which is coupled to proton gradient
Lactate dehydrogenase reaction

Pyruvate + NADH
Lactate + NAD+
Among all of the pathways in the cellular metabolic

reaction network, glycolysis has the highest flux
in terms of moles of substrate flowing through
it. For cells in culture, carbon flux (based on
the number of moles of carbon atoms, or the
number of carbons in the compound multiplied
by the number of moles of the compound) or
molar flux (based on the number of moles of each
compound) of glycolysis is normally a few times
higher than that of the TCA cycle. PPP flux usually
constitutes only about up to 5% of glucose intake.
Glycolysis and the TCA cycle also supply precursors

to build cellular components. Culture media do not
necessarily supply cells with the right balance of
all of the components that they need to synthesize
cell mass. The three main pathways for energy
metabolism also serve as key distribution centers
for carbon skeletons needed for other cellular
functions. Glycolysis supplies glycerol phosphate,
which is for the synthesis of phospholipids.
Glucose-6-phosphate and fructose-6-phosphate
are both a source of nucleotide sugars for glycan
synthesis, such as UDP-galactose, UDP-glucose, and
GTP-mannose. Except for liver cells (hepatocytes),
cells in culture have little gluconeogenesis activity;
that is, they cannot make hexose from lactate or
amino acids. So, even if cells can derive energy
from lactate and amino acids, they will still
need hexose to synthesize ribose and glycans.
Cells in culture take up a large quantity of amino
acids, especially glutamine. The intake of amino
acids exceeds what is needed to make cell mass and
product. The surplus of nitrogen is either excreted
as ammonia or transferred to pyruvate to form
alanine, and excreted. Since alanine is much less
growth inhibitory than ammonium, pyruvate has
some moderating effect on ammonium toxicity.
NADH Balance
While glucose is oxidized in Glycolysis and TCA cycle,
its carbons never react directly with O2
The energy is preserved to NAPH/FADH2, which is
then reacted with O2 in oxidative phosphorylation in
mitochondria to generate ATP
Altogether 12 reducing equivalents (NADH/FADH2) are
generated to react with 6O2, generating 6H2O
2 of the 12 reducing equivalents, two are generated in
cytosol (in glycolysis) 10 in mitochondria
Glucose 6 Phosphate
NAD
NAD
NADH
NAD
Pyruvate
Lactate
malate NAD
aspartate
shuttle
reducing
equivalent
NADH
cytosol
mitochondrion
oxidized
NAD
Election transfer
chain
Pyruvate
Equal moles of
Pyruvate and
NADH reducing
equiavlent enters
mitochondria
NADH- Pyruvate Balance
Fig. 3.5: NADH/NAD balance in cytoplasm. NADH generated

in glycolysis is recycled back to NAD via lactate production
and malate-aspartate shuttle to transfer the reducing
equivalent into mitochondria.
A total of 12 moles of reducing equivalent (10

NADH and 2 FADH2) are produced when 1
mole of glucose is oxidized to CO2. The 12 mole
reducing equivalents consume 6 moles of O2 in
oxidative phosphorylation, consistent with the
stoichiometry of glucose oxidation (1 glucose/6
O2). Among the 12 NADH/FADH2, 10 are produced
in the mitochondria and the other 2 NADH are
produced in cytosolic glycolysis. The two reducing
equivalents produced in the cytosol must then be
transported into the mitochondria where they drive
the reaction that consumes the 6th molecule of O2.
NADH does not pass through the inner membrane

of mitochondria. Rather, it passes its reducing
potential through a carrier system called the malateaspartate shuttle. This system takes the reducing
equivalent into mitochondria through an exchange
of molecules between the mitochondria and the
cytosol. On the cytosolic side, NADH is oxidized
to NAD and transfers its reducing equivalent to
malate by reducing oxaloacetate (OAA). Malate
is then transported across the mitochondrial
membrane. Once inside the mitochondria, the
reducing equivalent in malate is transferred back
to NADH by being oxidized to again become OAA.
The net result of the transfer must be that only one

NADH in the cytosol can become one NADH in the
mitochondria. To maintain that balance, cells employ
two transporters: one for transporting malate and
the other for transporting aspartate (hence the
name malate-aspartate shuttle). To maintain the
charge balance in the transport process, each of the
two transporters transfers a pair of compounds in
opposite directions (malate and -ketoglutarate;
aspartate and glutamate). In this fashion, there
is no net change in total carbon or nitrogen on
either side of the mitochondrial membrane.
On each side of the membrane, the same glutamateOAA to -ketoglutarate-aspartate amino-transfer
reaction occurs, but it is in the opposite direction.
The transfer of the reducing equivalent of NADH into
the mitochondria is therefore not only dependent

on the NADH concentration but also linked to amino
acid metabolism and the activities of the TCA cycle,
through the concentrations of associated carriers.
Glycolysis and the TCA cycle are thus connected by a
cytosolic balance of pyruvate and NADH. The flux of
both pyruvate and NADH are stochiometrically linked
in glycolysis, as well as through the LDH reaction. As
a result, the fluxes of pyruvate and NADH entering
the mitochondria are also stoichiometrically
related. Regardless of whether cells are at a
lactate-producing or -consuming state, the molar
ratio of pyruvate flux to NADH flux is always one.
1,2-Bisphospho
glycerate
Glyceraldehyde
3-Phosphate
Glucose
NAD+
Pyruvate
NADH
NAD+
QAA
Lactate
Malate-Aspartate
Shuttle
Aspartate
Mitochondria
Glu
Aspartate Glu
KG
Malate
KG
Malate
NAD+
Pyruvate
NADH
OAA
Oxidative Phosphorylation
ATP
Fig. 3.6: Detailed reactions of malate-aspartate shuttle.
Each enzyme reaction/co-transportation is given in the same color

Direction of reaction in NHDH reducing equivalent transport is indicated by an arrow
Net transport of one reducing equivalent-1 NADH in cytosol, and +1 NADH in mitochondria
Transport of 1 malate + 1 glutamate into mitochondria and 1 KG + asparate in cytosol
Fig. 3.7: Transfer of NADH reducing equivalent from cytoplasm into mitochondria requires
continuous transport of four malate, aspartate, glutamate and -ketoglutarate.
Fig. 3.8: Biochemical reactions in glycolysis, TCA cycle and Pentose phosphate pathway
Energetic Yield of Oxidation of Glucose

Cytosol
glu cos e
2pyruvate + 2NADH + 2ATP
Mitochondrion
2pyruvate
2acetylCoA
2acetylCOA + 2NADH + 2ATP + 2CO2

6NADH + 2FADH2 + 2GTP + 4CO2
The overall energetic yield is ~30ATP, considering the cost of ATP transport out of mitochondria
The NADH generated in cytosol (Glycolysis) is recycled back to NAD+, either through reduction of
pyruvate to lactate, or by NADH/NAD+ shuttle into mitochondria for oxidation.
Role of Transport and Transporters in Metabolism
Glucose Transporters
Two Main Types of Glucose Transporters
GLUT transporters mediate facilitative diffusion across
plasma membrane.
SGLT, the sodium dependent glucose co-transporters
are expressed primarily in small intestinal absorptive
cells or renal proximal tubular cells. They use Na+-K+
ATPase pump for active transport of glucose.
Glucose Transporter Isoform

GLUT1 is highly expressed in all cells
Km is small for GLUT1. At culture glucose concentration
it operates at maximum rate.
Energy metabolism takes places in multiple

compartments, separated by lipid bilayer membranes.
First, glucose must cross the cytoplasmic membrane
to undergo glycolysis in the cytosol. The products
from glycolysis (pyruvate and NADH, in the form
of reducing equivalents) are transported into the
mitochondria. This exchange of molecules across the
membrane is mediated by membrane transporters.
Glucose
transporters
mediate
the
influx
of glucose across the cytosolic membrane.
Generally speaking, there are two types
of glucose transporters: GLUT and SGLT.
The GLUT transporters are uniporters for

facilitated transport, allowing glucose to move
along its concentration gradient. They have twelve
transmembrane regions and intracellular carboxyl
and amino termini. According to common sequence
motifs, they are divided into three subclasses.
GLUT1 is ubiquitous, found in almost all cells. It can

transport glucose and galactose in a concentrationdependent manner that is described by MichalisMenten kinetics. The Km for glucose is very low (1
2 mM). At the glucose concentration used in culture
medium, the flux of GLUT1 is at its maximum. In
some cells, GLUT1 is under the regulation of the
transcription factor HIF-1 (hypoxia inducible factor).
Under hypoxic conditions, the expression of GLUT1
is up regulated to increase the uptake rate of glucose.
The Km of GLUT1 for galactose is rather high.
When galactose is used as the only sugar, even
at a concentration of 25 mM, the uptake rate
is so low that only a little lactate is produced.
Fig. 3.9: Michaelis-Menten kinetics plot for GLUT1 transporter
A few other notable GLUT transporters are: insulin

responsive GLUT4 and fructose transporting GLUT5.
In addition to GLUT1, cells in culture and in different
tissues may express other GLUT transporters at
different proportions. The expression of different
transporters will give them different responses
to the concentration of glucose or other sugars.
Table 1: Glucose Transporters

Tissue
Expression
ubiquitous
glucose, K m = 1 - 2 mM
Liver, pancreas,
intestine, kidney
glucose, K m = 16 - 20 mM
glucosamine K m = 0.8 mM
glucose, K m = 0.8 mM
GLUT5
brain, neurons
heart, muscle,
adipose
intestine, testis
GLUT7
intestine, testis
GLUT9
kidney, liver
GLUT11
heart, muscle
brain, spleen,
leukocytes
testis, brain, liver
liver, pancreas
heart, muscle,
prostate
GLUT1
Class 1
GLUT2
GLUT3
GLUT4
Class II
GLUT6
Class III
GLUT8
GLUT10
GLUT12
Affinity
glucose, K m = 5 mM
fructose K m = 10 - 13 mM
glucose, Km = 0.3 mM
fructose K m = 0.1 mM
fructose K m = ? mM
fructose K m = ? mM
glucose, K m = 5 mM
glucose, K m = 6 mM
glucose, K m = 0.3 mM
not well known
Lactate Transport
Fig. 3.10: Monocarboxylate transporter for lactate

and pyruvate
The second type of glucose transporter, SGLT, is a

co-transporter with Na+. It transports two sodium
ions and one glucose molecule into the cell. The
Na+ concentration is low intracellularly but is high
in the medium and in body fluid. The large sodium
concentration difference and negative electric
potential across the cytoplasmic membrane gives
rise to a high propensity of Na+ to enter the cell.
Thus, the chemical potential energy of the sodium
gradient and electric potential is used to drive the
uptake of glucose against a concentration gradient.
SGLT transport is abundant in intestinal epithelial
cells and is responsible for moving glucose
from the gut into the intestinal epithelial cells.
The glucose is then exported into the blood
stream on the other side of the cellular barrier.
Lactate and pyruvate are transported by

monocarboxylate transporters. These transporters
exist in two forms: one on the cytoplasmic membrane
and one on the mitochondrial inner membrane.
Lactate and pyruvate are both negatively charged.
Their movement across the cellular membrane will
cause a charge unbalance and create an electric
potential, unless measures are taken to counter
that imbalance. The monocarboxylate transporters
(MCT), which are responsible for their transport, are
a family of co-transporters that couple the transport
of lactate or pyruvate to the transport of a hydrogen
ion in the same direction to neutralize the charge
transfer. MCT is thus a symporter; its mechanism of
transport is facilitated diffusion.
Lactate transport is enhanced by a large difference
in lactate concentration between intracellular
and extracellular environments. pH also affect the
flux of lactate through MCT, however, whether the
Transport Across Mitochondria
effect is enhancing or retarding is dependent on the

direction of the proton gradient. MCT allows for
lactate transport in both directions, for excretion as
well as uptake. Keeping medium pH at a lower level
reduces lactate production during rapid growth
period, but enhanced lactate consumption in the
stationary phase.
Cells in culture typically channel about 1/20 of the

carbons from their glucose intake to the TCA cycle
The chemical potential energy generated by pyruvate
and oxidize them to CO2. The molar flux of pyruvate
oxidation/TCA cycle, i.e. NADH and FADH2, is not
necessarily all converted to ATP. The process can be into the mitochondria is thus about 1/10 of that of
decoupled to generate heat instead of ATP, as occurs in the glucose consumption rate. Each mole of pyruvate
hibernating mammals.
entering the TCA cycle via acetyl CoA generates
The mitochondrion is also the main site of molecular about 15 moles of ATP. These are exported to the
interconversion and degradation of amino acids and cytosol and require the import of equal moles of ADP
the main source of acetyl CoA. The excess glutamine and PO 3- for their synthesis. Each mole of pyruvate
4
consumed enters the TCA cycle through -ketoglutarate.
generated from glycolysis or lactate consumption
For some cells, asparagine acts as a sink (in addition
to alanine), which is formed through oxaloacetate and is also accompanied by one mole of NADH, whose
reducing equivalent is transferred into the
aspartic acid.
mitochondria through the malate-aspartate shuttle.
Additionally, cells in culture consume glutamine at
a high rate (approximately 1/5 to 1/10 of glucose
in molar ratio). Nearly half of the glutamine enters
the TCA via -ketoglutarate. CO2 produced in the
TCA cycle is then exported out of the mitochondria.
Besides these major species, many other molecules
(including amino acids and nucleotides) are
transported into the mitochondria for DNA, RNA,
and protein synthesis. As will be described later, the
precursor for fatty acid and cholesterol synthesis,
acetyl CoA, is generated in the mitochondria, while
fatty acid and cholesterol synthesis occurs outside
the mitochondria. Acetyl CoA is very reactive
and does not get transported directly across the
inner membrane of the mitochondria. Rather, it
is transported out of the mitochondria as citrate.
After cleaving off acetyl CoA, the remaining four
carbons are returned to the mitochondria as
pyruvate or malate. Thus, the citrate and malate
flux across the mitochondrial membrane is also
substantial to sustain lipid and cholesterol synthesis.
The transport across the mitochondrial inner
Regulation of Glucose Metabolism
membrane is dynamic and complex. Many

compounds
crossing
the
membrane
are
charged, yet their transport must not perturb
the proton and electric potential gradient. The
transport across the mitochondrial membrane
must be tightly regulated. Our understanding
of its regulation is still rather limited.
Under physiological conditions, the flux of glycolysis

and the TCA cycle is not controlled by one or a small
number of rate-limiting enzymes. Glucose flux is
the result of mutual constraints of many enzymes
in the pathway, through their feed-forward and
feedback inhibition and activation. A large number
of pathways are highly inter-connected and crosstalk
with each other through shared common substrates
or regulators. In mammals, different tissues serve
different metabolic roles to maintain the overall
homeostatic state of the organism. The partition of
metabolic roles is largely accomplished by giving
different tissues a different set of isozymes. Different
isozyme sets allow cells to respond to environmental
fluctuations
or
cellular
cues
differently.
Isozymes and Differential Allosteric Regulation
Cells express different isozymes in different tissues,

under different conditions
different isozymes have different kinetics and
regulation
Different isoforms (isozymes) of glycolytic

enzymes catalyze the same reaction step, but have
very different kinetic properties and are often
subjected to contrasting regulations. Isozymes
of four glycolysis enzymes, hexokinase (HK),
phosphofructokinase (PFK), pyruvate kinase (PK),
and phosphofructokinase/fructose biphosphatase
(PFKFB), play key roles in controlling glycolysis
flux. Their different allosteric regulations give
them very distinct reaction characteristics.
The isoforms of those enzymes making up glycolysis

are different in proliferating and quiescent cells.
They are thought to be responsible for endowing
the high glycolysis flux and high lactate production
in proliferating and cancer cells and are thought
to be related to oncogenic transformation. They
also give tissues their metabolic capabilities.
Fig. 3.11: Allosteric regulations in glycolysis. Note strong

activation of glycolysis flux by F16BP and inhibition by
lactate.
PFK is pivotal in modulating the overall rate of

glycolysis and is a key node in energy metabolism.
Its activity is subjected to allosteric inhibition
by ATP and citrate, and is activated by AMP. PFK
has three isozymes: liver (PFKL), muscle (PFKM),
and platelet (PFKP). Among the three isoforms
of PFK, the muscle and the liver isozymes are
activated allosterically by fructose 1,6-bisphosphate
(F16BP). Muscle phosphofructokinase is inhibited
by lactate, a characteristic which may facilitate
reduced glycolysis flux at high lactate levels.
Fructose-2,6-phosphate is a shunted glycolytic

intermediate that plays a key regulatory role in rapidly
modulating the activity of PFK. All three PFK isozymes
are activated by fructose 2,6-bisphosphate (F26BP).
F26BP activates PFK1 by allosterically increasing its
affinity for F6P, even in presence of inhibitors such
as ATP, lactate, etc. The synthesis and degradation
of F26P is catalyzed by a bi-functional enzyme,
PFKFB (also known as PFK2); its kinase activity
catalyzes the synthesis of F26P and its phosphatase
activity catalyzes the hydrolysis of F26P to F6P.
PFKFB has four isozymes, each with different
kinase and phosphatase activities, allowing each to
F2,6P plays a key regulatory role

Key regulatory enzyme of glycolysis
PFK (PFK1)
PFK2
PKM
Cancer cells (fast growing cells) and quiescent
cells have different isozymes
respond differently to regulators. The brain isoform,

PFKFB3, has the highest kinase to phosphatase
activity and is expressed in several tumor cells. This
suggests that PFKFB3 can be accountable for the
glycolytic phenotype of reported cancerous cell lines
by allowing them to have high cellular F26P levels.
The enzyme catalyzing the penultimate step of

glycolysis, pyruvate kinase, has three isozymes
in mammalian systems. The muscle isozyme is
expressed as either of the two splice variants, M1
or M2. The M1 isoform is mostly expressed in
adult tissues whereas the M2 isoform is expressed
exclusively in rapidly growing tissues, such as
fetal and tumor tissues, and also is thought to be
a critical player in the transformation leading to
cancer. In addition, the M2 isoform is known to be
under positive feed forward regulation by F16P.
Isozymes are often named after the tissue in which they
are the dominant isoform. However, it is important
to remember that the expression of isozymes is not
limited to one form in a cell type. Different isoforms
are often expressed in the same tissue or the same
cell. Different combinations of isozymes give rise
to different kinetics and regulatory behaviors
that may meet different physiological needs.
With the available genomic tools, we can easily

determine the relative expression of different
isoforms of the key enzymes of glycolysis, and further
evaluate how to influence cellular metabolism.
Signaling Pathways and Regulation by

Growth Control
The regulation of cellular metabolism is tightly

linked to the control of cell growth. The signaling
pathways that regulate growth rate (involving key
Signaling pathway/growth rate control of players like p53 and cMyc) also play regulatory roles
Glycolysis
in regulating glucose metabolism. Transformation
Insulin signaling
that causes cells to switch from a quiescent state to
a proliferating state also triggers metabolic changes
positively regulating growth rate
through AKT regulate glucose, amino acid metabolism to increase their glucose uptake and glycolysis flux.
P53 (tumor suppressor) supresses glucose uptake and p53 is a major tumor suppressor that plays key
glycolysis
roles in cell cycle arrest, senescence, and apoptosis.
Its role in the regulation of glucose metabolism and

Fast growing (tumor) cells have fast glycolysis (and oxidative phosphorylation was not well understood
until only recently. It induces the overexpression of
lactate production)
TIGAR under mild oxidative stress conditions. TIGAR
contains a fructose-2,6-bisphosphatase catalytic
Regulation of Metabolism
activity domain, which mediates the degradation of
Glucose
fructose 2,6-phosphate, leading to a decrease in PFK1
GLUT1,4
p53
activity and the attenuation of the glycolytic flux.
cMyc (proto-oncogene) stimulates glycolysis
Glucose
H2O
Glucose-6TIGAR Phosphate
Fructose-2,6Bisphosphate
PPP
NADPH
ROS
Fructose-6Phosphate
PFK2
PFK1
Transcriptional Inhibition
Transcriptional Activation
Allosteric Activation
p53
PGM
Lactate
Pyruvate
NADH
SCO2
p53
COX
ATP
Mitochondria
Fig. 3.12: Tumor suppressor p53 negatively regulates

glycolysis flux.
Akt and Myc Regulation of Metabolism

Glucose
Akt
GLUT1
Glucose
HK
Glucose-6Phosphate
Activation by phosphorylation
or localization
Transcriptional Activation
Allosteric Activation
PFK2 Fructose-6Phosphate
PFK1
Myc
LDH
Pyruvate
Malate
Lactate
GLS
Glutamate
Glutamine
TCA
ASCT2
SN2
Glutamine
(Extraacell
ular)
p53 can also modulate glycolytic rate by regulating

the activity of PGM, GLUT1, and GLUT4 transporters.
Furthermore it up-regulates mitochondrial
oxidative phosphorylation by upregulating the
expression of SCO2 (synthesis of cytochrome c
oxidase 2), which mediates the assembly and
activity of the cytochrome c oxidase complex.
Pro-oncogenic genes may also provide the link

between increased glycolytic flux and oncogenic
transformation. Akt is a pro-oncogene which,
when deregulated, affects many cellular functions
including metabolism, cell proliferation, and protein
synthesis. In tumor cells, constitutively activated
AKT is sufficient to shift the cells from a primarily
oxidative state to a primarily glycolytic state.
Myc
is
another
proto-oncogene
whose
pleiotropic regulatory roles include energy
metabolism. Glycolytic enzymes have Myc
canonical E-boxes in their promoter and are
deregulated when Myc is overexpressed.
Mitochondria
Fig. 3.13: Signaling kinase, AKT and transcription factor, Myc,

positively regulate energy metabolism.
Metabolic Homeostasis and Lactate Consumption

Cells in culture, including those used in bioprocessing,
have all been selected for their capability to proliferate.
Their particular set of glycolytic isozymes directs
their metabolism to a high glucose flux and a high
rate of lactate production. However, even the exact
composition of isozymes is not monolithic among
different cells. Most cells express multiple isoforms
of the same enzyme in glycolysis and their proportion
is not identical. Consequently, even though the
Metabolism is a balancing act of many constraining metabolism of all cultured cells share the common
trait of a high glucose flux and high lactate production,
reaction nodes
their metabolic behaviors are not all identical.
Glycolysis flux influenced by:
Lactate accumulation in culture inhibits cell growth
Regulation (feed back and growth)
and hastens the decline of cell viability in the
Pyruvate/NADH flux into mitochondria
stationary phase. A key to achieving a high cell
LDH (NAD recycle rate)
concentration and high productivity is to direct cell
Lactate export
metabolism to minimize the accumulation of lactate.
Glucose intake
One may aim to alter the cells metabolism to reduce
or eliminate lactate production. However, it is
important to keep in mind that virtually all rapidly
proliferating cells produce lactate. Such a metabolic
state might be a default state that is essential
for cell growth and cannot be easily altered over a
long period without affecting growth behavior. It
is also important to keep in mind that through in
the course of culturing cells for manufacturing, the
duration of the actual production period is only
a very small fraction of the total process lifetime.
The long duration of the process time is mostly for
expanding cell numbers to reach a production scale.
Altering the metabolism from the cells default
state may have an unforeseen effect on cells.
Another approach for alleviating lactate inhibition
is to tamper with the cell metabolism only in the
production scale or in the final stages of production.
Increasing evidences show that the switch from
lactate production during the growth stage to
lactate consumption in late stages is correlated
with a high productivity. One may seek to alter
cell metabolism to change lactate consumption
in the last stage to be more reproducible.
The LDH reaction that catalyzes pyruvate to lactate

is reversible. The direction of the flux depends
on the concentrations of NAD, NADH, pyruvate,
and lactate. Those concentrations, in turn, are
related to fluxes of reactions that produce or
consume them, including pyruvate entry into
mitochondria and lactate export (or import).
Further observations related to lactate consumption:
Cycle
Lactate consumption occurs in post-rapid stages of

growth, when the glucose consumption rate is low.
The pyruvate flux into the mitochondria is rather
limited. When cells are at a high glucose flux, the excess
pyruvate must be converted into lactate to regenerate
NAD+. Therefore, a high glucose flux state is always
linked to a lactate production state. Furthermore,
rapid proliferation is generally associated with
high glucose flux. Lactate consumption occurs
only when the growth rate is slow. Experimental
observation has demonstrated that the absence
of rapid growth and a low glucose consumption
are necessary conditions for lactate consumption.
Glucose is still being consumed while cells

consume lactate. While lactate is being taken
up by cells, its specific consumption rate is
relatively small. Lactate is converted to pyruvate,
and then it enters the mitochondria for further
oxidation and energy generation. Even though
energy is being derived through lactate oxidation,
cells still need many constituents derived from
glucose for glycan synthesis, including NADPH
(from PPP), glucosamine, and galactose. These
compounds cannot be derived from pyruvate in
most culture cells, since they lack key enzymes for
gluconeogenesis. They must be supplied through
glucose. The glucose consumption rate in the lactate
consumption stage is small, but it is never zero.
Cycle
Fig. 3.14: Metabolic fluxes in lactate production and consumption metabolism.
The conversion of lactate to pyruvate by LDH does

not occur in isolation. It is coupled to the reduction
of NAD to NADH (also catalyzed by LDH), and linked
to the reverse transport of lactate and H+ by MCT
from the medium to the cytosol. The propensity and
rate of lactate consumption is affected by the pH
of the media. In addition to transporting pyruvate
formed from lactate into mitochondria, NADH

has to be transported from the malate-aspartate
shuttle into the mitochondria for oxidation.
Glutamine and Its Relation to Energy Metabolism
For most cultured cells, glutamine is the second

highest consumed nutrient. Its molar consumption
Glutamine
rate is about 1/5 to 1/10 of that of glucose for most
Consumed by growing cells at a high rate
cells. Glutamine is a major amino acid constituent
Non-essential in vivo. Glutamine synthesase decreases of cellular proteins. It supplies amino groups for the
upon in vitro culture
synthesis of purine and pyrimidine bases, which are
Supply TCA cycle intermediates
the backbone of nucleic acids. However, the amount
causes NH3 release
of glutamine consumed by cells far exceeds what
is needed for synthesizing cellular components.
used in nucleotide and protein synthesis
Glutamine is not an essential amino acid for
mammals; it is only essential for cultured cells. Cells
in some tissues, like the liver, synthesize glutamine
by the incorporation of an ammonium into glutamate
at the expense of an ATP using glutamine synthase.
The transcript level of glutamine synthase in cultured
cells is low. It actually decreases by nearly two orders
of magnitude when liver cells are put into culture.
A large portion of glutamine is converted to glutamate

by glutaminase, and then to -ketoglutarate before
entering the TCA cycle. Through -ketoglutarate,
glutamine is a major contributor to central metabolic
flux. There is increasing evidence that glutamine
is needed to drive the TCA cycle for faster growth.
Fig. 3.15: Entry of glutamine into TCA cycle. Glutamine

is converted to glutamate by glutaminase. Glutamate is
then convert to -ketoglutarate by either transamination
reaction (transfer its -amino group to pyruvate or
oxaloacetate) or via oxidation by glutamate dehydrogenase
(converting NADP to NADPH and releasing ammonium).
In the course of being converted to glutamate

and -ketoglutarate, the nitrogen is released
as ammonium or transferred to an amino acid,
such as alanine or asparagine, and excreted.
The ammonium that is released from glutamine
contributes to the waste metabolite accumulation.
Amino Acid Metabolism

Mammals can synthesize only fewer than 10
of the 20 amino acids used in the translational
synthesis of proteins. 11 or 12 (depending on
the species) essential amino acids that they
cannot synthesize must be acquired through diet.
Fig. 3.16: Major amino acid transporters
Amino acids are transported with overlapping

transporters.
All IgG antibodies produced by mammalian cells
are glycoproteins, with an N-linked oligosaccharide
attached to each heavy chain in the hinge region at
Asn-297
Amino Acid Degradation

Amino acids are consumed at a rate of three to five
times what is needed for making biomass and product.
Excess amino acid consumed must be excreted.
The nitrogen (amino group) is removed from the carbon
skeleton by transamination, oxidative or non-oxidative
deamination. The excess nitrogen is excreted as
ammonium ion or as excreted amino acids (e.g. alanine,
proline, asparagine)
The carbon skeleton mostly enters the TCA cycle to be
converted to excreted non-essential amino acids or to
be converted to pyruvate and lactate, or to contribute
to acetyl CoA in cytosol through citrate.
Large excess of glutamine consumed by cells in
culture is converted to glutamate in the cytosol,
or is transported into the mitochondria and then
converted to glutamate. Glutamate is then converted to
-ketoglutarate and enters carbon metabolism.
Glutamine (amide group) is used as an amino group
donor in adenosine (AMP), guanosine (GMP) and
Cytosine (CTP) biosynthesis.
Aspartic acid and glycine are also used in nucleic acid
synthesis
Methionine is methyl group donor. Tryptophan is used
in NAD synthesis
Glutamate participates in a large number of reactions.
The flux of its synthesis or supply is expected to be high.
Cells in culture have more essential amino acids

requirements than the organism; a number of
non-essential amino acids become essential for
in vitro culture, including glutamine, tyrosine,
serine under some growth conditions, proline
for some cells. All essential amino acids must be
supplied in medium, while non-essential amino
acids can be derived from other amino acids.
Amino acids are taken up by cells through a large
number of amino acid transporters. Most amino acid
transporters transfer a family of amino acids with
similar characteristics, such large neutral (uncharged
side chain) amino acids, cationic or anionic amino
acids. The rate of transfer for particular amino acids
is thus not only dependent on the concentration
difference of itself between intracellular
environment and extracellular medium, but also on
the competition with other amino acids for the same
transporter. One amino acid may be taken up via more
than one transporter, albeit with different affinity.
The amino acids taken up by cells are not likely

to be in the right stoichiometric ratios to
meet their need. Excess amount of amino acids
is converted to the non-essential amino acids
which are not supplied in sufficient quantities,
or is degraded. The interconversion and
degradation of amino acids takes place through
the intermediates of glycolysis and TCA cycle.
Before the carbon skeleton of an amino acid enters

carbon metabolism pathways its amino group or
other nitrogen containing functional groups is
first removed. Glutamine and glutamate become
-ketoglutarate, asparagine and aspartate become
oxaloacetate; all are catabolized through the TCA cycle.
Alanine becomes pyruvate, while leucine, isoleucine
and others enter carbon catabolic pathway through
acetyl-CoA. Although the degradation of amino

acids yields energy, it is not a major energy source.
Alanine
Cysteine
Glycine
Serine
Threonine
Tryptophan
Pyruvate
CO2
Acetyl-CoA
Acetoacetate
Isoleucine
Leucine
Lysine
Threonine
Citrate
Asparagine
Aspartate
Oxaloacetate
Aspartate
Phenylalanine
Tyrosine
Isoleucine
Methionine
Valine
Isoleucine
Leucine
Lysine
Threonine
Citric
acid
cycle
Fumarate
The excess nitrogen from amino acid degradation

is excreted as ammonia or as the amino group of
non-essential amino acid, especially as alanine or
proline. In some cell culture processes ammonium
accumulates to growth inhibitory levels. Although
the adverse effect of ammonium on cell growth
is usually not as severe as lactate accumulation.
Isocitrate
CO2
-Ketoglutarate
Succinyl-CoA
CO2
Arginine
Glutamate
Glutamine
Histidine
Proline
Fig. 3.17: Entry of amino acids into catabolism
Lipid Metabolism
Functions of Lipids
Contributes to the membrane fluidity
Storage of precursors metabolized to second
messengers (diacylglycerol, inositol triphosphate)
Lipids (such as, polyphosphoinositide) involved in
protein traffic and membrane fusion events
Anionic lipids (such as, PS) involved in attachment of
cytoskeletal proteins to membranes
Cholesterol and sphingolipids form microdomains
or rafts: enriched in specific subsets of membrane
proteins
Subcellular Localization of Lipid Metabolism in

Animal Cells
Cytosol
NADHP synthesis (pentose phosphate pathway, malic

enzyme)
Isoprenoid and early cholesterol synthesis
Fatty acid synthesis
Mitochondria:
Fatty acid oxidation
Lipids play key roles in many physiological functions

critical to cellular properties of particular interest
to bioprocessing. In addition to forming the bilayer
membrane, lipids are also involved in cell signaling.
Lipid content in bilayer membrane affects membrane
fluidity and permeability. Lipid bilayer membrane
partitions various organelles from the cytosol.
Secreted recombinant proteins are processed first in
endoplasmic reticulum (ER) and the Golgi apparatus.
They are excreted via membrane vesicles. For
optimal protein secretion capacity, the membrane
homeostasis and biogenesis among organelles,
secretory vesicles and plasma membrane is critical.
Not all lipid bilayer membranes are the same. The
lipid composition of ER, mitochondrial and plasma
membrane differ from each other. The plasma
membrane of hepatocytes is enriched in cholesterol;
however, the amount of cholesterol in the rough
and smooth inner ER is lower. There is very little
cholesterol in the inner mitochondrial membrane.
Acetyl-CoA synthesis
Ketone body synthesis
Fatty acid elongation
Endoplasmic Reticulum:
Phospholipid synthesis
Cholesterol synthesis (late stage)

Fatty acid elongation
Fatty acid desaturation
Peroxisome:
Cholesterol precursors on synthesis

Final steps of cholesterol synthesis also occurs in
peroxisome
Lipid Transport
Lipid Transport Processes
Intramembrane Transport:
Transbilayer movement of lipid molecule
Transbilayer movement at eukaryotic ER: relatively nonspecific, ATP-independent process
Three classes of transport:
Aminophospholipid translocases (also called
flippases)
Bidirectional transporter or scramblases
ATP Binding Cassette (ABC) pumps

Intermembrane Transport:
Movement of lipid molecules from one distinct membrane
domain to another
Possible mechanisms:
Monomer solubility and diffusion (for molecules such as
free fatty acids)
Soluble carriers such as lipid transfer proteins
Carrier vesicles
Although some cells can be cultured in lipid-free

medium, most cell culture media contain some
fatty acids and lipids. Cells readily take up fatty
acids, phospholipids, and cholesterol from the
medium and incorporate them into cellular lipids.
Lipids can be supplied as serum lipoproteins in the

form of a complex with albumin or liposomes, or as
solubilized conjugates, such as sorbitol-fatty acid
esters. The cellular uptake of fatty acids is a passive,
non-energy dependent process. After being taken
up by cells, fatty acids quickly become esters; the
intracellular levels of free fatty acids are quite low.
Cholesterol is complexed to low density lipoprotein
(LDL) in the body and is taken up by cells through
the LDL receptor. For cells in culture, cholesterol
is often supplied as a conjugate with serum
albumin, or as complexes with cyclodextrin.
Membrane apposition and transfer

Membrane fusion
Fatty Acid Metabolism
Most cells have the capability of synthesizing

various fatty acids. Under starvation conditions,
cells also perform -oxidation to degrade fatty
acids into acetyl CoA in the mitochondria or
peroxisomes. Acetyl CoA then enters the TCA cycle
to generate energy. Typically, cells in culture do not
need to derive energy from fatty acid oxidation.
Fatty acids are synthesized from acetyl CoA in
the cytosol. The first step of fatty acid synthesis
involves adding a CO2 to acetyl CoA to form malonyl
CoA, which then reacts with acetyl CoA to become
a four-carbon fatty acyl CoA. It is noted that even
though CO2 is a catabolic product, it is also an
essential nutrient for biosynthesis. Fatty acid
synthesis, thus, involves the step-wise elongation
processes of using three-carbon malonyl CoA to
add a two-carbon unit to fatty acyl-CoA in each
cycle. NADPH is also consumed in this process.
There are a number of fatty acid synthetases that
can synthesize fatty acids to different lengths.
The fatty acid products from elongation
reactions are all saturated fatty acids. Double
bonds are then synthesized by unsaturation
reactions after saturated fatty acids are made.
Acetyl CoA shuttle

Acetyl CoA shuttle
On mitochondria inner membrane
Citrate transporter: transport citrate into cytosol

Malate, -ketoglutarate transportor: transport into
mitochondria
Pyruvate transporter: transport into mitochondria
In Cytosol
Citrate lyase: citrate +CoA+ATPOAA+acetylcoA
Malate dehydrogenase: OAA+NADH+Malate +NADH+
Malic enzyme: malate+NADP Pyruvate +NADPH+CO2
In mitochondria matrix
Pyruvate carboxylase: Pyruvate+CO2+ATPOAA+ADP

Malate dehydrogenase: malate+NAD+OAA+NADH+
Citrate synthase: OAA+acetylCoACitrate +CoA
CO2
Pyr
NAD
Malate
NADPH NADP
Pyr
CoA
NADH
OAA
CO2
acetyl CoA
ADP+Pi
Citrate
Acetyl CoA is the building block of fatty acids.

It is generated primarily through the oxidative
decarboxylation of pyruvate in the mitochondria.
However, fatty acid synthesis takes place in the
cytosol. Acetyl CoA does not pass through the bilayer
membrane. Rather, it is exported to the cytosol via
an indirect process. Citrate, formed by condensation
of OAA and acetyl CoA in the TCA cycle, is then
transported into then cytosol, where it is split into
oxaloacetate and acetyl CoA. Oxaloacetate gets
reduced to malate at the expense of one NADH. Malate
is then converted to pyruvate, losing one carbon and
consuming one NADPH. Pyruvate then recycles into
the mitochondria. The process of making fatty acids
using acetyl CoA is thus energetically expensive.
Fatty acid
cholesterol
synthesis
ATP CoA
acetyl CoA
OAA
Citrate
KG
Malate
Fum
Suc
Glu
NHg
Glu
Gln
NH3
Fig. 3.18: Citrate and acetyl CoA shuttle.
Cholesterol & Its Biosynthesis
Fig. 3.19: The structure of cholesterol
Cholesterol is a 27-carbon molecule. Mammals

require cholesterol as a constituent of membranes
and as a precursor for the synthesis of steroid
hormones, bile acids, and lipoproteins. Cholesterol
is relatively insoluble and resides exclusively
in various cell membranes. Its regulation is
particularly important since excess cholesterol
forms solid crystals, leading to cell death.
Proper cell function requires that cellular
membranes have the appropriate composition,
Cholesterol metabolism is a tightly regulated pathway

subjected to stoppage of cholesterol biosynthesis in the
presence of excess cholesterol.
Animal cells obtain cholesterol from both de novo
biosynthesis and receptor mediated uptake.
Cholesterol

Component for membrane biogenesis and precursors

for the synthesis of steroid hormones, bile acids and
lipoproteins.
Cholesterol resides exclusively in cell membranes and

its regulation is particularly important since excess cholesterol forms solid crystals leading to cell death.
Cholesterol is ~10% of dry weight of plasma membranes. Precise lipid composition of the plasma membrane has been controversial since it is difficult to isolate high yield of pure samples from tissues or cultured
cells.
Cells obtain cholesterol by de novo synthesis or through

receptor-mediated uptake of plasma lipoproteins.
including cholesterol, in order to maintain bilayer

fluidity, impermeability and other characteristics
specific to different organelles. Cholesterol
constitutes ~10% of the dry weight of plasma
membranes, and plasma membrane cholesterol
accounts for 65% to 80% of total cellular cholesterol.
Cells in culture obtain cholesterol either by de
novo synthesis or through receptor-mediated
uptake of exogenous low-density lipoproteins.
Cholesterol is synthesized from acetyl CoA, which
is condensed by 3-hydroxy-3-methylglutaryl
(HMG)-CoA synthase (HMGCS) to form HMGCoA. HMG-CoA is converted to mevalonate by
HMG-CoA reductase (HMGCR).
This enzyme
is the target of statins, the class of drugs that
suppress cholesterol synthesis in patients.
Further synthesis of mevalonate to farnesyldiphosphate takes place in peroxisomes.
Subsequent condensation of two molecules of
farnesyl diphosphate to form squalene, lanosterol,
lathosterol and finally cholesterol occur in the ER.
Out of the 18 key enzymes taking part in

cholesterol biosynthesis, 5 enzymes reside in the
peroxisome and 13 reside in the ER. HMGCS is
upstream of HMGCR and is found in the cytosol.
Thus, there are at least three different sub-cellular
compartments involved in cholesterol biosynthesis.
Although cholesterol in mammals is synthesized

primarily in the liver, most cells have the
capability of synthesizing cholesterol for their
own growth requirements. NS0 cells lack an
enzyme, 17--hydroxysteroid dehydrogenase,
which converts lanosterol to lathosterol.
In
NS0 cells, 17--hydroxysteroid dehydrogenase
is silenced through methylation of a CpG island
upstream of its promoter, leading to the cell
lines dependency on cholesterol for growth.
Fig. 3.20: Cholesterol biosynthesis takes place in mitochondrion, peroxisome

and endoplasmic reticulum.
Cholesterol Biosynthtic Pathway

Cholesterol is synthesized from acetyl CoA, which is condensed by 3-hydroxy- 3-methylglutaryl (HMG)-CoA
synthase (HMGCS) to form HMG-CoA.
HMGCS exists in a mitochondrial form in hepatic tissue and is present in cytosol as 53kD protein for other
tissues.
HMG-CoA is converted to mevalonate by HMG-CoA reductase (HMGCR)
HMGCR exists as a 97-kDa glycoprotein in the endoplasmic reticulum (ER) and is exemplified as the rate
determining the enzyme of the biosynthesis pathway.
Mevalonate is further metabolized to farnesyl-diphosphate by a series of peroxisomal enzymes as shown in the
figure.
First committed step to cholesterol byosynthesis is typified by condensation of two molecules of farnesyl
diphosphate to form squanlene by farnesyl diphosphate farnesyl transferase-1 (Fdft1), or (squalene synthase) a
47-kDa protein residing in the ER
Squalene is converted to the first terol, lanosterol by the action of squalene epoxidase.
Conversion of lanosterol to lathosterol involves a series of oxidations, reductions and demethylations. It is a 17
Glycan Biosynthesis and Protein Glycosylation
N-linked Glycosylation
attachment of oligosaccharide to the protein through the
amine group of an asparagine

O-linked glycosylation
attachment of oligosaccharide to the protein through the
hydroxyl group of a serine or threonine
All IgG antibodies produced by mammalian cells are
glycoproteins, with an N-linked oligosaccharide attached to
each heavy chain in the hinge region at Asn-297
Importance of Glycan
Effect of Glycan
Facilitate protein folding in the ER
Increase solubility
Affect biological activities
Fucose for ADCC activity
Affect half-life in circulation and pharmacokinetics
Heterogeneity In Glycoforms
Macroheterogeneity: when multiple sites of
glycosylation are present in a protein, the occupancy on
different sites differs on different molecules
Microheterogeneity: the structure of the glycan
occupying the same site differs among different
molecules
A vast majority of recombinant therapeutic proteins

are glycoproteins. The extent of glycosylation and
the structure of the glycans on those glycoproteins
may have a profound effect on their activities
and circulatory half-life. Glycans are classified as
O-linked or N-linked glycans. O-glycans attach to
the polypeptide through the -OH group of serine
or threonine. N-glycans link to protein through the
amide group of asparagine. For N-linked glycan,
the asparagine is in an Asn-X-Thr/Ser recognition
sequence, where X indicates no specificity.
N- and O-glycans attached to proteins are

structurally heterogeneous. The glycans attaching
to the same attachment site of different glycoprotein
molecules are often structurally different. Such
heterogeneity is called microheterogeneity.
Multiple glycosylation sites are often present on

a protein molecule. Not all glycan attachment
sites on a protein molecule may be occupied.
Different protein molecules may have different
combinations of occupied and free sites;
such difference in the occupancy on different
attachment sites is called macroheterogeneity.
Glycosylation starts while the protein is still being

translated and undergoing protein folding in the
ER. The presence of glycans affects the solubility,
aggregation, and the stability of the folding protein.
The glycan structure affects the half-life of blood

circulation and immunogenicity of a therapeutic
protein. The presence of carbohydrates delays
clearance from blood, as demonstrated by a
comparison of glycosylated protein and its nonglycosylated counter-part. Higher sialic acid content
increases the circulation half-life of erythropoietin
(EPO). Under-sialylated glycoproteins are thought
to be cleared by a higher liver uptake via the
hepatic asialoglycoprotein binding protein. It was
also postulated that the glycosylated recombinant
proteins are better trapped by the extra-cellular
matrix, thus having a longer bioavailability
in vivo than their unglycosylated variant.
Protein Folding and Glycosylation
Fig. 3.21: Glycosylation of secretory proteins
Fig. 3.22: N-glycan processing of glycoproteins

in endoplasmic reticulum
Glycan on glycoprotein may also affect its biological

activities. Many therapeutic antibody IgG molecules
facilitate killing of target cells through antibody
dependent cellular cytotoxicity (ADCC). ADCC
activities of those antibodies have been reported
to be affected by its glycan structure. Molecules
without a fucose on its mannose core have a fifty
fold higher ACDD activity than those with a fucose.
O-glycosylation initiates in either the ER or the Golgi.
Overall, our understanding of O-glycosylation is far
less than N-glycosylation. N-glycosylation is initiated
by the transfer of a preassembled oligosaccharide
(Glc3Man9GlcNAc2, an oligosaccharide of three
glucose, nine mannose, and two N-acetylglucosamine)
to an asparagine in the recognition sequence of
a nascent protein in the ER lumen. The addition
of glycan occurs during the process of protein
synthesis, prior to the completion of protein folding.
The first part of N-glycan synthesis involves

the assembly of a high mannose backbone on
a membrane anchored dolicol molecule, on the
outside surface of the ER. The glycan is linked to the
dolicol carrier through a pyrophosphate group in
an activated form. After the seven-sugar backbone
is formed (with five mannose and two N-acetyl
glucosamine), it flips over to the interior of the ER.
No transporters are needed for the transport of
backbone glycan; rather, a flippase catalyzes their
translocation into the ER lumen. Inside the ER,
the backbone acquires an additional four mannose
and three glucose to become a mature core.
The mature core is then transferred to a binding
site (Asn-X-Thr/Ser) on a nascent protein
molecule. However, not all glycan binding sites
may be occupied. This could be due to competition
between protein folding and core glycan transfer.
Hence, different permutations of site occupancies
in different protein molecules exist, giving rise to
a macroheterogeneity of glycosylation patterns.
After being synthesized, chaperones surround
protein molecules to facilitate their folding process.
The three glucose molecules on the glycan core
serve as a quality control signal for the proper
folding of these glycoprotein molecules. Upon
completion of folding, the three glucose molecules

are removed from glycan, signaling the proteins
readiness to be transitioned to the Golgi apparatus.
Proteins that are not folded properly, with glucose
still in their glycan, may be recycled through the
calnexin pathway for refolding. Some unfolded
proteins are sent to the proteosome for degradation.
Glycan Extension in Golgi Apparatus
Fig. 3.23: N-glycan extension in Golgi apparatus
The well-folded glycoprotein molecules are

enclosed in membrane vesicles of ER and then
bud from the ER and translocate to the Golgi
apparatus. Once there, they fuse with the Golgi
body membrane and the glycoprotein cargos are
released into the lumen of the Golgi apparatus.
Inside the Golgi, mannose is trimmed from

the N-glycan core, effectively reducing the
number of mannoses from nine to three, before
extension takes place. Three main carbohydrate
molecules constitute most of the extended glycan:
N-acetyl glucosamine, galactose, and sialic acid.
Different glycosyltransferases add a different
monosaccharaides to the growing core glycan on the
protein. The incoming monosaccharide provides
the activated carbonyl group and the receiving
carbohydrate moiety on the growing glycan on the
protein may have more than one hydroxyl group
available for extension. The glycosyltransferases
recognize different incoming monosaccharaides
and catalyze different glycosidic bond formations.
However, a number of glycosyltransferases do allow
for some flexibility in glycosidic bond formation.
Intermediate glycans on the protein have more than
one sugar that can be extended and each sugar may
have more than one hydroxyl group that is available
for extension. Importantly, the extension reaction
does not take place on all of the available reaction sites.
Each growing glycan, thus, has multiple available
reaction paths for extension. The glycans can
grow into different numbers of branches, creating
structures such as biantennary and triantennary
glycans. In some cases, the reactions of adding
those sugars to different branches of the glycan
may occur in different sequential orders, but
lead to the same product. In other cases, the
addition of a particular glycosidic linkage may

hinder the reaction of the others; thus, the
reaction, itself, leads to different glycan structures.
Consequently, a very large number of glycan structures

can be formed in the N-glycosylation pathway;
however, the number of glycosyltransferases
for the extension reactions expressed in a cell
or tissue is relatively small (only a dozen or
so). However, diverse patterns of glycosylation
are often seen throughout many glycoproteins.
Glycan Types and Microheterogeneity
Fig. 3.24: Different types of N-glycans in glycoproteins
The web of glycan extension reactions forms

a complex network which, when drawn out
graphically, indeed resembles a network of
diverging and converging paths leading to a number
of different fully-extended N-glycan structures.
N-glycan structures are generally classified into

three principal categories: high mannose, complex,
and hybrid. All of them share a common trimannosyl (Man3GlcNAc2) core structure. The high
mannose glycans have five to nine mannose (Man5GlcNAc2) sugars. Those with two GlcNAcs attached
9
to the tri-mannosyl core are called complex. As its
name implies, the hybrid types are a combination
of high mannose and complex glycans, and have at
least three mannose sugars but only one GlcNAc
on one nonreducing mannose. This diversity of
glycan sugar composition on each glycosylation
site is referred to as microheterogeneity.
N-glycan microheterogeneity arises through
alternative reaction paths of extension in the Golgi
apparatus, as described above. The Golgi apparatus
consists of stacks of membranous compartments
commonly grouped into cis, medial, trans, and
trans-Golgi network (TGN) cisternae.
These
cisternae are not biochemically homogeneous. As
the secretory glycoproteins traverse through these
Golgi compartments, the glycan extension reactions
are catalyzed by varying the composition of
glycosylation enzymes in each compartment. Adding
to this diversity, not all protein molecules spend an
equal length of time in different Golgi compartments;
some exit early while others linger. Some glycans
Synthesis and Transport of Nucleotide

Sugar Precursor
do not acquire a terminal sialic acid, while others

have multiple sialic acid molecules on them.
Glycosidic bond formation is mediated by nucleotide

sugars. The nucleotide (NTP) reacts with an
activated sugar (glucose-1-phosphate, or glalatose1-phosphate,
N-acetyl-glucosamine-phosphate,
mannose-1-phosphate) to form an NDP-sugar. Uracil
is used for glucose- and galactose-based sugars
(e.g., UDP-glucose and UDP-galactose), guarnyl for
mannose and fucose, and cytidyl for sialic acid.
Mannose, galactose, and fructose are synthesized in
branches of the glycolysis pathway. All three sugars
are activated at their first carbon. Therefore, they
link to glycans through the formation of (1n)
glycosidic bonds. For example, UDP-NAcGlc is added
to a growing core by forming an N-acetylglucosamine
(1n) mannose bond; there can be a number of
possible positions on mannose (e.g., 2, 3, 4, or 6).
Fig. 3.25: Transport of nucleotide-sugar precursors

into organelles
For sialic acid, the second carbon is activated; thus

CMP-2-sialic acid will form a sialyl (2n) bond
with galactose. The synthesis of all the precursor
sugars occurs in the cytosol, including a ninecarbon sialic acid and N-acetyl neuraminic acid.
Similarly, all nucleotide sugars are formed in this
way, except for sialic acid. The activation of sialic
acid to CMP-sialic acid occurs in the nucleus.
The backbone of N-linked glycan is synthesized

on the cytosolic side of the ER membrane through
the membrane-anchored dolicol. The nucleotide
sugars used in the formation of the backbone, GDPmannose, UDP-N-acetyl glucosamine, and GDPglucose are synthesized in the cytosol and directly
react with dolicol or with the growing glycan
backbone. The assembled backbone is then flipped
into the ER and the subsequent reactions occur
inside the ER. Transporters for these nucleotide
sugars have been reported to be present in the ER.
The nucleotide sugars for the extension reactions
in the Golgi apparatus are also transported
through transporters. These include CMP-sialic
acid, GDP-fucose, UDP-N-acetyl glucosamine,
Glycan synthesis use nucleotide sugar as monomer unit

Nucleotide sugar highly charged, require specific
transporter into organelles
Sialic acid synthesis occurs in cytosol, activation to CMPSialic acid to places in the nucleus.
UDP-galactose, and the activated sulfate donor

(3-phosphoadenosine, 5-phosphosulfate). All of
these transporters are antiporters, meaning that
a stoichiometric exchange of nucleotide sugars
with NMPs is responsible for the charge balance.
Fig. 3.26: Biosynthesis of precursors of glycans
Glycan Diversity Among Species
Glycosylation
Enzymes are somewhat different among species
Possible immunogeneity (e.g. high mannose glycan from
most yeast
Unless
intended
for
vaccination,
the
immunogenicity
elicited
by
recombinant
proteins is a concern. An antibody elicited by
and against the protein therapeutic can result
in neutralization of the therapeutic protein and
may result in an unintended drop in efficacy,
thus causing serious adverse clinical effects.
The potential immunogenicity of recombinant

therapeutics may arise from the aglycosylated
protein core or from the glycan associated with
them. There are at least two mechanisms by which
glycans on a protein may affect the immunogenicity
of a human therapeutic: 1) by being a foreign
glycan structure, or 2) by shielding a segment of
the protein that is otherwise antibody inductive.
Different recombinant human therapeutic proteins

that are produced in different organisms are differently
glycosylated (such as those from CHO versus yeast)
or aglycosylated (such as from CHO versus E. coli).
Comparison of those proteins indicates that the
shielding effect of minimizing immunogenicity is
affected by the nature of the protein, as well as by
the source of the protein. The concern about the
immunogenicity of different glycoforms of the rDNA
proteins produced in insect cells and in transgenic
plants has hindered those technologies application
for rDNA therapeutic protein production.
Glycosylated proteins produced in CHO and
mouse myeloma cells are minimally immunogenic.
The glycosylation pathway is highly conserved
in mammals. Nevertheless, a divergence among
different species does occur. Host cells derived from
other species may possess a set of glycosylation
enzymes that are different from humans. Human
glycans
have
terminal
N-acetylneuraminic
acid (NANA), whereas other mammals have
N-acetylglycolylneuraminic acid (NGNA). NGNA
is the hydrolytic product of CMP-sialic acid
hydroxylase, which is present in nearly all mammals
but absent in humans. Thus, glycoproteins produced
in CHO have some NGNA present among all sialylated
glycans. Similarly, glycoproteins expressed in CHO
CHO Glycan
Has NGNA instead of human NANA
Has only (2,6) sialic acid, human has both (2,3) and,
(2,6).
Concluding Remarks
In this chapter, we presented a brief overview of
broad areas of cellular metabolism. We explored the
core of energy metabolism, the process of glucose
utilization through glycolysis, PPP and the TCA cycle,
and how all of these affect cell growth behavior
productivity. Through interconnected pathways,
the central corridor of energy metabolism also
influences the synthesis, and even the glycosylation,
of the product proteins. The excessive consumption
of glucose and glutamine and the corresponding
accumulation of lactate and ammonium in culture
all contribute to growth inhibition and low
productivity. Lactate consumption in the late stage
culture has been positively associated with a high
productivity. There are, therefore, ample incentives
cells have only terminal (2,3)-linked sialic acids,

in contrast to (2,6) and (2,3) seen in humans,
due to a varied composition of sialyltransferases.
Such differences in glycan composition have posed
a concern; however, the antigenicity of recombinant
proteins directly caused by glycans is still scant.
to better understand cell metabolism and to possibly
find different ways of manipulating cell metabolism
to better redirect the process. In recent years, we
have developed a better understanding of the link
between glycolytic regulation and growth control.
We have also established better tools to probe the
relationship between metabolic flux distribution
and other aspects of physiology that influence both
productivity and product quality. With the benefit
of global physiological perspectives, we continue to
gain a deeper understanding of metabolism. Global
views at a systemic level will significantly enhance
our capacity to manipulate cell metabolism, and
thus increase productivity and product quality.
Medium Design for Cell

Culture Processing
Optimization of Cell Growth Environment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
A Guide for Medium DesignBody Fluids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
Types of Media and Classes of Medium Components. . . . . . . . . . . . . . . . . . . . . . . . . . . 101
Types of Medium. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
Basic Components of Cell Culture Medium. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
Classes of Medium Components. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
Components of Basal Medium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
Water. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
Sugars and Energy Source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
Amino Acids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
Vitamins. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
Nucleosides. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
Optimal Concentration of Organic Nutrients. . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
Fatty Acids and Lipid Precursors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
Non-Nutritional Medium Components. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
High Molecular Weight and Complex Supplements . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
Medium for the Industrial Production Culture. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Optimization of Cell Growth Environment

Medium, like the food we eat, exerts a fundamental
influence on the well-being of cultured cells,
Industrial Cell Culture Process
profoundly affecting their growth, metabolic
1. Cell expansion
activities, and other biological capability. The
2. Production/differentiation
question of how best to devise culture medium
emerges whenever new in vitro cultivation Cell expansion stages last much longer than production
based science or technology is on the horizon,
Medium design for both stages
as occurred three decades ago driven by the
revolution in recombinant mammalian cell based
biotechnology, and as is occurring now concomitant
with the emergence of stem cell science.
Cells are the heart of cell technology; however,
without proper medium, cell cultivation cannot
accomplish process goals. Most cell types share
common basic nutritional requirements although
their needs for growth factors and cytokines may
MEDIUM DESIGN |97
Classical Medium Design - Optimize Cell Growth

Optimization for production - squeeze cells last
productivity out
Resurgence of media research in stem cell culture
opportunities in growth factors, antogonists, and
signaling pathway manipulation
differ. In the nearly five decades since scientists began

to isolate and cultivate cells, the focus of medium
design has been to optimize cell growth, maintain
growth potential and sustain the differentiated
properties in cultures of differentiated cells.
With the growing importance of biologics, the

focus of medium design has been extended to
enhancing production characteristics, such as
productivity or product quality. However, even
with the focus shifting to production, optimizing
medium for cell growth remains important. In
fact, the time that cells spend in the production
stage in a manufacturing reactor is a comparatively
small portion of their life span. A cell spends the
majority of its lifespan in growth, yielding progeny
to generate a sufficiently large number of cells for
producing product. Providing cells with an optimal
medium during the expansion stage is critical.
Recent advances in stem cell science have spurred

renewed interest in elucidating the nutritional needs
of cells in culture. Although the fundamental aspects
of nutritional requirements of a stem cell are not
different from other cell types, their requirements
for growth factors, surface matrices, and other
microenvironmental factors makes medium design
for stem cells far more complex than that for any
cell lines used in protein biologics production.
Furthermore, stem cell applications require that
the stem cell progeny be directed to differentiate
to specific lineages, for which the growth factor
requirements pose an even greater challenge.
Regardless of traditional biologics-based cell

technology or stem cell bioprocessing, the culture
process will involve both cell expansion and product
formation or differentiation. Medium optimization
strategy for cell expansion and for production may be
rather different. For expansion, the long term healthy
state of a cell while proliferating must be safeguarded.
In contrast, for production of biologics, the cells are
approaching their final stage of utility, and after all
products have been released into the medium, cells
and product molecules must be separated. Hence,
even conditions that might ordinarily hamper
MEDIUM DESIGN |98
growth or harm cells, such as reduced temperature

or increased osmolality, are sometimes used.
For stem cell processing, or for other cell therapy,

the distinction between cell expansion and production is not as significant. Even though cells are no
longer being expanded during the differentiation
stage or other final stage of preparation for clinical applications, cells must not to be subjected to
deleterious conditions. Since the final product is
cell mass itself, their survivability and functional
capability after the cell culture process is critical.
In the following sections we will focus on nutritional

needs for cell expansion first, as that is best known.
We will then discuss how optimal medium for production compares with that required for growth.
MEDIUM DESIGN |99
A Guide for Medium DesignBody

Fluids
Table 1. Cellular Chemical Environment in vivo
Approximate Concentrations in Cellular Environment
Interstitial Intracellular
(mM)
(mM)
Na+
140
14
4.0
140
1.2
0.01
0.7
20
108
28.3
10
Ca
2+
Mg
2+
CI-HCO3
HPO4 , H2PO4
11
0.5
Lactate
1.2
1.5
Glucose
56
Protein
02
Total Chemical Species

(mmole/L)
301.8
302.2
Corrected osmolar activity (mM)
281.3
281.3
3-
2-
SO43Amino Acids
To design an optimal medium for cell growth, it is

instructive to examine the chemical environment of
their natural niche. The ultimate objectives of medium
design are cell expansion, differentiation, and
production, not necessarily to reproduce their niche.
Understanding their native chemical environment
provides us with a starting point from which to
devise an environment suited to process goals.
The vast majority of cells in the body are not in direct

contact with blood, but are surrounded by interstitial
fluid. The chemical composition of interstitial
fluid, especially the protein and hormone content,
varies with tissues. However, the general chemical
composition of small molecular weight solutes in
interstitial fluid and in intracellular fluid is similar.
The low molecular weight solute composition
of interstitial fluid bears a few important
characteristics. The total osmolarity is around 280300 mM (or mOsm). A couple of percentage points of
error notwithstanding, the osmolarity can be taken
as the sum of molarity of all dissolved species in the
fluid. The largest contributor to the final osmolarity
is Na+, followed by Cl. In addition to Na+, other
inorganic species are present; notable are K+, Mg2+,
and Ca2+. However, those positively charged ions are
all present at low concentrations. Since the net charge
in a solution must be neutral, the total molarity of
positively and negatively charge ions must be equal.
In general Cl concentration is lower than Na+ because
bicarbonate (HCO3) is also present at ~30 mM
contributing to the negative charge in the solution.
For many ion species the interstitial concentration and

intracellular concentrations are strikingly different.
Both Na+ and Cl are present outside the cell at a tenfold higher concentration than inside the cell, as is
Ca2+. In contrast, K+, Mg2+ and PO43 concentrations
are much higher on the intracellular side.
Cells can tolerate deviations from optimal
conditions for some period of time. The estimated
range of non-lethal physiological concentrations of
key compounds vary considerably. Keep in mind
MEDIUM DESIGN |100
Table 2. Non-Lethal Range of Medium Constituents

Normal Range
Approximate Nonlethal Limits
Oxygen
35 45
10 1,000 mm Hg
Carbon dioxide
35 45
5 80 mm Hg
138 146
115 175 mmol / L
Potassium ion
3.8 5.0
1.5 9.0mmol / L
Calcium ion
1.0 1.4
0.5 2.0mmol / L
Chloride ion
103 112
70 130mmol / L
75 95
20 1500 mg / dL
Sodium ion
Glucose
Body
temperature
98 98.8 65 110 (18.3 43.3)

(37.0)
F (C)
pH
7.3 7.5
that the non-lethal range for the human body can

be rather different from that for cultured cells.
For example, cells in culture can tolerate low (80
mM) or high (140 mM) sodium concentrations, or
high osmolarity (400 mOsm), for a period of days,
whereas these extremes would not be tolerated by
most aquatic animals for more than a few hours.
6.9 8.0
Types of Media and Classes of Medium Components

Basal Medium
Sugar
Amino acids
Fatty acids, lipid precursors
Vitamins, nucleosides
Bulk salts, trace elements
pH buffer
Supplements
Serum, hydrolysates
Growth factors
Carrier proteins
A complete cell culture medium often has two

major categories of components: basal medium
and growth supplements. The basal medium is the
nutrient mixture consisting of the small molecular
weight components including sugar, amino acids,
vitamins, various salts, etc. The basal medium does
not merely provide a nutritional source for deriving
energy and making new cell mass and product,
it also provides balanced salt concentrations
and osmolarity to allow for cell growth.
However, most cells will not grow if provided

with basal medium alone, as basal medium
does not contain growth factors or other factors
necessary for optimal growth conditions.
Growth supplements that may be added to basal
medium include growth factors, phospholipids,
soy hydrolysate, serum, etc. These supplements
may promote cell growth by providing constituent
components for specific signaling pathways, or may
supply special nutritional needs (such as delivering
cholesterol), and may direct cellular differentiation
MEDIUM DESIGN |101
or maintain cells at a particular differentiation state.
There are numerous formulations of basal

media that are made into powder, packaged
and sold for commercial or routine use. Sterile
filtration is the standard means of sterilizing
basal medium, but heat sterilization (as is
used for microbial media) is also possible.
Types of Medium
Complex Medium vs. Chemically Defined

Medium
Chemically Defined Media

Progress in this area will likely be accelerated by:
an urgency to demonstrate control over all aspects
of production and downstream processing for
licensing by the FDA
the availability of recombinantly produced (in E. coli)
tissue culture supplements (i.e. insulin, EGF)
genetic engineering of cells to produce their own
growth factors
development of small, synthetic peptides that can
mimic the action of the larger, naturally occurring
protein (i.e., RDG sequence)
acceptance of continuous bioreactor systems and
the operation of these systems in a maintenance
mode
Serum-Free and Animal Component-Free

Medium
Traditional cell culture medium contains up to

15% animal serum in addition to basal medium.
Serum is a highly complex fluid in terms of its
chemical composition. Such a medium, containing
a largely undefined chemical composition, is
called a complex medium. Many supplements
commonly used in industrial processes, e.g.,
plant hydrolysates, soy phospholipids, also
fall into this category. Their use renders the
chemical composition of the medium undefined.
A chemically defined medium contains only

components whose chemical composition is
known and characterized, and has all of its
chemical species specified. It does not contain
any mixture of components with unknown or
undefined composition. For example, lipids or
phospholipids are not well defined compounds,
but are mixtures of a class of compounds and
are not chemically specified. A chemically
defined medium often contains growth factors,
cytokines, and carrier proteins. Thus, a chemically
defined medium is not necessarily protein-free.
A large number of industrial production processes

of rDNA proteins has eliminated serum from the
medium in the past decade. The use of serumfree and animal-component-free medium has
become the industrial manufacturing norm, with
the intent to minimize animal virus or prion
contamination. Recently, serum-free medium has
been increasingly used even during the cell line
development stage to eliminate exposure of cells
to serum and animal components throughout the
MEDIUM DESIGN |102
Serum-Free Media
Serum-free medium consists of nutritionally
complete basal medium supplemented with an
empirically determined mixture of hormones, growth
factors, attachment factors, attachment proteins and
binding proteins.
Many serum-free media contain a complex mixture
of undefined components, such as soy-meal
hydrolysate, peptone or beef hydrolysate, or other
plant hydrolysate.
entire cell banking and manufacturing process.
Nonetheless, the use of serum-free medium is not

yet universally practiced. For example, bovine
serum is still widely used in manufacturing viral
vaccines. As well, it is often not an easy task to
eliminate animal serum from processes involving
the cultivation of primary differentiated cells.
By definition, protein-free medium contains no

protein. Most protein-free media are, as well,
chemically defined. However, a protein-free medium
may contain undefined lipids or fatty acids, and thus,
may not be chemically defined. Cells that grow well
in chemically defined medium are likely to be highly
adapted or transformed, which eliminates any
dependence on mitogenic molecules or lipid sources.
Protein-Free Medium
Basic Components of Cell Culture Medium

Classes of Medium Components
Stoichiometric vs. Habitation-Conducive
Basal Medium Components
Stoichiometric: glucose, amino acids, vitamins,

nucleotides, lipids, fatty acids, some growth factor,
some salts
Habitation conducive: Bulk salts, such as sodium,
chloride, and some proteins, albumin, some
growth factors.
Medium serves two important roles: to provide a

chemical environment in which cells can grow, and
to supply the components cells need to generate
energy and convert to cell mass and products. Some
medium components are taken up by cells and
utilized to make more cells and products; other
components provide the chemical environment
but are not appreciably taken up by cells. The
majority of medium components, including glucose,
amino acids, lipids, and vitamins, are consumed
stoichiometrically. The more cells are made,
the more of those components are consumed.
In a most precise sense, nearly all medium
components are utilized to generate more biomass.
Even water is taken up by cells; as the cell volume
expands water must be taken up along with other
components that constitute cellular materials.
However, in practical terms, many medium
MEDIUM DESIGN |103
components are consumed in such small quantities

that their consumption may not be measurable.
Stoichiometric medium components must be

supplied in sufficient quantities to reach the
cell, and must meet the product concentration
target of the production. To reach a higher
goal, more stoichiometric components must be
supplemented. For the unconsumed components
of the mediumcomponents whose role is
only to provide a conducive environmentthe
key issue is to maintain their concentration.
Examples of conducive, unconsumed medium

components include chloride ion, sodium ion, and
even phosphate ion. Under ordinary conditions, only
a negligible amount of these components are taken
up by cells. However, these normally unconsumed
salts may be taken up in substantial quantities
in fed-batch cultures where fortified feeding is
used to add more glucose, amino acids, and other
nutrients to grow cells to a higher concentration.
In that case, these will also need to be replenished.
Thus,
unconsumed
medium
components
may
become
stoichiometric
components
under high density cultivation conditions.
Example
PO4-3 concentration in medium is typically 1 mM, while its cellular concentration is 11 mM. The
intracellular ATP/ADP concentration is about 2mM, which represents around 6mM of phosphate.
Total DNA and RNA contain another 35 mM equivalent of phosphate. A culture with cells of an
average volume of 2 x 10-12 L and at a cell concentration of 1010 cells / L take up about [(11+6+35)
mmole / L x 2 x 10-12 L / cell x 1010 cell / J= ~ 1 mmole / L of PO4-3.
Cell growth will certainly cause PO4-3 concentration in the medium to decrease drastically at such a
cell concentration.
For instance, phosphate and trace metals are consumed in very small amounts, although their
depletion may not lead to stoppage of growth immediately, they must still be supplied in
stoichiometric quantities.
MEDIUM DESIGN |104
Stoichiometric vs. Catalytic

Macromolecular Components
Some high M.W. medium components are
interalized and consumed
Others are not internalized when exerting their
functions and are degraded only slowly
Cell culture medium often contains macromolecules

such as insulin, fibroblast growth factor (FGF),
serum albumin, etc. These constituents serve a
variety of purposes; some are carrier proteins
carrying ligands into the cell, while others are growth
factors that bind to receptors on the cell surface.
Molecules that are taken up by cells may need to
be replenished to maintain their concentration,
whereas molecules that transmit signals by
binding to cell surface receptors may not need to
be replenished as frequently during cultivation.
Many macromolecules are internalized, and

some are degraded (consumed), while others
are recycled. For example, the ferric ion carrier,
transferrin, binds to transferrin receptor and is
internalized. After being internalized, transferrin is
translocated to lysosome, and after releasing ferric
ion in the low pH environment of the lysosome,
transferrin is recycled to the extracellular medium.
As long as ferric ions are available (i.e., replenished),
transferrin can continue its role as ferric ion carrier.
Insulin, another commonly used growth factor,
binds to the insulin receptor and triggers a
signaling event that does not involve its own
internalization. However, when present at a high
concentration (a few microgram per ml), insulin is
rapidly internalized by hepatocytes and degraded.
Therefore, even though as a growth factor it is
not consumed, its concentration does decrease.
An example of a consumable macromolecule is lowdensity lipoprotein (LDL). After an LDL particle

binds to an LDL receptor on the plasma membrane,
the receptor-ligand complex is internalized in a
clathrin-coated pit that pinches off intracellularly
to become a coated vesicle. The clathrin coat then
depolymerizes, resulting in an uncoated (smooth
surfaced) vesicle, often called an endosome. The
endosome then fuses with an uncoupling vesicle
that has an internal pH of about 5.0, which causes the
LDL particles to dissociate from the LDL receptors.
MEDIUM DESIGN |105
The LDL receptors are then recycled back to the

plasma membrane. The vesicles containing the LDL
particles fuse with lysosomes, in which the cholesterol
esters are hydrolyzed to fatty acids and cholesterol.
Cholesterol is incorporated into cell membranes.
Components of Basal Medium

Water
Types of Contaminants in City Water:
Inorganicsheavy metals, iron, calcium, chlorine
Organicsby-products of plant decay, detergents
Bacteriaendotoxin or pyrogen
Particlescolloids or particles
A typical water preparation process involves filtration,
Reverse Osmosis
WFI used to be employed, now mostly pure water.
Mammalian cells are exceedingly sensitive to the

quality of water used for media preparation. City water,
the usual source of water for medium preparation
contains particulates, bacteria which are the source
of endotoxin, trace organics, and various inorganic
ions including harmful heavy metals. Typical water
preparation processes include deionization through
ion exchange, microfiltration to remove particulates
and bacteria, and finally reverse osmosis to reduce
conductivity (or increase resistance) to > 20 Mcm.
In some cell therapy applications, because the
product (i.e., cells) is subjected to little purification
process before administration to the patient, cell
culture medium may be prepared using water
for injection (WFI) to avoid the entry of any
pyrogenic contaminants. WFI is prepared by low
evaporation rate distillation, minimizing the chance
of any water droplet in the evaporating stem from
carrying over any solute or particle from the water.
Sugars and Energy Source
Glucose and glutamine are the primary nutrients

that supply a cells energy needs in culture. The
physiological concentration of glucose in blood
is 0.8 g / L. In culture, glucose is typically present
from 1 g / L (5.5 mM) to 5 g / L (27.5 mM). In the
production reactor, sometimes a high level of
glucose, as much as 15 g / L (82.5 mM), is used.
In this case, glucose is a large contributor to the
osmolarity of the medium, and adjustment of the
composition of the medium must be made (by
reducing sodium and chloride concentration) to
MEDIUM DESIGN |106
maintain osmolarity in a growth permissible range.
Six-Carbon Sugars
Glucose:1-5 g / L,
Fructose, galactose, may also be used.
Galactose and glucose can both be transported by the
GLUT1 transporter, which is present in most cells.
Fructose is transported by a different transporter
(GLUT5); unless the transporter is expressed, the cell
may not be able to use fructose efficiently.
The alternative sugar is often taken up by cells at a
slower rate, which may reduce lactate production and
be good for cell maintenance.
Pyruvate and ribose are sometimes supplied in small
quantities, insufficient to supply cell's energy needs.
Glutamine Serves Two Roles:

source of amino acid for protein
nucleoside synthesis, energy source.
Glutamine is consumed in large quantities,

approximately 1/5 to 1/10 of glucose in molar amount.
Glutamine is spontaneously degraded in aqueous
solution.
All cultured cells express the GLUT1 transporter

at a significant level, and take up glucose readily.
GLUT1 also transports galactose. Thus, galactose
can be used as an alternative sugar to glucose. The
galactose km for uptake is higher than for glucose. In
the concentration range used for glucose, galactose
is taken up by cells at a lower rate, resulting
in lower lactic acid production in the culture.
Fructose is also transported by the GLUT5
transporter. The Km for fructose transport by
GLUT5 is also high. Thus, similar to galactose, the
uptake rate for fructose is lower than for glucose
unless a high concentration of fructose is used. At
a low uptake rate, the use of fructose also results in
lower lactic acid production compared to glucose.
Glutamine is an essential amino acid for cells in

culture. Most cells in tissues express glutamine
synthase, and make glutamine from glutamic
acid. In culture, they consume glutamine at
roughly 1/5 to 1/10 of the consumption rate of
glucose. Glutamine supplies the amino group
for nucleotide biosynthesis. It also supplies the
carbon backbone of TCA cycle intermediates by
converting to -ketoglutarate. Through carbon
13 tracer experiments, it has been shown that
glutamine contributes to lactate production
during cellular energy metabolism in culture.
Glutamine spontaneously degrades in aqueous
solution, and releases ammonium. Consequently,
to avoid degradation, glutamine is typically
added to culture medium immediately prior to use.
MEDIUM DESIGN |107
Amino Acids
Table 3. Essential and Non-Essential Amino
Acids
Essential amino acids
Non-essential amino
acids
L-arginine*
L-alanine
L-cysteine*
L-asparagine
L-histidine
L-asperatic acid
L-isoleucine
L-glutamic acid
L-leucine
L-glycine
L-lysine
L-proline
L-methionine
L-serine
Amino acids are classified as essential or nonessential

based on nutritional studies using animals or tissue
culture cells. Cells lack the biosynthetic pathways for
making essential amino acids, and rely on exogenous
supply to meet growth needs. Some amino acids are
essential only for cells in culture, but not for animals.
In animals, different tissues may cross feed each

other; amino acids synthesized in one tissue (e.g.,
liver) may be transported to cells in other tissues.
Some enzymes involved in amino acid biosynthesis
are expressed at lower levels in cultured cells in vitro
than in cells from their tissue of origin. Glutamine
is non-essential for animals; it is synthesized from
glutamic acid through glutamine synthetase. The
expression level of glutamine synthetase decreases
when cells are cultured in vitro. There is also a CpG
island in the promoter region of glutamine synthetase,
which may be methylated to cause glutamine
synthetase to be silenced in some cultured cells.
L-phenylalanine
L-threonine
L-tryptophan
L-tyrosine*
L-valine
L-glutamine*
*Essential for cells in culture, not for animals
Most culture media contain all twenty amino acids.

proline is required by mutant CHO cells;
serine is frequently required at clonal densities;
asparagine is required by certain malignant cells;
glycine sometimes needed in case of borderline folic
acid deficiency or in the presence of folate analogues
(methotrexate and aminopterin)
Small peptides can serve the same function as amino
acids--some of these are more stable (e.g., glycineglutamine) or are transported more readily than their
free amino acids counterparts.
Some non-essential amino acids are excreted into culture medium; alanine is most commonly seen. Asparagine and proline may also accumulate in medium.
Upon cell isolation for in vitro culture, expression

of some amino acid synthesis enzymes is
suppressed, but expression of some amino
acid transporters is elevated, which allows for
a faster transfer rate to meet growth needs.
Cell culture media developed in the 1960s

and 1970s contained at least the 14 essential
amino acids. Those media were designed to be
used with serum supplementation, which also
supplies some amino acids. Media designed
for serum-free culture include all amino acids.
In medium preparation, amino acid mixtures are often
prepared as concentrated stock solutions organized
into groups: neutral, acidic, basic, etc. Although the
solubility characteristics of amino acids indicates
that solubilizing them is feasible, the kinetics of their
dissolution is slow. Some stock solution preparation
methods rely on pH adjustment using acid or base to
increase the rate of dissolution. Those stock solutions
also introduce additional salts into the culture
medium. Thus, when using such stock solutions,
it is important to assess the osmolarity of culture.
MEDIUM DESIGN |108
Vitamins
Ascorbic acid may be beneficial for cells that synthesize
collagen.
Vitamin A can have a pronounced effect on growth and
differentiation of some cell types.
Vitamin K is required for gamma-carboxylation and
correct processing of vitamin K dependent proteins.
Vitamin E functions as an antioxidant.
Vitamin D regulates Ca+2 and is regarded by many as a
hormone rather than a vitamin. Most toxic of all vitamins when present in excess.
Thiamine pyrophosphate catalyses the transfer of carboxyl group, transketolase, transaldolase.
Pyridoxal phosphate (pyridoxine vitamin B6) catalyses
transamination.
The biological activities of vitamins vary. Even

though they are classified as a common class of
nutrient, their biological roles are diverse. They all
share the common feature of being essential for the
vitality of humans, and are needed only in minute
quantities compared to glucose and amino acids.
Some vitamins are cofactors involved in biochemical
reactions, and are required by all cells. These include
biotin, thiamine pyrophosphate (or its precursor),
riboflavin and cobalamin. Some vitamins such as
vitamin D, vitamin K, vitamin A, and vitamin C are
required by only certain differentiated cell types.
Biotin is a carrier of activated CO2, and is involved in

pyruvate dehydrogenase, pyruvate carboxylase, and
fatty acid synthesis.
Cobalmin (B12) is involved in free radical reactions of
intramolecular C-C bond rearrangement, methylation,
and conversion of ribonucleotides to deoxyribosenucleotides.
Nucleosides
Nucleosides are not included as essential components

of basal media when serum is supplemented.
Inclusion of small quantities of nucleosides is
common in serum-free medium. The small quantities
included reflect the low nucleoside content in
cell mass: nucleic acids (RNA and DNA together)
constitute only about 5% of dry cell mass. A purine
source (adenosine or hypoxanthine) together with
thymidine is beneficial when folic acid is in limited
supply (e.g., in the case of methotrexate selection).
Table 4. Nucleosides in Basal Medium

RNA
DNA
Adenosine
Thymidine
Cytidine
2deoxyadenosine
Guanosine
2deoxycytidine
Uridine
2deoxyguanosine
Fatty Acids and Lipid Precursors
Lipids constitute a significant portion of biomass of

mammalian cells. Lipid bilayer membrane forms
vesicles which play key roles in protein secretion
and virus replication. The composition of lipids in
membrane affects its fluidity. However, the effect
of lipid supply in medium is usually rather subtle.
MEDIUM DESIGN |109
Unlike the supply of essential amino acids or sugar,

whose deficiency cause observable effects on cell
growth almost immediately (with only a delay of
one doubling time), the effect of supplementing
or removing lipid is not as easy to assess.
Fatty Acids
Cells can synthesize fatty acid, but cant introduce
double bond beyond C9.
Some cell lines benefit from cis-unsaturated
fatty acids, such as oleic acid, linoleic acid and
arachidonic acid (a precursor for prostaglandin
formation)
Phospholipid Precursors
Choline
Ethanolamine
Inositolprecursor for phosphatidyl-inositol

biosynthesis
Cholesterolrequired by some cell lines (e.g.
NS-0 myeloma)
Mammalian cells can synthesize almost all

lipids required for their growth, including fatty
acids, phospholipids and cholesterol except for
auxotrophic mutants. Cells can make fatty acids of
all different carbon length in membrane, including
most unsaturated fatty acids, which constitutes
almost half of the fatty acids in phospholipid.
However, mammals do not introduce double bonds
beyond C9 into fatty acids. Linoleate (18:2 cis-9,
12) and linolenate (18:3 cis-9, 12 15) are thus
essential fatty acids.
Hence complete medium
for serum-free culture usually contains oleic acid,
linoleic acid, sometimes also arachedonic acid.
However, the removal of essential fatty acids from
culture medium, does not cause immediate cessation
of growth, subtle changes in membrane properties
is often masked by its residual amount in the cell
and visible effect may take a long time to emerge.
Ideally supplementing cells with phospholipids is
a good practice. However, non-animal source of
phospholipids with reproducible quality is not easily
available. Often precursors of phospholipids (choline,
ethanolamine, inositol) are supplied. Cholesterol is
supplied to its auxotrophic mutants, such as NS0 cells.
Cholesterol and fatty acids have very low solubility.
Fatty acids are conjugated in serum albumin when
serum supplement is used. In albumin-free medium
fatty acids may mixed in with other hydrophobic
supplements. Cholesterol can be supplied as
derivatized ester with a organic acid attached to
its hydroxyl group to increase its solibility, or can
be supplied as cyclodextrin conjugated complex.
MEDIUM DESIGN |110
Optimal Concentration of Organic

Nutrients
Optimal
Growth Rate
Suboptimal
Inhibitory
Optimal range
typically spans
over 10 fold or
more
Starvation
Nutrient Concentration
Fig. 4.1: Optimal range of nutrients for cell growth
Bulk Salts and Trace Elements
Upon determining the essential and beneficial

components of medium, it is necessary to decide on
the optimal concentration of those nutrients. The
experimental determination of concentration range
is complicated by cell growth and consumption
that causes medium composition to change.
Furthermore, metabolites accumulate as cell grow,
making culture conditions change as time goes by.
This problem was resolved by using clonal growth

as originally demonstrated with ells of human,
chicken and other species.. Cells were seeded in
small petri dishes at very low density of about 1050 cells/cm2 as opposed to 104 cell/ cm2 At such
a low cell concentration, the amount of nutrients
consumed is negligible compared to the amount
present in the medium. It was further assumed
that nutrient utilization is completely independent;
thereby allowing the effect of one component to be

tested when all the other components are present
in their optimal concentrations determined by
the point of experimentation. After plating cells
are dispersed as single cells on the surface. A
couple weeks later, the extent of growth from all
the cells initially plated is estimated by the total
surface area covered by cells after cells on the dish
is stained with a dye. The optimal concentration
is one which gives largest cell growth area.
The optimal concentration for almost all
nutrients (amino acids and other small molecular
weight molecules) span over a wide range of
at least ten fold. One can thus conclude that
for the purpose of cell cultivation, most cells
have a wide range of optimal concentration
for most organic basal medium components.
Mineral elements are also essential components of

cell mass. Phosphate constitutes part of nucleotides
and nucleic acids; magnesium is present in high
concentrations in the cell as it is conjugated to ATP,
which is present in the mM range; calcium, which is
MEDIUM DESIGN |111
Table 5. Concentrations of bulk ions in basal medium

(mM)
DMEM/
F12 (1:1)
Na+
Williams
E
DMEM
RPMI
F12
150.31
143.71
155.12
137.74
144.03
4.18
5.37
5.37
5.37
3.00
Mg2+
0.71
0.81
0.81
0.41
0.60
Ca
1.05
1.80
1.80
0.42
0.30
126.66
125.33
118.48
108.03
134.83
1.02
1.17
0.78
5.63
1.17
HCO3
29.02
26.19
44.04
23.81
14.00
SO42-
0.41
0.81
0.81
0.41
0.00
0.85
282
288
2+
ClPO43-
NO3
Total
313
302
327
Role of Bulk Ion

Maintenance of membrane potential (Na+, K+)
Osmotic balance (Na+, Cl-, HCO3-etc.) contribute most
of the osmolality of fresh medium. Optimal range is 280 310 mOsm / kg.
Biological roles:
Mg+ conjugate with ATP
Ca2+, Mg2+ for cell adhesion
PO43- for nucleotides
Signaling (Ca2+ )
Buffering (HCO-3, HPO3-4)
Table 6. Trace Elements in MCDB

104 (serum-free medium for human diploid cells) (mM)
CuSO45H2O
1.0 x 10-6
FeSO47H2O
5.0 x 10-3
MnSO45H2O
1.0 x 10-6
(NH4)6Mo7O244H2O
1.0 x 10-6
NiCl26H2O
5.0 x 10-7
SeO2
3.0 x 10-5
Na2SiO39H2O
5.0 x 10-4
SnCl22H2O
5.0 x 10-7
NH4VO3
5.0 x 10-6
ZnSO47H2O
5.0 x 10-4
essential for signaling in some differentiated cells,

is present in high concentrations in endoplasmic
reticulum (ER) and in some organelles. Calcium
and magnesium are also involved in cell-cell and
cell-substrate adhesion. It must be noted that
sodium and potassium, through their reverse
abundance in cytosol and medium, balance
membrane potential, which is fundamental to life.
Many trace metals play key biological roles. Ferrous
ion plays a key role in electron transfer complexes, as
does copper ion. Zinc is present at high concentration
in pancreas and is conjugated with insulin. Selenate
serves as an antioxidant. These elements are
required by cells in minute quantities; however,
their long term deprivation is detrimental to cells.
There is a wide range of concentrations of bulk ions
that is conducive to cell growth. What is apparently
most important for growth is the balance of
osmolarity. The most important contributors
to osmolarity are sodium ions, chloride ions,
and bicarbonate; while the concentrations of
these can be varied, the total osmolarity must
be maintained in the range of 270-330 mOsm.
When bicarbonate is not used in culture medium its
contribution to osmolarity must be replaced by other
salts. It is common practice to add a mixture of NaCl
and KCl to maintain their molar ratio at about 30.
There is a wide range of conducible concentration for

bulk ions. What is apparently most important is the
balance of osmolarity during growth stage. The most
important contributors to osmolarity is sodium and
chloride ions, the concentration of both can be varied,
so is bicarbonate. However, the total osmolarity
must be maintain in the range of 270-330 mOsm.
MEDIUM DESIGN |112
Trace Elements
Those clearly required by cultured cells are: iron, manganese, zinc, molybdenum, selenium, vanadium, copper.
Trace elements are ubiquitous contaminants of chemicals and supplements used in preparation of medium.
Some media contain rare trace elements such as rubidium, cobalt, zirconium, germanium, molybdenum, nickel,
tin and chromium; may be needed for long-term growth in
protein-free medium.
When bicarbonate is not used in culture medium

its contribution to osmolarity must be replaced by
other salts. It is common practice to add a mixture
of NaCl and KCl to keep their molar ratio at about 30.
What are the roles of heavy metal ions? The tables

provide a reference of optimal starting
Non-Nutritional Medium Components

Some components of medium are additives that
make it operationally easier to grow cells. They can be
removed from medium without harmful effect, and
do not appear to be taken up by cells. However, their
presence in the medium, especially under process
conditions, can minimize operational deviations.
Sodium Bicarbonate Buffer
Sodium bicarbonate provides a pH buffer in our body

fluid, and was used to buffer cell culture medium
in the early days of in vitro culture development.
However, the pKa of bicarbonate is 6.1 making it less
than ideal as a buffer for neutral pH. The buffering
capacity of bicarbonate derives from the equilibrium
with soluble carbon dioxide as shown in the inset. The
buffering action of bicarbonate requires the presence
of CO2. With a given bicarbonate concentration,
the pH is inversely proportional to the CO2 level;
thus, as gas phase CO2 goes up, pH goes down.
Typical cell culture medium contains 14-44 mM

NaHCO3, which at equilibrium requires 10% CO2 to
maintain pH at 7.4. As pH decreases due to lactate
production, the CO2 level in the gas phase can be
reduced, which takes the proton to the left hand
side of the equilibrium equation, to maintain pH.
Conversely, as cells begin to consume lactate in the
late stage of a fed-batch culture, the CO2 concentration
in the gas phase is increased to maintain pH.
MEDIUM DESIGN |113
Bicarbonate Concentration in medium

44mM in DMEM, 14 mM in F12, 26 mM in circulating blood.
It is necessary to use 5 10% CO2 in the incubation chambers; media that contain bicarbonate become
alkaline very rapidly due to loss of CO2 when removed from the incubator.
The low pKa of bicarbonate (6.1) results in suboptimal buffering throughout the physiological pH range.
NaHCO3 buffer requires appropriate CO2 concentrations in the gas phase. The reactions are:
CO2 dissolves in aqueous solutions.

The CO2 concentration in liquid is described by Henrys Law.

H: Henrys law constant

CO2 in an aqueous solution forms a bicarbonate ion.

The equilibrium is described as:

From the definition of pH and pKa

The pH of the solution is affected by PCO2 and HCO3-.
From the equation, one can plot the relationship among HCO3-, PCO2 and pH.
MEDIUM DESIGN |114
How do bicarbonate and CO2 work together as a pH

buffer?
The equilibrium equation, with pKa = 6.36 can be used
to plot the relationship between pH and bicarbonate
concentration. The relationship is also dependent on
CO2(g). Two lines are shown for two different CO2(g)
concentrations. Each point on the line represents the
corresponding pH and carbonate concentration. To
buffer the medium at a given pH one has to select a
combination of bicarbonate and CO2(g) concentrations.
Figure 4.2: Relationship between sodium bicarbonate

concentration, atmospheric carbon dioxide and pH
Alternative Buffer
Sodium beta-glycero-phosphate (20 mM) also
functions as a detoxifier of ferric chloride hydroxo
compounds (i.e., Fe3+ chelator)
Zwitterionic buffers: HEPES (N-2hydroxyethylpiperazine-N-2-ethane) is used
between 10 50mM.
Alternative buffers can be growth inhibitory at high
concentrations (>25 mM)
Requires balance of osmolality (by adjusting bulk
salt levels, e.g. if 2 mM is used and 40 mM of
NaHCO3 is eliminated, the osmolality must be
balanced with NaCl / KCl. Need to maintain Na+ / K+
ratio (~30:1).
Table 7. Cell Culture Tested Biological Buffers

Description
pKa value
at 37 C
Anhydrous
Mol. Wt.
Working
Concentration
(mM)
Glycine 1.0M
9.53
75.0
50 200
Glycylglycine
7.95
132.1
10 20
HEPES
7.31
238.3
10 28
MOPS
7.01
209.3
10 20
Sodium
Bicarbonate
6.28
84.0
2 26
TRICINE
7.80
179.2
<50
Under some conditions, bicarbonate buffer is

insufficient for maintaining pHfor instance,
when a CO2 incubator is not available. There are a
number of buffers known to sustain cell growth.
Some of these buffers have a pKa in the vicinity of
7.0, and are more suitable for sustaining neutral
pH than bicarbonate. Unfortunately, most of them
have a lower maximum concentration that can
be used without a negative effect on cell growth,
which limits the buffering capacity they can provide.
It should be remembered that CO2 is a metabolite, a

medium component for buffer, and is also a nutrient
required for some biochemical reactions, such as
carboxylation reactions in fatty acid synthesis.
Normally, the role of CO2 as a nutrient is not
apparent because, as cells respire, the CO2 produced
is sufficient to supply CO2 needed for carrying out
carboxylation reactions in the cell. However, when
cultivating cells in a bicarbonate-free medium at a
low cell concentration, CO2 may become limiting for
cell growth. If continuous air is continuous flown
over the culture and thus stripping of CO2 occurs,
CO2 produced by cells may be stripped away and
may not be sufficient for supporting growth. Under
conditions such as these, using an air supply with
a small supplement of CO2 (0.2-0.5%) will suffice.
MEDIUM DESIGN |115
Antioxidants
Physiologically Relevant Antioxidants
Vitamin E
Taurine
Beta-carotene
Transferrin
Ceruloplasmin
Selenium - Selenium-deficient cells are more
sensitive to oxygen toxicity (selenium is a cofactor
for glutathione peroxidase, an enzyme which
helps remove peroxidases from cells).
Catalase
Superoxide Dismutase
Reduced glutathione
-mercaptoethanol
Mechanical Damage Protective Agents
Table 8. Synthetic Protective Agents Used in Cell

Culture
Common name
Chemical identity
Pluronic F68 or F88
Block copolymer glycols

of poly(oxyethylene) and
poly(oxypropylene) (M.W. ~8350)
PEG
Poly(oxyethylene) glycol (or polyethylene glycol) (M.W. ~20,000)
PVA
Polyvinyl alcohol (M.W. ~20,000)
MC
Methylcelluloses (15 cps methocel)

0.1- 0.2%
CMC, Edifas B50
Sodium carboxymethylcellulose
HES
Hydroxyethyl starch
PVP
Polyvinylpyrrolidone
Dextran
Dextran (M.W. ~78,5000, 20-60 g / L)
Reactive oxygen species (ROS) arising from

biochemical reactions cause transient presence of
superoxide radical (O2-) and hydrogen peroxide
(H2O2). Their reactions with lipids, proteins, and DNA
cause damage to cells and to media components.
Physiologically, cells deal with ROS through
reducing reactions that use cellular reducing
agents such as glutathione. Conventionally, the
issue is dealt with through medium design either
by providing antioxidant compounds to cells to
facilitate their capacity to cope with ROS, or by
supplying components to minimize oxidation
of labile medium components in culture fluid.
Except for mercaptoethanol, few other antioxidants

are routinely used in cell culture. However, for
the cultivation of differentiated cells or for the
cultivation of cells under stress conditions, some
additional antioxidants are included to prolong the
period before irreparable damage to cells occurs.
Some medium components are added with the

intention of modulating the physiochemical
environment of the cell. In addition to osmolarity,
viscosity was speculated to be important for
sustaining cell growth in culture in the early
days of serum free culture medium development.
Most process cell culture media contain a

protective agent, Pluronic F68, which was first
employed in the 1960s in growing BHK cells
in a stirred tank bioreactor. Many Pluronic
surfactants are available; all are block copolymers
of polyoxyethylene and polyoxypropylene. The
larger the POE (polyoxyethylene) group, the more
hydrophilic the molecule is and the greater its
detergent-like activity and cell cushioning effects.
The larger the POP (polyoxypropylene) group, the
greater the toxicity and the anti-foaming ability.
Presently, Pluronic F68, within a concentration
range of 0.010.1%, provides adequate cell
cushioning, but the degree of foaming is high.
Therefore, it is desirable to determine if a suitable
MEDIUM DESIGN |116
replacement is available. The only Pluronics other

than F68 known to provide suitable growth/
productivity are F88 and F77. Both have somewhat
different micelle formation characteristics than F68.
Polyethylene
glycol
or
cellulose have been used
agents,
but
are
seldom
Antibiotics
Table 9. Antibiotics for Cell Culture

Antibiotic
Chlortetracycline
Gentamicin sulfate
Nystatin
Penicillin G
Spectinomycin
Concentration
Antibiotic
Spectrum
Grampositive and
5 mg / L Gramnegative
bacteria
Grampositive,
Gramnegative
50 mg / L
bacteria and
mycoplasma
50 mg / L (or 100
Fungi and yeasts
U / mL)
100 U / mL
Grampositive
bacteria
carboxymethyl
as protective
used
now.
The use of antibiotics in cell culture is unevenly

practiced. For manufacturing processes, antibiotics
are rarely used, although in vaccine production,
the use of neomycin, polymyxin B, streptomycin,
and gentamicin are sometimes seen. For the
cultivation of primary cells, which are not subjected
to a long duration of quality control prior to
processing, the use of gentamycin reduces the rate
of contamination significantly. Some antibiotics
are only moderately more inhibitory to bacteria
than to cultured cells. Toxicity testing is necessary.
Grampositive and
20 mg / L Gramnegative
bacteria
High Molecular Weight and Complex Supplements

Some calls such as NS0, some hybridoma/
myeloma cells, and hepG2 cells are capable of
rapid growth using only basal medium without any
further supplement. However, such extraordinary
capability is the exception rather than the norm
of cells in culture. Most cells in culture require
supplementation of at least a number of growth
factors and carrier proteins. Many stem cells
and other normal diploid cell lines need high
concentrations of serum to grow. In fact, for isolation
of cells from tissue, serum is still commonly used, at
least in the early stage of cell cultivation. In discussing
the role of various supplements it is instructive
to start by examining the role of animal serum.
MEDIUM DESIGN |117
Serum or Biological Fluids
Functions of serum in cell culture medium:

Protease inhibitors (alpha 2 macroglobin) neutralize
proteases used in trypsinization or produced by cells.
Provides hormones and growth factors.
Provides carrier proteins
for low molecular weight substances
(e.g.,transferrin)
for nutrients which dissolve poorly (e.g., fatty
acids, cholesterol, apolipoprotein)
Binds compounds which are toxic when present in
excessive amounts, and releases slowly
Binds and/or neutralizes toxic substances (e.g.
detergents).
Serum is the blood fluid left behind after coagulation;

it is free of blood cells and most coagulation
proteins. Serum is an extremely complex mixture
that contains nutrient substances, metabolites,
hormones, plasma proteins, substances released
from damaged cells (e.g., hemoglobin and growth
factors from platelets), antibody molecules against
various antigens to which the animals have
been exposed, and may even harbor infectious
agents such as viruses carried by the animal.
Fetal bovine serum (FBS) is the most widely used

serum in animal cell culture because it contains higher
concentrations of growth stimulatory factors and
lower concentrations of growth inhibitory factors.
Other commonly used sera are human, bovine calf,
newborn bovine, donor bovine, and donor horse.
Serum serves many different and important roles

in cell culture. In addition to providing nutrients
not sufficiently present in basal medium (e.g.,
cholesterol), serum provides factors for cellsubstrate attachment (e.g., vitronectin, fibronectin)
for adherent cells, modulates colloid osmolarity,
a physiological property of medium with respect
to viscosity. Serum contains protease inhibitors,
and neutralizes trypsin used in cell detachment
as well as other enzymes released by dead cells.
Serum contains the carrier proteins transferrin
and serum albumin. Carrier proteins function
to chaperone components that are in very low
concentration or are otherwise poorly soluble,
such as fatty acids (carried by albumin), or are
unstable such as ferric ion (carried by transferrin).
Serum is rich in bulk proteins (e.g., serum
albumin) that can prevent non-specific adsorption
of critical factors to culture vessels. Serum also
plays an important role as a scavenger. In the
course of cell cultivation, various contaminants
may arise from numerous sources. For example,
this could result from leaching of minute chemical
components from reactor parts, from the filter
used in medium preparation, or even from medium
MEDIUM DESIGN |118
Disadvantages of serum in cell culture medium

potentially introduce animal viruses into cell culture
and other undesirable contaminants (e.g. adventitious
agents, antibiotics, proteases).
Availability of high quality serum
high running costs and unnecessary capital outlay.
Normally purchased in large lot sizes and costly
storage.
Serum lot testing tedious and costly.
Increase complexity of downstream processing and
final product characterization.
supplements. Some of these chemicals may be

detrimental to cell growth or may have other
negative effects on cells. In the presence of serum in
the medium, those compounds may be sequestered
by adsorption to serum proteins before they can
act on cells, thus minimizing potential damage.
Animal serum, however, has numerous disadvantages

in addition to cost and the difficulty of controlling
consistent quality. The most serious concern is the
possibility of contamination with animal viruses or
prions. The presence of serum in culture medium also
makes downstream processing more complicated,
and makes the task of final product characterization
more complex. Serum carries antibodies, some of
which may be against viruses that the animal donors
had been exposed to. For virus production processes,
if serum antibodies cross-react with the product
virus, the production will be drastically affected.
MEDIUM DESIGN |119
Insulin and Insulin-like growth factor (IGF1)

Insulin stimulates glucose uptake by adipocyte
and other cells, also has a mitogenic effect at high
concentrations
Insulin is used in culture at 1 - 10 g /ml range.
Blood insulin level is 4U/ml or 1.3 g / ml (1
U=0.33 g) ; IGF1 level is is 100 to 200 ng / mL.
Insulin and IGF have overlapping signaling
pathway through IR and 1GF1R
IGF1 regulates cell growth, IGF2 invovles in
development. IGF1 can replace insulin in cell
culture at a much lower concentration
Insulin plays a key role in regulating glucose uptake

by many cell types. In addition to modulating
glucose metabolism, insulin also exhibits mitogenic
effects, and stimulates cell growth through an
overlapping pathway with IGF-1. IGF-1 has an
acute effect on protein and carbohydrate anabolism
by increasing cellular uptake of amino acids and
glucose, and by stimulating glycogen and protein
synthesis. IGF-1 also affects cell proliferation,
differentiation, and apoptosis. It is a potent mitogen
acting to increase DNA synthesis and to stimulate
the expression of cyclin D in a wide variety of cells.
Both insulin and IGF-1 bind to the insulin

receptor (IR) and IGF receptor (IGF1R) but with
different affinities. After insulin binding, IR or
IGF1R is phosphorylated, leading to activation
of an insulin receptor substrate (IRS). There are
multiple isoforms of IRS that are distributed
differently in cells of different tissues. The signal
is then relayed to downstream signaling pathways.
The response of the cell to insulin and IGF is
dependent on the abundance level of the different
IRS isoforms. Most cells, including CHO cells, express
both IR and IGFR. However, NS0 cells express only
IGFR. Differential binding to IR and IGF1R, as well
as differential activation of various IRS isoforms
leads to different responses to insulin and IGF1.
Insulin is used in cell culture at concentrations

that are nearly a hundred-fold higher than found
in blood. At such high concentrations insulin can
trigger a mitogenic response. IGF has a much
stronger affinity for IGF1R and stimulates cell
growth at much lower concentrations than insulin.
MEDIUM DESIGN |120
Industrial NS0 cells or other myeloma lines are

grown without insulin supplementation, and many
CHO cells have been adapted to grow without
insulin. It is likely that the signal transduction
pathways have been altered downstream
of IR or IGFR in those adapted cell lines.
Insulin was one of the first recombinant proteins to

be made available for therapeutic use. Recombinant
insulin has long been used in cell culture. IGF1 is
also used in cell culture, often in a commercial form.
in cell culture, often in the commercial form.
Transferrin
Typical concentration: 1 30 g / mL (MW 80kd, 10 g
/ mL=0.1 M)
80 kDa glycoprotein with homologous N-terminal and
C-terminal iron-binding domains. Binds to iron very
strongly with a dissociation constant of approximately
1022M-1.
Low interspecific potency; for human and rodent cell
lines can be replaced by other iron binding protein (i.e.
hemoglobin),
May be replaced by an iron-chelating agent, such as
citrate.
Table 10. Some Iron Chelators as Transferrin

Replacements
Ferric citrate
0.1 - 0.5 mM
Ferric iminodiacetic acid
0.001 M
Ferric ammonium citrate and

Tropolone
2 - 10 M
Transferrin is the iron carrier glycoprotein in

mammals. Ferric ion is highly oxidative, and exists
primarily bound to heme and other proteins with
iron centers. In circulation, ferric ion is bound to
transferrin. Transferrin has a very high binding
constant for iron, but dissociates readily at low
pH. Transferrin receptor-bound iron is taken up
by cells and translocated to lysosomes, where iron
is released for incorporation into cellular proteins.
For industrial processing, it is desirable that all
growth factors supplemented to culture medium be
derived from recombinant DNA expression rather
than having been isolated from human or animal
sources. Although recombinant insulin and other
growth factors have long been available, recombinant
transferrin became available only in recent years.
The specificity of transferrin binding to crossspecies transferrin receptor is not universally high.
The recombinant transferrin used commercially
is primarily of human origin. The concentration
required for cells of different species may
differ and must be empirically determined.
Transferrin can be replaced by iron chelating

agents, including citrate. Most chelating agents
have a much lower binding constant for iron than
does transferrin, and they are used in higher
concentrations than are needed for transferrin.
MEDIUM DESIGN |121
Serum albumin and other carrier proteins

Serum Albumin
used at 0.1 5 mg / mL
High interspecific potency
Fatty acid composition and content depend on method
of preparation and species.
Most defined medium use fatty acid-free albumin
coupled to specific fatty acids, particularly oleic acid or
linoleic acid.
Table 11. Transport and Carrier Proteins

Transport
proteins
Source
Structure
Effects
Serum albumin
Supplies free fatty

1-chain
acids
Plasma
(MW=68000) Detoxifyer contains
trace elements
Transferrin
Plasma
High density
lipoprotein
(HDL)
Particle
(multiple
Plasma
protein
subunit)
Low density
lipoprotein
(LDL)
Plasma
Trenscobalamin
Plasma
Binds vitamin B12
Ceruloplasmin
1-chain
Plasma (MW=
135000)
Binds copper
Hemoglobin
Red
cells
1-chain
Supplies iron
(MW=77000) detoxifyer
Particle (Apo
B)
4 subunits
(MW
~65000)
Cell Adhesion Molecules
Accepts and
transports
cholesterol and
cholesterol esters
Transports
cholesterol and
cholesterol esters
The most abundant protein in serum is albumin.

Albumin is a versatile molecule and is a carrier for
many compounds that may have low solubility in
aqueous solution. Most notably, albumin is a carrier
for fatty acids, which can be toxic when present
in free form at high concentrations. Albumin also
binds bilirubin, heavy metal ions, and other agents
that may harm cells. Albumin is probably the
most important protein that mediates scavenger
functions of serum in cell culture medium.
Recombinant forms of human albumin are also

available. However, unlike other recombinant
proteins, which are easily characterized for use in
a chemically defined medium, the diverse binding
capacity of albumin makes its complete chemical
properties harder to define. Not all albumin
preparations are the samealbumin from different
preparations may be bound to different amounts
and varieties of fatty acids or other compounds.
In addition to transferrin and serum albumin,
serum also contains other carrier proteins as
shown in the table. Most are rarely used in culture.
Transports O2
Anchorage dependent cells are employed in the

vaccine industry and in emerging stem cell and
cell therapy-based technologies. In this section we
will address adhesion of cells only to conventional
stationary surface cultivation, not microcarriers.
Although stainless steel or chemically modified
polymer surfaces are still used, the vast majority
of cells are grown on conventional plastic or glass
surfaces. Many suspension-adapted cells revert to
attachment when grown on adhesive (positively
MEDIUM DESIGN |122
charged) surfaces. Adhesion can be enhanced

in the presence of serum or by coating surfaces
with adhesion molecules. For the cultivation of
industrial cell lines, there is, in general, no need
for coating the surface with adhesion molecules.
Cells clumping and cells sticking to vessel surfaces can

be partially corrected by adding to medium:
Pluronic F68 (0.01 0.1%)
Adhesion molecules are used to promote

adhesion of various stem cells, differentiated
cells, or some highly transformed cells that do
not attach well to tissue culture flask surfaces.
Commonly used adhesion molecules, as shown
in the table, include biological molecules
(fibronectin, laminin), ECM extract (matrigel,
which is rich in laminin), and synthetic molecules
(poly-L-lysine or RGD peptide (arg-gly-asp).
These adhesion molecules may be used to

revert suspension cells to an adherent state for
cell cloning. Under adherent conditions, cells
form colonies on the surface of culture dishes
and can be easily isolated using a cloning ring.
Heparin (10100 ug / mL)
In some cases, one might want to prevent cell

adhesion to a surface. Inclusion of heparin, heparin
sulfate, or Pluronic in the medium, along with use of
a non-adhesive surface can minimize cell adhesion.
Table 12. Adhesion Molecules Used for Cell Culture

Adhesion proteins
Source
Effects
Fibronectin
Plasma, cell lines
Promotes attachment growth of mesenchymelly derived cells
Laminin
Extracellular matrix
Promotes attachment and growth of ectodermally and endodermally derived cells
Collagens (I-IV)
Skin, extracellular
matrix, placenta
Promotes attachment and growth either directly or through

the binding of other adhesion proteins
Vitronectin
Plasma
Promotes attachment and growth of a variety of cell types
Fetuin
FBS
Promotes attachment of cells to glass and plastic
Poly-d-lysine
Synthetic Polymer
Promotes attachment of many cell types (even in the presence

of serum)
MEDIUM DESIGN |123
Protein Hydrolysates
Soybean hydrolysate and peptone are commonly

used
Peptones derived from acid or enzyme hydrolysates
of casein, gelatin, meat, soy, egg and lactalbumin
have been used as supplements in cell culture
Contain a mixture of amino acids, small peptides,
inorganic ions, carbohydrates and vitamins
Possible roles of hydrolysate
Source of amino acids in the form of
oligopeptides
Some oligopeptides may mimic analogues of
signaling molecules by having non-specific
binding various cell surface receptors
Such effects may be growth-promoting or
apoptosis-retarding
Hydrolysates or extracts from animal or plant tissues

were commonly used in cell culture processes to
reduce the dependence on serum. A beef extract,
Ex-Cyte, was an excellent source of phospholipids;
however, the use of animal extracts has been largely
discontinued, and has been replaced by hydrolysates
from soy, rice, and other plants derived by enzymatic
or acid hydrolysis. Lot to lot variation among such
extracts is huge. In a transcriptome analysis of CHO
cell samples grown under different reactors, pH,
and other conditions, it was found that the specific
lot of hydrolysate overrode other experimental
variables in sample clustering. In a data mining
experiment encompassing data from more than 100
manufacturing runs, it was found that hydrolysate
lot had a strong correlation with productivity.
The roles of hydrolysates are not completely
understood. Hydrolysates are complex mixtures
of amino acids, peptides, derivatized peptides,
carbohydrates, and some lipids. They provide some
nutrients and minerals and may also act as scavengers
through undefined molecular interactions with
possible contaminants. Hydrolysates may have some
growth-stimulating or anti-apoptotic activities.
Given that synthetic peptides have been found to
interfere with signaling pathways by binding to
signaling intermediates, the possibility cannot be
excluded that some hydrolysates have similar effects.
MEDIUM DESIGN |124
Lipid Supplements
Lipids are sometimes added to serum-free culture

medium, although they must be used with caution.
Most lipids in circulation are associated with
carriers. Too much free lipid in culture medium is
not desirable. Phosphatidyl choline, phosphatidyl
ethanolamine, and/or sphingomyelin may be added
to cell culture medium in the form of liposomes
or may be dissolved in DMSO. Reconstituted lipid
supplements containing a defined composition of
phospholipids are commercially available for use in
chemically defined medium.
Medium for the Industrial Production Culture

Medium design for industrial cell culture
processes encompasses two aspects, cell
expansion and production. For cell expansion
the focus is on optimal growth and sustained
viability; for production, the objective is rapid
growth to production cell density, sustained
viability, and productivity at high cell density. For
the production of non-cell products including
recombinant proteins and viruses, the viability of
the culture at the end of production may not need
to be high; thus an extreme composition that favors
production at the expense of viability may be used.
In the past decade it has become a common practice

to use high osmolarity in the final stage of production
culture. Glucose concentrations as high as 15 g/L
(83 mM) may be used at the initiation of production
culture. At such a high level, glucose contributes
significantly to overall osmolarity; thus, the
concentration of NaCl is often reduced to compensate
for the additional glucose. High osmolarity
affects the growth rate of most cells, although a
high glucose concentration does not appear to
affect growth of some CHO and myeloma cells. As
discussed in the section on Fed-Batch cultures, most
industrial production is initiated at about 70% of
reactor volume, and concentrated nutrient mixtures
of amino acids, glucose, etc. are added during
cultivation. The concentrated nutrient mixtures
usually carry extra salts used to dissolve amino
MEDIUM DESIGN |125
acids, which additionally increase the osmolarity.

As a consequence, it is not unusual to reach an
osmolarity of 400 mOsm by the end of production.
In recent decades, the concept of industrial medium

design has undergone a major shift. The focus on
providing cells with optimal growth medium is now
restricted to applications involving cell line isolation,
banking, and expansion. A new focal point is the
design of medium for production cultures. In some
cases, even in the expansion of the seed train toward
a production scale, the medium used has been altered
from traditional optimal formulations; for example,
sometimes high glucose concentrations are used. It
has not been shown that such conditions provide
any advantage for cell expansion, but new points of
view have replaced the notion that optimal growth
conditions are likely to be found by recreating
the native niche of a particular tissue of origin.
For industrial processes, the purpose of medium

design is not to provide conditions that cells like, but
to devise conditions that can be imposed on cells to
maximize their productivity. Significant advances
toward that objective have been accomplished.
MEDIUM DESIGN |126
Methods and Strategies in Cell Line Development
Cell Line Development
With Contributions From Gargi Seth

Host Cells and Recombinant Protein Production. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Transient Transfection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Stable Cell Line Development. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
What Makes a Hyper-Producing Cell Line? . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Basic Steps for Generating a High-Producing Cell Line . . . . . . . . . . . . . . . . . . .
Basic Elements on a DNA Plasmid. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Amplification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Cell Adaptation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Stability of Selected Clones . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Automation and High Throughput Technology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Concluding Remarks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
127
128
129
129
131
134
138
142
143
145
146
Host Cells and Recombinant Protein Production

Animal cells are powerful vehicles for the production
of therapeutic recombinant proteins. Therapeutic
proteins, along with viral vaccines and cell therapy
products, are generally called biologics to differentiate
from small-molecule drugs. The process of
transforming the host cell line is critical to producing
cells of high productivity. Cells must be screened for
those that contain a superior level of production
and growth characteristics. This chapter discusses
the process of generating cell lines suitable for the
large-scale production of recombinant proteins.
Industrial expression of recombinant proteins is
generally categorized into two types: transient or
stable. Transient expression is frequently used for
the generation of small quantities of protein (up
to gram quantities), for protein characterization
METHODS AND STRATEGIES IN CELL LINE DEVELOPMENT |127
Transient Expression:
Not suitable for long term production or commercial
manufacturing
Useful for rapid production of small quantities of
research materials:
Drug candidate identification and in-vivo
evaluation
Assay development (e.g. binding assays)
or animal testing. This process does not generate

a clonally derived cell line; it temporarily
transduces an existing cell line (e.g. HEK293 or
COS) with plasmids encoding the product protein.
Conversely, stable cell lines are generated for the
industrial production of therapeutic proteins. These
Structural studies
Toxicology studies
Transient Transfection
Transient Expression Protein Production

Transgene in plasmid DNA or virus (advenovirus or
vaccinia virus) is introduced into the cell, the nucleus
where it exists as an extrachromosomal unit.
Cells are heavily loaded with vectors to increase the
production
The heterologous DNA is not integrated into the host
genome.
plasmids generally do not replicate
viral vector may be maintained inside the cell as an
autonomous replicon
cell lines must be capable of producing product of

the same quality in different batches, and in different
locations, throughout the years. Once a production
line is transduced and selected, cells are expanded to
establish a master cell bank, which is stockpiled. Cells
from a mater cell bank are then expanded further
to create a working cell bank from which frozen
vials of cells are taken for use in the production.
The banked cells, typically stored in liquid nitrogen,
are used for manufacturing purposes throughout
the production cycle. In generating the production
line, both the choice of the host cell and the choice
of the vector carrying the products gene are pivotal.
The overexpression of exogenous gene(s) in target

cells is an important technique used to understand
the functional significance of an unknown gene.
Often, the desired effects can be observed with a
transient expression of the protein, so in these cases,
stable transduction is often un-warranted and unnecessary. Transient expression is also frequently
used to produce and isolate protein products encoded
by a transgene. The aim, in this instance, is to simply
obtain a sufficient quantity of the product protein
in a short amount of time. The scale of production
in these cases is often small, but could also be fairly
large (e.g., up to tens of liters of reactor volume).
For protein production through transient

expression, the host cells are loaded with a very
large amount of exogenous DNA, such as a plasmid.
In transduction, only a small fraction of exogenous
DNA is actually taken up by cells and even smaller
fractions actually enter the nucleus where the
transgene gets transcribed. Thus, to see high levels
of exogenous protein expression, researchers
employ the use of a strong promoter to drive gene
Host Cells Most Frequently Used for Transient

Expression:
HEK-293 (human embryonic kidney fibroblasts)
COS (from green monkey kidney cells)
BHK cells
CHO cells (to a lesser extent)
Production Life of the Transient Expression System is

Usually Limited by:
Loss of DNA from the cell with time
Deleterious effect of foreign DNA on cell viability
Typical titers ranges from >1 to 100mg / l in 5 - 10 days
(~ 0.1-1 pg / cell / day)
Stable Cell Line Development
transcription and transduce large quantities of

starting material. However, if too successful, the
overexpression of some genes can be toxic to cells.
When transiently expressed, the exogenous vector

does not get integrated into the genome. After
transfection, they exist as extra-chromosomal
elements. Most plasmid vectors, except for some
episomal vectors, do not self-replicate and are
gradually lost in about a weeks time. Even some
viral vectors, such as adenoviral vectors, do not
persist over a long time and are largely lost in a
month. This is fitting for the overproduction of
protein, where the aim is to generate a short and
intense, rather than a long and sustained, burst of
protein production. Thus, in transient situations, the
percentage of cells actually taking up and expressing
exogenous DNA (i.e., the transfection efficiency)
can significantly affect the overall expression level.
HEK-293 (293; human) and COS (African green

monkey) cells are frequently used for the transient
expression of proteins, but product quality
parameters, such as glycoform profile, may differ for
materials derived from different species (e.g., human
versus Chinese hamster). However, CHO cells are
unique. Transient material produced in CHO cells is
more representative of the material produced from
stable CHO cell lines, so there has been an increasing
use of CHO cells for stable expression. Unfortunately,
the transient transfection productivity in CHO is
relatively low compared to that of 293 host cells.
What Makes a Hyper-Producing Cell Line?

Two types of cells are most commonly used as
host cells for the generation of stable production
cell lines: CHO cells and myeloma cells. Virtually
all therapeutic proteins made with these cell lines
are secreted out of the cell and can be harvested
from their culture fluid. CHO cells, which do not
secrete any protein in an appreciable amount in
their native state, must be made to develop the
capability of protein secretion while becoming
producers. In contrast, myeloma cells have well-
developed protein secretion machinery, left over

from their original purpose of secreting antibodies.
Host Cell
Protein
Secretion
CHO DG44
serum dependent, adherent, DHFR-/-
nil
CHO DXB11
serum dependent, adherent, DHFR
+/-
nil
CHO K1
serum dependent, adherent, proline

dependent
nil
SP2/0
serum dependent, suspension
plasma
cell
Myeloma
NS0
serum dependent, suspension,

cholesterol-dependent
plasma
cell
Typical Characteristics of Producing Lines

Serum independent
Suspension growth
Highly secretory, associate physiological change
Energetic
Secretory pathway
ROS-redox balance
Sustained growth in stationary phase
Ready switch to lactate consumption
Glycosylation capacity
Many scientists have studied the changes that

occur during the maturation of B-cell to plasma
cell, both at the proteome and transcriptome
levels. The cellular alterations appear to
include elevated energy metabolism, higher
protein secretion and glycosylation capacity,
as well as an increased redox balance, to
counter the effects of reactive oxygen species.
B-cells and plasma cells have only one functional

immunoglobulin gene in their genome. In the
developmental maturation process, one of
the two gene copies in the diploid alleles is
inactivated, thus preventing the production of
two immunoglobulin molecules in a single cell
(i.e. allele exclusion). Even a single copy of
a transgene is sufficient for a cell to become a
super secretor, like the plasma cell. Therefore, to
transform a myeloma cell into a high producer of
protein, one needs only to introduce a single copy
of the product gene into the appropriate locus of
the genome. The pre-existing cellular machinery
is already tuned to secrete a high level of protein.
CHO cells, contrary to myeloma cells, must have

modifications to enhance their protein secretion
machinery and to acquire characteristics of highproducing cells. In addition to bolstering the
protein secretion machinery, a high-performing
cell line should also have superior growth and
metabolic characteristics. Manufacturing conditions
differ profoundly from laboratory cultivation.
For example, the prolonged stationary phase in
a fedbatch culture, in the presence of lactic acid,
allows for a longer period of protein production
at a stage when the cell concentration is high, thus
leading to a maximal product titer. The important
question, in this case, is how to identify candidate
cell lines that harbor those desirable traits.
In the past decade, numerous transcriptome
and proteome studies have been conducted to
examine the traits leading to hyper-productivity.
Biotechnologists now realize that there is
Best Producer
Cumulative effect of
combination of
different changes
may not be the
same, some may not
be additive
Potentially a
large
combination
may give rise to
similar
productivity.
Different changes
may lead to the
same incremental
improvement of cell
characteristics
necessary for high
productivity
Energy
Metabolism
Secretion
Redox
Growth/Death
Control
Fig. 5.1: Hyperproductivity of recombinant protein in producing

cells contributed by multiple cellular functions. Many alternative
combinations of superior traits may lead to high-productivity.
probably no single master regulator that can be

turned on to make a CHO or NS0 cell a hyperproducer. Hyper-productivity is the culmination
of multiple changes in multiple cellular pathways,
such as metabolism, secretion, redox balance, and
growth/death control. The acquisition of hyperproductivity is more likely to involve diminutive
gene expression changes on a vast scale, rather than
larger alterations in only a few master switches.
Basic Steps for Generating a HighProducing Cell Line
Steps in Cell Line Creation

Transfection
Selection
Amplification
Single cell cloning
Screening
Adaption
A few basic steps are generally followed to make

a host cell become a high producer of the desired
product. First, a transgene coding for the product
protein is typically introduced to the host cell using
a plasmid. In addition to the transgene, the plasmid
carries a gene that confers a selectable trait, such
as antibiotic resistance, so that after transfection,
a selective pressure can be applied to enrich for
those cells which have internalized the plasmid. The
plasmid does not replicate in mammalian cells and
would otherwise be gradually lost as cells multiply.
By applying selective pressure over an extended
period, all of the cells that are selected for would
have the plasmid integrated into the chromosome.
After a stable cell population is obtained,

preliminary in vitro screening is then performed
to screen for clones with the highest production
levels. This is often fulfilled by assaying the product
concentration in the supernatant of a 96-well plate.
Another common practice is to grow cell clones
as small colonies on soft agar and then perform
immunoassays in situ to identify those with a larger
immunoprecipitation zone around the colony.
The next step, the amplification of transgene copy
number, is practiced in some instances (such as
when CHO cells are used), but not all (such as when
myeloma cells are used). To do this, the stable cells
are subjected to a high concentration of an inhibitor
to the amplification marker, which the cells require

for survival. This high concentration of inhibitor
kills the vast majority of cells, except those that have
multiple copies of the marker and resistance genes.
As the amplification marker multiplies, the adjacent

integrated transgene(s) are also co-replicated.
The number of copies may increase dramatically
in different regions of the genome, thereby giving
rise to high levels of transcription and translation.
Through this process, in some high-producing cells,
the transcript level of recombinant IgG heavy chain
becomes the most highly expressed transcript in the
cell. An excessive expression of protein can overwhelm
cells protein folding capacity and lead to an unfolded
protein response in the endoplasmic reticulum
and, thereby, induce apoptosis. Consequently, cells
which have not developed appropriate machinery
to handle the increased production may not survive.
Fig. 5.2: Typical steps in introducing a transgene

for generating high producing cell lines for
manufacturing.
After amplification, the selected cells not only have

multiple copies of the transgene and a high level
expression of resistance and product genes, but they
also have developed the secretory capacity to allow
for enhanced protein secretion. These clones have a
high propensity to become high producers, but not all
of them do. Hyper-productivity also requires many
other traits, such as the capability to quickly grow to
a high cell density and the capability to sustain a high
viability over a long duration in the stationary phase.
Subsequently, single-cell cloning is performed

on those surviving cells, typically by sorting
single cells into culture wells by flow cytometry.
Cloning can also be performed by dispensing cells
(approximately 0.2 cell per well) into multi-welled
culture plates, so that the probability of having
more than one cell in each well is very low. Thus,
all cells that arise from a given well all originated
from the same cell. Single cell cloning is necessary
because a pool of stable cells would have vastly
different genetic backgrounds, perhaps in the loci
of exogenous gene integration, and potentially have
different mutations caused by those integration
and other unknown aberrations. Such a mixture of
cells of genetic heterogeneity is called a cell pool.
After single cell cloning, the productivity of each clone

is then assessed and those with high productivity are
isolated. The selected clones are further expanded
for stock preparation, growth characterization and
product quality assessment. In some cases, the
producing cells are then adapted to growth conditions
more amenable to manufacturing conditions.
Methods of Gene Transfer
DNA-Calcium Phosphate Co-Precipitation

DNA and calcium chloride is added dropwise into a HEPES
buffer with sodium phosphate (1 mM). A fine precipitate
forms in 5-30 min, and is added directly to the cells (~240g/106).
Electroporation
Expose cells to a high-voltage electropulse in the presence
of DNA solution. This introduces pores in the plasma
membrane, allowing entry of DNA. Duration of pulse and
strength of electric field varies with cell type.
Lipofection/Lipid Mediated Gene Transfer

A mixture of DNA with amphipathic compound (DOTMA,
DOPE, etc), that simultaneously interacts with DNA and
hydrophobic portions of the membrane, allowing passage
of DNA into the cell.
Single cell cloning is considered critical for

establishing a production cell line, as the cells arising
from a single cell are genetically homogenous.
It is practiced regardless of whether there is an
amplification step or not. Although the host cells used
for establishing production lines are all aneuploid
and can be prone to further genome reorganization
and epigenetic reprograming, a culture of cloned
cells are much more homogenous than cell pools.
From a single cell at the beginning of single cell
clone to the end of a production run, the production
cell may have gone through more than 60 doublings.
If that is extended to the entire production lifetime,
the number of cell doublings may be greater than 80
doublings. It is important to minimize the outgrowth
of mutated cells in the original pool, during the
course of cell expansion, by single cell cloning
during the generation of production cell lines.
A number of different methods are commonly used
to introduce expression vectors into host cells.
The choice of method is dependent on cell type,
the available quantity of cells and plasmids, and
the experience of the lab practicing it. Although
different methods depend on different mechanisms
of plasmid uptake by cells, all methods require the
cytoplasmic membrane to first become permeable
to plasmids. Plasmid delivery by calcium phosphate
precipitation, cationic polymers, and liposomes
relies on direct interactions of the particles, or
lipid vesicles containing plasmid DNA, with the
cellular membrane through an endocytosislike of mechanism. In electroporation and
microinjection, physical force is used to introduce
openings in the cell membrane for DNA entry.
The Methods of Gene Transfer table lists
DNA-Calcium Phosphate Co-Precipitation
commonly used methods of DNA introduction. All

DNA and calcium chloride is added dropwise into a HEPES methods use a very high plasmid-to-cell ratio, but
buffer with sodium phosphate (1 mM). A fine precipitate only a moderate DNA concentration. While up to
forms in 5 - 30 min, and is added directly to the cells (~2- a thousand copies of the plasmid can enter cells,
40g/106).
only a small proportion actually translocate to
the nucleus and are transcribed to allow for the
Electroporation
expression of selectable marker resistance gene.
Expose cells to a high-voltage electropulse in the presence
of DNA solution. This introduces pores in the plasma
membrane, allowing entry of DNA. Duration of pulse and
strength of electric field varies with cell type.
Table 2. Estimates of Number of DNA Molecules per Cell for

Commonly Used Non-Viral Gene Transfer Methods.
Method
DNA
concentration
(g/mL)/pM
Cell
concentration
(106 cells/mL)
Number
of DNA
molecules
per cell
Calcium
phosphate
50 / 15
1.8*106
DEAE dextran
10 / 3
3.6*105
Lipofection
40 / 12
1.5*106
Electroporation
40 / 12
10
7.3*105
Basic Elements on a DNA Plasmid
Fig. 5.3: A typical vector for introducing transgene into

host cells for recombinant protein production.
Plasmids facilitate the introduction of specific

genes into mammalian cells. They also contain
elements that enhance the transcription and
translation of the product gene. Both viral and
bacterial plasmid vectors are commonly used for
introducing transgenes into mammalian cells for
research. However, the prevailing vehicle used for
introducing product gene for recombinant protein
production is a plasmid vector rather than a viral
vector. Although viral vectors are used often in gene
therapy, they are rarely employed for establishing
production cell lines for therapeutic proteins.
The expression of the product gene may be
driven by a constitutive, inducible, or conditional
promoter. Conditional promoters, driven by specific
endogenous factors or events, are frequently used
in the research of differentiation and development.
It allows a reporter gene or selectable gene to be
expressed after a particular differentiation event. It
enables the selection or sorting of differentiated cells
Elements in a Vector
Promoter
Coding sequence (often with introns) of gene of
interest
poly A signal
Selectable marker
Other elements of plasmid (cloning site, origin of
replication)
Target Gene
Positional cloning: screen library for chromosomal
markers known to flank the gene of interest, and then
walk, testing individual genes identified in the region.
PCR amplify
Promoter
Constitutive
Conditional
Inducible
Native dynamic regulated
in a population. For example, the albumin promoter

is expressed and activated specifically in liver cells.
When driving the expression of green fluorescent
protein (GFP), cells of the liver lineage become green.
For protein production, the vast majority of

vectors employ a constitutive, and very strong,
promoter. Traditional viral promoters, such
as SV40 late promoter or the CMV promoter,
are frequently used. In the past few years,
constitutive promoters, such as the promoters
of elongation factor 1 (EF-1) and glyceraldehyde
dehydrogenase (GAPDH), from CHO, have been
isolated and used in product gene expression.
In addition to a strong promoter, enhancer
elements can also be included in the intron of
the transgene construct, to ensure a high level of
transcription. It is rather common to see that at
least one intron is included in the product gene
construct. Furthermore, the DNA sequence of
the product gene is often codon optimized to
match the efficiency of translation machinery
of the host cell. Through codon optimization,
one can replace the codon for a rare tRNA with
a more abundant tRNA, to avoid the translation
rate being limited by the supply of the rare tRNA.
In addition to the promoter, enhancer, and
the product gene, the vector must also has
a selectable marker, and if amplification is
to be involved, the amplification marker.
In the course of introducing DNA into host cells,

only a fraction of them actually express the
plasmid. Although the number of plasmids entering
each cell is likely to be very high (probably in the
order of hundreds to thousands of plasmids),
the probability of their entry into the nucleus
and subsequent integration onto the genome is
very low. These plasmids are not capable of selfreplication; they only replicate after integrating
into the host chromosome. Free plasmids in the
cell are, thus, gradually degraded or otherwise lost.
After transfection a very large fraction cell
population are untransfected. To identify and select
those that have the transgene, a selectable marker is

included in the plasmid. Cells that have at least one
copy of plasmid integrated into the chromosome
and express the selectable marker gene will
survive in the presence of selectable marker.
Recessive Selection
DHFR (dihydrofolate reductase)

On DHFR deficient background
TK (thymidine kinase)
On TK deficient background
Dominant Selection
Antibiotic resistance
Neomycin
Hygromycin
MDR (multi-drug resistance)
DHFR
Reporter Gene

gfp family
-galactosidase
luciferase
secreted alkaline phosphatase
Selectable markers are classified into two

categories: dominant and recessive. A recessive
marker resides in cells with a particular genetic
background and causes a growth deficiency.
Then, the introduction of a compensatory gene
leads to overcoming the deficiency. For example,
a cell line without a functional dihydrofolate
reductase (DHFR) gene requires the additional
supplementation of thymidine and glycine in the
culture medium for cell growth. The introduction
of a functional DHFR enables it to grow without
the supplements. Therefore, after the transfection
of plasmid containing DHFR, only the transfectants
will grow in the absence of thymidine/glycine.
Similarly thymidine kinase (TK)-defective mutants
require thymidine to be included to the culture
medium. The introduction of a functional TK gene
allows for cell growth in the absence of thymidine.
Conversely, the presence of a dominant selective

agent is lethal to the cell. By introducing the selectable
marker gene to the cell, the cell is endowed with a
resistance to the selective agent. The most frequently
used selectable markers, and their mode of action in
mammalian cells, is listed in the adjacent table. Each
resistance gene encodes an enzyme, which modifies
the selective chemical agent to destroy its activity.
The phosphorylation and acetylation reactions
employed for inactivating antibiotics require
intracellular reactants (ATP, acetyl group donor).
Those enzymes are, thus, only effective in destroying
the selective agents intra-cellularly. The hydrolysis
enzyme, on the other hand, may be active even when
released into medium after cell lysis. In any case,
in the selection process, the concentration of the
selective chemical agent decreases with time and the
rate of decrease is dependent on the concentration
of transfected cells. Thus, the optimal concentration
for selection for each agent is not only dependent
on cell line but also on its concentration. Clonal

selection and population selection may have rather
different optimal concentration of selective agent.
Another class of selective agents interferes
with the uptake of a toxic selective agent. The
multidrug resistance gene (MDR) confers cells
with resistance by increasing their ability to
pump toxic substances, such as colchicine, out
of cells. Its overexpression allows for selection
from a background of cells not expressing MDR.
Table 2. Commonly Used Drugs for Selection of Stably Transfected Mammalian Cells
Antibiotic
Family
Mode of Action
Resistance gene
Mode of
resistance
Gene
Drug
size (bp) concentration
range
(g / mL)
Geneticin
(G418)
Aminoglycoside
Block protein synthesis Neomycin

by inhibiting elongation phosphotransferase
(npt)
Phosphorylation
of Geneticin
795
100 - 800
Hygromycin B
Aminocyclitol
Inhibit protein
synthesis by disrupting
translocation
and promoting
mistranslation
Hygromycin
phosphotransferase
(hpt)
Phosphorylation
of Hygromycin B
1011
10 - 400
Puromycin
Aminonucleoside
Block protein synthesis

by causing pre-mature
chain termination
Puromycin
N-acetyltransferase
(pac)
Acetylation of
Puromycin
603
0.5 - 10
Blasticidin S
Peptidylnucleoside
Inhibit protein
synthesis by interfering
with peptide bond
formation
Blasticidin S
deaminase (bsr)
Deamination of
Blasticidin S
396
1 - 10
Zeocin
Bleomycin
(Glucopeoptide)
Intercalate into and

cleave DNA
Bleomycin
resistance protein
(ble)
Bind
stoichiometrically
and prevent
Zeocin from
binding DNA
375
0.1 - 50
Amplification
The strategy is to use a mutated form of the enzyme

that has a lower catalytic activity or to use an enzyme
inhibitor
DHFR with methotrexate (MTX)
DHFR:
dihydrofolate+NADPH tetrahydrofolate+NADP
GS with methionine sulphoximine (MSX)
The metabolic enzyme (amplifiable marker) needs to
be amplified in order to supply sufficient reaction rate
for cell survival
Glutamine Synthetase:
Glutamate + ATP + NH3 Glutamine + ADP + P
Two systems are commonly used for gene

amplification in mammalian cells: DHFR and
glutamine synthetase (GS). The most commonly
used gene amplification system is based on the DHFR
gene, whose chemical antagonist, methotrexate
(MTX) can be used to drive gene amplification. DHFR
is an enzyme which catalyzes the conversion of folate
to tetrahydrofolate, a compound required for the
biosynthesis of glycine, thymidine monophosphate
and purine. Methotrexate, a folate analogue, binds
and inhibits DHFR, thereby leading to cell death in
the absence of thymidine and purine in the medium.
When cells are selected for growth in methotrexate,

the surviving population contains increased levels
of DHFR which results from an amplification
of the DHFR gene. This is also accompanied by
amplification of 10 10,000 kilobases of DNA
surrounding the site of integration. Therefore,
by introducing a gene of interest (i.e. protein
product gene) alongside the DHFR gene, coamplification of the product gene can be achieved.
The amplification process can be perform as

a single step of MTX exposure over one to two
weeks, or in multiple step-wise increases in
methotrexate concentration. As MTX concentration
is increased, surviving cells with higher degrees
of DHFR gene amplification are obtained.
Highly methotrexate resistant cells may contain
several thousand copies of the DHFR gene.
DHFR based amplification is more efficient in a

DHFR defective genetic background. Otherwise,
endogenous DHFR may get amplified without
concurrent amplification of the product gene.
Chinese hamster ovary cells deficient in DHFR
were isolated after ethyl methanesulfonate- and
UV irradiation-induced mutagenesis. These DHFRdeficient cells require the addition of thymidine,
glycine, and hypoxanthine to the media. These cells
do not grow in the absence of added nucleosides
unless they acquire a functional DHFR gene.
The glutamine synthetase (GS) selection system is
based on the biosynthetic pathway of glutamine
from the substrates glutamate and ammonia.

Most mammalian cell lines require glutamine
supplementation in their culture media to grow
since their endogenous GS activity is low. The
promoter region of GS in CHO cells is rich in
CpG and evidences indicate that GS in CHO is
silenced possibly by methylation of cytidine.
DHFR Amplification
More efficient in DHFR- background
CHO DXB11; one DHFR is deleted, the other has
missense mutation
CHO DG44: both DHFR deleted
With sufficiently high levels of MTX, amplification can
be carried out in DHFR+ background
The GS expression vector contains a glutamine

synthetase gene along with the gene of interest,
allowing for selection by growth in glutaminefree cell culture media. The GS gene is usually
driven by a weaker promoter, typically the
SV40 promoter. With a high concentration of
the glutamine synthetase inhibitor, methionine
sulphoximine (MSX), it is possible to select
for transfectants with gene amplification.
Several variations on the systems described above

have been developed for increasing achievable
expression levels. DHFR is used in conjunction
with an impaired neomycin resistance gene. After
transgene induction and under G418 selection only
cells with the vector integrated in a transcriptionally
active region will express neomycin resistant
transcripts at high enough level to survive. Since
the locus of integration is transcriptionally
active, the high expressing clones isolated after
amplification have only a few integrated gene copies.
After amplification, lasting for a week to two weeks

typically, the concentration of antagonist selective
chemical agent is reduced. With a lower level of
selective pressure the number of copies of transgenes
may also decrease and resulting in a decrease in its
transcript level and productivity. The propensity for
losing transgenes is probably affected by the loci of
integration on the chromosome, with those near the
distal end of the chromosome arm being more prone
to dislodge from genome. Usually a clone becomes
more stable after the initial drop of copy number
and product titer upon the reduction of selection
pressure. In most cases the remaining transgenes are
stable in subsequent cell cultivation. Nevertheless a
low level of selection pressure is often maintained
to suppress any possible deleterious mutants which
may have a lower copy number and faster growth rate.
Classical DHFR Amplifiable Vector
There have been a number of variations of this method.

Some methods use another resistance marker, such
hygromycin or neomycin resistance, for the first selection
of cells co-transfected with DHFR and the gene of interest,
followed by amplification by selection of amplified
cells in the presence of high concentration of MTX.
Fig. 5.4: A vector for DHFR based amplification of transgene.
Classical DHFR transfections employ a plasmid in

which DHFR is tethered downstream to the gene of
interest driven by a promoter (e.g., SV40 enhancer/
promoter). The plasmid is used to transfect CHO cells
(such DXB11) deficient in DHFR. The transfected cells are
selected in nucleotide-free medium, in the presence of
methotrexate (MTX). Subsequently, MTX concentrations
are increased to enrich for cells that have multiple
copies of DHFR and, concomitantly, the gene of interest.
The DHFR/MTX system has been widely used for the

generation of antibody-producing cells. The example
shown uses a two-plasmid system. DHFR resides on the
plasmid containing the gene encoding the light chain. Note
that instead of cDNA, the gene of interest is interspersed
between two intron sequences. The heavy chain gene,
on another plasmid, has Neo as the drug resistance
marker. The two plasmids are co-transfected into CHO
cells, often at a stoichiometric ratio that slightly favors
the heavy chain plasmid. The transfected cells (which
co-express DHFR/light chain plasmid and the heavy
chain plasmid) are then enriched using MTX treatment.
Fig. 5.5: A gene construct for introducing heavy chain

and light chain molecules of immunoglobulin gene.
Glutamine Synthetase (GS) as Selectable Amplifiable Marker

Glutamine synthetase (GS) is selectable marker in
most mammalian cells, as they require glutamine
for growth in culture. It is used frequently in myeloma and CHO cells. The transfectants are selected
by growth in a glutamine-free medium. Vector amplification can subsequently be achieved by using
the inhibitor of GS, methionine sulphoximine (MSX).
Fig. 5.6: A vector with GS selectable marker.
Directing Integration to a Transcriptionally Active Region

The locus
effect on
are more
of transgene integration influences the expression level, due to a position

gene expression. There are strategies to select clones whose transgenes
likely to have integrated into transcriptionally active region (hot spots).
An impaired neomycin phosphotransferase, which confers neomycin G418 resistance, has

been used successfully for hot spot integration. In the impaired enzyme, the translation
initiation site has been mutated to reduce its translation initiation efficiency; thus more
transcripts are needed to synthesize the same levels of proteins that confer G418 resistance.
Moreover, an artificial intron is introduced to decrease the level functionally active Neo transcripts, by increasing the
frequency of unsuccessful splicing. As a result, only those clones which have the Neo inserted into a transcriptionally
active region of the chromosome will have a high enough expression level to overcome drug selection.
Most of the selected resistance clones have only one copy of the transgene. The selected clones can be further
amplified using the traditional DHFR system. With the increased probability of transcription, fewer clones will
need to be screened to obtain high producers, and most obtained clones have low levels of DHFR amplification.
Furthermore enhancement is obtained using a single promoter (CMV) for heterologous protein and selectable
(amplifiable) marker, while having an IRES between the heterologous gene and selectable marker gene.
Cell Adaptation
From transfection to working bank takes about 4 - 6

months.
Host
cell
Target
gene
adaptation to suspension growth

anti-apoptotic gene
glycosylation modulation
Transfection
Cell pool,
selection,
cell clone
Selection of
higher and
stable producer
Initial drug testing

(chemistry, biological,
and animal)
Adaptation to
suspension
growth
Initial process
development
Adaptation to
serum-free/
Animal-component-free
growths
Working cell banking
Process development & Scale-up

feed batch/perfusion
metabolic shift
Fig. 5.7: Typical steps in cell line development.
In a typical cell line development process, a large

number of product-secreting clones are selected
and subjected to further screening of their growth
characteristics, product titer, and product quality,
glycosylation patterns, and other post-translational
modifications. In some cases, the cells do not grow
well in the culture environment used in production
(e.g., high agitation rates or altered medium
composition). They often have to be adapted to
new culture conditions by long-term cultivation
with a gradual change of environmental factors. Over
time, the cells gradually develop the ability to grow
under the new chemical and physical environment.
The presence or absence of a physical surface
for cell attachment is probably the most drastic
difference in cultivation conditions. Most normal
diploid cells used for virus production are strictly
anchorage dependent. Some cells, like myelomas
and hybridomas, are suspension cells that can
be readily grown in a mixing vessel. Many cell
lines commonly used for recombinant protein
production, including CHO, BHK, and HEK293
cells, are derived from adherent cells. Although,
not being strictly anchorage dependent, they often
prefer adherent growth given a compatible surface.
When cultivated in suspension, the unadapted

cells either fail to grow or form large aggregates
with extensive cell-cell contacts and intercellular
adhesion. They can be adapted to grow in suspension
by being cultured in shaker flasks or spinner flasks.
The initial growth rate is slow, and dispersing agents
like heparin sulfate, reduced serum (if any), or
calcium incorporation may be necessary to prevent
adhesion to the wall of the culture vessel in the early
stage of adaptation. Gradually, the growth rate is
increased and the cells eventually adapt and appear
indistinguishable from regular suspension cells.
The adaptation of cells to a new nutrient environment
is also commonly practiced in cell line development.
For example, the requirement of complex lipid
additives and growth factors may be reduced or even
eliminated through adaptation, although sometimes
these adapted cells exhibit lower productivity.
Fig. 5.8: A timeline for generating an industrial producing cell line.
Stability of Selected Clones
From thawing (2x108 cell) to production (20m3) needs

at least 16 doublings.
Stability is tested over 40 doublings
Normal diploid cells used in vaccine production are

stable within the accepted population doublings
Continuous cell lines have variable karyotypes in
culture
Mutations and epigenetic alterations occur in cultured

cells at relatively high frequencies. Some of these
events are affected by culture conditions. For example,
many types of stem cells are prone to differentiation,
even by changes in cell density, oxygen tension,
growth factor concentration, and cell aggregation
state, depending on the particular cell type.
Cells used for human vaccine production are
mostly diploid and are considered to be stable
under established cultured conditions. These cells
do not exhibit any visible phenotypic alterations
or macroscopic chromosomal abnormalities
within the accepted range of doubling, or until
they reach senescence. The stability of these
cells is not a general concern in bioprocessing.
In contrast, the extensively selected, hyper-producing

recombinant cell harboring transfected, and often
amplified, transgenes have a higher propensity to lose
their high productivity. To begin with, the aneuploid
host cells used to generate those hyper-producing
cells are less stable than their diploid counterparts.
In addition to having abnormal chromosomal counts,
many of the chromosomes also have macroscopic
structural
aberrations.
Such
chromosomal
alterations accumulate over time, generating cells of
divergent karyotypes from the same parental line.
Divergent karyotype and chromosomal abnormality
Cell Line Stability Issues

Productivity decrease over time
Product quality changes over time
Karyotype changes over time
Microscopic chromosomal aberration occurs over time
The last two issues may not be critical for protein
biologics, but are of concern for cell therapy
applications
Possible Causes of Instability in Recombinant Protein

Productivity
Mutation, especially in intergenic region in pivotal
product gene loci
a producing cell may contain many copies of
product genes, but maybe only a dominating few
contribute to most product transcript
Epigenetic silencing of pivotal genes, at DNA level or at
histone reprogramming level
Loss of copies of product gene due to deletion or
chromosomal rearrangement
is, thus, an inherent nature of producing cells derived

from continuous cell lines. Even though production
lines may not be stable (in relation to ploidy), they
must maintain their production characteristics
related to growth, productivity, and product quality
over the number of population doublings required
for creating sufficient cell banks. They must also
be hardy throughout the thawing process, as well.
Assuming that a product life time of 10,000 runs
at 10,000 liter scale, with a final cell concentration
of 1010 cells/L, the selected cell clone will have to
double nearly 80 times. This number of doublings
greatly exceeds that required of a fertilized
egg to grow into a hamster, a mouse, or even a
human adult. In that long duration of time, the
occurrence of mutations, epigenetic changes,
and chromosomal rearrangement in some cells
of the population is unavoidable and probably
cannot be eliminated with todays technology.
In discussing cell line stability, we decided to
focus on property changes that affect productivity
and product quality. The critical component for
sustaining productivity over time is to prevent
any cell with a lower productivity from overtaking
the population, or to prevent a very high rate
of productivity loss in the majority population.
A gross and rapid change of productivity in a large

fraction of cells may occur upon the removal or
reduction of selective pressure after transgene
amplification. This problem is alleviated by
employing a lower degree of amplification and by
establishing the clone only after the copy number of
transgene has stabilized. For long-term stability, one
could carry out serial cultures for 40 to 60 doublings
and examine the productivity, as well as the transgene
copy number. From a thawed liquid-nitrogen frozen
cell bag of 1010 cells to a production reactor, cells
may undergo 15 doublings, so a 40-doubling test
of stability will certainly give sufficient margin.
The stability of product quality is more difficult to

assess. Mutations in genes affecting product quality,
such glycosyl-transferases, may occur and lead
to a subpopulation of cells that produce inferior
Production cell lines are tested for their ability to

retain the product gene in the genome and produces
the product
The focus is production stability, not cell/genome
stability
product. Mutations causing protein sequence

alteration may occur in one or more copies of
product genes in all, or a subpopulation of, cells.
Since not all copies of product genes in the cells
are transcribed and translated at equal efficiency,
not all mutations of product genes in a production
cell population will be manifested to the same
degree. With deep sequencing technology, one may
be able to detect such mutations, even at minute
level, in the consequent producing cell population.
Automation and High Throughput Technology
PerkenElmer - Basic Model
Tecan -- more automated with multi-plate capabilities

and computer interface
Fig. 5.9: Examples of high throughput cell clone

screening system.
Developing a cell line for production purpose is

a very labor-intensive process. To increase the
probability of obtaining a very high producer, a
large number of cells need to be isolated at every
step, which involves productivity variation from
successive rounds of selection and amplification.
In the past decades, lab automation and highthroughput technology have become an integral part
of bioprocess development. Liquid and cell handling,
both cell pool and cell clones, quantification of
product titer, data acquisition, data processing and
analysis, and archiving have all becoming automated.
Many of the automated systems are based on culture

plates, or wells, and resemble other liquid handling
systems for high throughput chemical screening.
The difference is that a incubation system, with
temperature and atmospheric control for gas mixture
and humidity, is necessary. Robotic arms are often
used to move plates onto the working stage and
allow multiple manipulations to be performed on
multiple plates without human interference. In many
cases, the system is installed inside a clean room or
clean hood to minimize microbial contamination.
The culture handling system is usually integrated

with an assay system to assess product titer and cell
growth. Multi-step assays and screening protocols
can be performed by transferring culture fluid
into automated assay systems. The results can be
directly integrated into culture handling systems
to further expand wells or plates selected for
Basic liquid handling manipulations

Distribution of liquid to 96 - 384 well plates
Sampling/removal of liquid
Transfer from plate to plate
Cherry picking transfer from well to well
More complex models are capable of:

Cherry picking
Multiple plate handing (i.e. movement of plates
from a hotel to the pipetting stage)
Multiple steps performed sequentially (e.g. DNA
preparation protocols)
Other add-ons like PCR machines, incubators,
spectrophotometers, etc
further investigation. An integrated microscopic

imaging system can provide the added capability
of determining anything from clonal colony
growth to colony morphology to ensuring that
cells picked from each well are single-celled clones.
Another type of automated system integrates cell

cloning with product titer assessment by performing
cell screening on agar plates. In this case, the
secreted product molecules (mostly antibodies) are
entrapped in agar that contains antibodies against
the product. A halo ring of immunoprecipitation
zone is formed around the colony. The size of the
halo ring reflects the amount of product secreted.
Image analysis is then used to extract the data
for selecting high producing clones to pick.
Concluding Remarks
Under best culture conditions, a hyperproducing
industrial cell line derived from CHO or myeloma cells
can secrete 50-100 pg protein/cell-day, a level which
rivals professional secretors in vivo. Such remarkable
cell lines are created through the combination
of optimized genetic constructs, selection,
amplification, cell clone screening, and the insight
of picking the right clones. Although the specific
productivity of the producing cells has increased
by about one order of magnitude, the methodology
of generating high producing cells has remained
largely the same in the past three decades. The entire
process is still empirical and very labor intensive.
High-throughput cell handling and screening
systems allow for the screening of a large number of
potential high producing clones in the early stages
of cell line development. However, subsequent steps
of adaptation, growth characterization, and testing
of stability is still labor intensive. The use of host

cell lines, which have been adapted or modified to
harbor all desirable growth characteristics, have
greatly reduced the need of adaptation. There is
an increasing movement toward miniaturizing
cell culture while still simulating large-scale
reactors, although this progress is still limited.
The advance in genomics has brought about a
fundamental change in the way we can study the
process of cell line development and brightened
the prospects that we can gain mechanistic insight
into hyper-productivity. This knowledge may allow
us to quickly select the right clone by examining
the transcriptome or genome of the candidate cells.
With more tools for genome engineering becoming
available, it may also become feasible to impart on
the cells favorable genome-wide modifications.
Stoichiometry and Kinetics of

Cell Cultivation
Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Cell Mass and Composition. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Cell Mass and Size. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Material Balance on Cell Growth. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Variation in Cell Volume . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Amino Acid Composition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Intracellular Fluid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Growth of Mammalian Cells. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Quantitative Description of Cell Growth & Product Formation . . . . . . . . . . . . . . . . . .
Stoichiometric Ratio and Yield Coefficient. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Integral Cell Concentration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Kinetic Model of Cell Growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
A Model Describing Growth and Production. . . . . . . . . . . . . . . . . . . . . . . . . . . .
Monod Model and its Derivatives. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Environment, Kinetics and Stoichiometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
147
148
148
149
150
152
153
154
156
158
159
162
162
164
165
166
Introduction
Cultured cells take up nutrients to generate energy
and to make more cell mass and products. For
manufacturing, it is important to supply a sufficient
quantity of nutrients. These nutrients allow cells
to grow and produce product while minimizing
the formation of waste product. To make the
manufacturing process efficient one must also
produce the targeted product quantity within a
given period. Therefore, one must not only know
how much nutrients to supply, but often also how
fast to deliver in order to sustain the production
environment. We use stoichiometric principles to
determine how to supply the correct quantity of
nutrients; and use kinetic principles to guide the
process along a desired path. Awareness of these
concepts is key to an overall understanding of how to
culture cells. This chapter discusses stoichiometry,
STOICHIOMETRY AND KINETICS |147
kinetics of cell growth, and also gives the typical

values of stoichiometric and kinetic parameters
commonly encountered in cell culture processes.
Cell Mass and Composition

Cell Mass and Size
Table 1. Typical Dry Weight of Cells
Bacterial
10-12 g / cell
Yeast
10-11 g / cell
Average Animal Cell
3-6 x10-10 g / cell
Table 2. Average Composition of an Animal Cell

Pg/Cell
Range
Percentage % of Dry
Biomass
Wet weight
3500
3000 - 8000
Dry weight
600
300 - 1200
Protein
250
200 - 300
10 - 15
~50 - 70
Carbohydrate
150
40 - 200
~1 - 5
~5
Lipid
120
100 - 200
~1 - 2
~5
DNA
10
8 - 17
~0.3
~2
RNA
25
20 - 40
~0.7
~4
Water
Volume
55 - 80
4x10-9cm3
Both microbial and mammalian cells vary widely in

size. In general, the dry biomass of bacteria, yeast,
and animal cells is in the order of 10-12, 10-11, and 5
x 10-10 g per cell, respectively. The most abundant
chemical species in a cell is water, accounting
for 90% of the volume of plant cells, 80 85% of
animal cells, and 70% of bacterial cells. Since cell
cultivation is carried out in aqueous environments
and the amount of water taken up by cells during
growth is extremely difficult to assess, the material
balance on cell culture is typically performed only
on dry matter, excluding water. Our discussion on
stoichiometry will be largely based on dry biomass
of cell number, as commonly practiced in cell culture.
Macroscopically, cells are made of a few classes

of macromolecules (protein, DNA, and RNA) or
macromolecular assemblies (primarily lipid bilayer
membranes). These organic matters constitute
the vast majority of the dry mass in a cell. Protein
molecules constitute the largest portion among
them, providing the machinery for DNA, RNA, and
protein synthesis. Protein molecules also serve as
the structural components of the cell and execute
all of the catalytic, transport, and communication
functions. The lipid content of an animal cell is
greater than in a bacterium. The abundance of
organelles contributes to their higher content and
their lipid bilayer membrane in an animal cell.
Intracellular carbohydrates exist as oligosaccharides

on many proteins and lipids. Carbohydrate also
exists as ribose in DNA, RNA, and nucleotides (e.g.,
ATP, GTP, etc). Only a small fraction exists in a free
(or phosphorylated) form. The cellular content
of carbohydrate is harder to estimate, because
it usually exists as a part of other molecules.
Material Balance on Cell Growth
Growth of Biomass Involves:

Consumption of nutrients
Production of new biomass
Excretion of products
Glucose
Amino acids
Other
macronutrients
(lipid,
nucleotides,
etc.)
Micro-organic
nutrients
(vitamins)
Bulk salts
Trace minerals
Oxygen
CELLS
Biomass
Product
Carbon dioxide
Water
Lactate, NH3
Excreted amino
acids
Example of the equation for

hybridoma growth:
C6H12O6 (glucose) + 0.15C6.14H12.36N1.50O2.08
(weighted average of amino acids)
+ 0.34C5H10N2O2 (glutamine) + 1.39O2
2.37CH1.97N0.26O0.49 (cell mass) + 0.0058CH1.83N0.14O2.06
(antibody) + 1.53CO2 + 1.28H2O + 1.44C3H6O3
(lactate) + 0.16NH3 + 0.13C4H7NO2 (alanine)
Such a formula is used when one performs metabolic
flux analysis.
The principle of material balance holds true in cell

culture. The total mass of inputs and outputs, and
the amount accumulated in a system, is always
in balance. The two most common, and most
abundant, nutrients in cell culture are glucose and
glutamine (although some cultures do not require
glutamine). They serve both as constituents of
cell mass and as sources of energy. Other common
inputs to cell culture processes include lipids,
lipid precursors, vitamins, and salts; these will
be discussed in the Medium Design chapter, as
these nutrients supply key cellular constituents,
but contribute less to generating energy.
To generate energy, glucose is converted to lactate,

CO2, and H2O through glycolysis, the TCA cycle,
and the pentose phosphate pathway. Glutamine is
deaminated, releasing NH3, before its carbon skeleton
is used for energy metabolism. Consequently, the
accumulation of metabolites, including lactate, NH3,
CO2, and H2O, is commonly seen in cell culture. In
many cases, the amino group from the metabolized
glutamine and other amino acids are exported as nonessential amino acids (such as alanine, asparagine,
and proline), in addition to being excreted as NH3.
Energy metabolism satisfies the energetic needs of

making biomass, through the biosynthesis of DNA,
RNA, protein, and organelles. Other important
energy-intensive aspects of cellular events are
the uptake of nutrients and the sustained balance
of cellular osmosis and membrane potential. As
will be discussed later, the cells cytoplasmic
and mitochondrial membranes have a negative
electric potential that must be maintained, at
the expense of energy, to sustain cell viability
The process of growing cells and producing
a product can be formulated into an overall
biomass synthesis equation. This equation can be
viewed as, essentially, the apparent composite
of all reactions involved in generating energy and
synthesizing biomass. At the center of the reaction
will be biomass, so a formula for the cell mass
can then be established, based on its elemental

composition. Usually, we use a formula that neglects
all elements except for carbon (C), hydrogen (H),
nitrogen (N), and oxygen (O). This formula is only
useful for describing the ratio of those elements.
One can arbitrarily assign the stoichiometric
numbers to give them different formula weights.
In the example shown, the stoichiometric
number of carbon is chosen to be 1. Others may
prefer to assign the formula mass to be 100.
The inputs in a cell growth process include
all nutrients consumed by cells to proliferate.
Since only C, H, N, and O are considered in the
stoichiometric equation, we consider only glucose,
glutamine, and other amino acids. Other minute
media components containing C and N (such as
vitamins or nucleotides) are neglected. Instead
of writing down all amino acids separately, one
may also use a weighted average, according to
the stoichiometric ratio of their consumption.
The outputs include new cell mass that has been

generated as the result of nutrient consumption,
as well as the metabolites and product that have
been excreted. One can write a molecular formula
describing the formation of the protein product, by
knowing its composition. Typically, the metabolites
excreted also include lactate and ammonia,
as well as some non-essential amino acids.
Variation in Cell Volume

Table 3. Size of Animal Cells
Cell Type
Volume (m3)
Hybridomas
Endothelial Cells
Trypsinized and reattached
before spreading occurs
12 - 20
1400 - 2500
17
2000
Chinese Hamster Ovary

cells (suspension)
1200 - 1800
Chinese Hamster Ovary

Cells (anchored)
1300 - 1800
Human foreskin fibroblasts

(FS-4)
Diameter (m)
7000
~14
The volume of a typical animal cell is a few pico

liters (about 1,000 times larger than bacteria). The
average cellular diameter ranges from 10 to 20 m.
Even for the same cell line, one can detect great size
variations over a range, since cells immediately,
before and after mitosis are about two times
different in their size. At a given time in a growing
culture, cells are in different stages of the cell cycle
and their size distribution is somewhat larger than
two fold. For aneuploid cells, the distribution of
size is typically greater than normal diploid cells.
The distribution of cell size changes with the
growth stage. Rapidly growing and quiescent
cells may have different sizes. Furthermore, in
Cell Volume exhibits a distribution at any point in

time
a culture, cells that lose viability often become

visibly smaller, as measured by flow cytometry.
Cell size varies amongst different cell types. Many

types of stem cells are fairly small. Their nucleus
spans more than 70% of the cell diameter and
their cytoplasm is relatively small. Liver cells and
antibody-secreting plasma cells are at the other end
of the cell size spectrum, and contain a significant
amount of cytoplasm for protein secretion.
Dead cells are often smaller

Varying with culture stage
It is instructive to remember that the volume of

a sphere (which is a reasonable approximation
of a cell) is proportional to its diameter raised
to the third power. Therefore, cells that are
twice as large in diameter are eight times
larger in cell volume. Although cell number is
traditionally used for the quantification of cell
concentration, it may not sufficiently capture
the difference when comparing different
processes, in which cell sizes are very different.
Fig. 6.1: Cell size change during a batch

culture
Amino Acid Composition

Table 4. Amino Acid Composition of Cells and IgG
Cell Composition Standard
IgG
Mean
deviation composition
ALA
9.03
0.32
5.31
ARG
4.74
0.32
2.43
10.08
0.59
0.26
0.04
12.62
0.63
GLY
9.14
0.57
6.98
HIS
2.22
0.07
1.67
ILE
5.73
0.35
2.43
LEU
9.00
0.68
6.83
LYS
6.85
0.49
6.98
MET
2.27
0.14
1.37
PHE
3.73
0.31
3.49
PRO
5.51
0.58
7.13
SER
6.19
0.16
12.90
THR
5.42
0.22
7.74
TYR
2.73
0.14
4.10
VAL
6.54
0.27
9.10
ASN
ASP
CYS
GLN
GLU
3.49
3.95
2.43
5.01
5.16
Microbial and plant cells often grow on simple

carbon sources supplemented with an inorganic
nitrogen source, such as ammonium or urea. The
cells convert inorganic nitrogen to all 20 natural
amino acids that are used to make proteins. Animal
cells lack the capability to make 11 to 12 of those
20 natural amino acids. These essential amino
acids must be supplied for animal cell culture,
to enable them to grow and make products.
Thus, knowing the amino acid composition of
cells, and of the protein product, is important.
The protein content and composition of cells change

under different growth conditions; however, they are
seldom measured. Nevertheless, literature values of
some cells are available, as well as the amino acid
composition product, IgG. Given target levels of
biomass and product to be produced, a stoichiometric
amount of all essential amino acids must be supplied.
In addition to essential amino acids, which must
be provided, non-essential amino acids are usually
also supplied. However, these can be derived
by metabolic transformation from other amino
acids and can be considered substitutable.
Intracellular Fluid
Table 5. Intracellular Concentrations of Amino Acids
mM
mM
Ala
0.2 - 2.0
Lys
0.1 - 0.6
Ang
<.05
Met
0.01
Asp
0.4 - 0.8
Ornithine
0.120
Asn
0.4 - 0.8
Phe
Asp
0.306
Proline
0.137
Citrulline
0.036
Ser
0.149
Glu
0.3 - 12
Thr
0.1 - 4
Gln
0.05 - 4
Tyrosine
0.059
His
<0.05 - 0.09
Valine
0.171
Ile
0.3 - 0.5
Leu
0.1 - 0.4
0.3 - 0.5
Table 6. Approximate Concentrations in Cellular

Environment
Interstitial
(mM)
Intracellular
(mM)
Na+
140
6 - 14
4.0
100 - 140
Ca2+
1.2
0.01
Mg
0.7
3 - 20
108
28.3
10
11
0.5
2+
Cl
HCO3
HPO42- H2PO4SO42Phosphocreatine
Camosine
Amino acids
14
2
Creatine
0.2
Lactate
1.2
1.5
Glucose
56
0.05
Protein
0.2
3-4
Adenosine triphosphate
1.5
Water constitutes 70 80% of total cell volume.

The soluble components in water also make up
a large fraction of the biomass. The intracellular
fluid
contains
electrolytes,
carbohydrates,
amino acids, metabolism reaction intermediates,
nucleotides (ATP, ADP, etc.), and many other
components. Most amino acids are present at
the 0.05 0.5 mM range in the intracellular fluid.
The vast majority of intracellular amino acids
reside in cellular proteins and only a tiny fraction
exists as free amino acids in intracellular fluid.
Typical concentrations of other major soluble

components in intracellular fluids are listed in
Tables 5 and 6, along with the typical extracellular
environment conditions, in vivo. Potassium,
magnesium, and phosphate are present at high
concentrations. A large fraction of Mg2 + is associated
with ATP, which is typically present at the 1 3 mM
range. In addition to free phosphate, phosphate is
also present in phosphorylated compounds (DNA,
RNA, nucleotides, phosphorylated sugars, lipids, etc).
Many inorganics, including phosphate, potassium,

and magnesium, are present at much higher
concentrations intracellularly, compared to the
extracellular fluid or culture medium. As cells grow,
they take up nutrients at large enough quantities
that they accumulate in the intracelluar fluid. It
is important to ensure those nutrients are also
supplied in sufficient quantities. By knowing their
intracellular content, the stoichiometric amount
required to produce the biomass can be estimated.
In addition to the major inorganic species (K+,

Na+, phosphate, Mg2+, and Cl-), many minute
inorganic elements are also constituents of cellular
components, including iron, copper, selenium, zinc,
cobalt, etc. Many primarily exist as a prosthetic
group of proteins. These elements must also be
supplied in enough quantities to generate biomass.
Unfortunately, the cellular content of those elements
are seldom reported and may vary widely among
different cell types or even under different culture
conditions. For example, the zinc content is much
higher in pancreatic cells than in other cells. Similarly,
Growth of Mammalian Cells
iron is rich in muscle and red blood cells. Without

quantitative data on the cellular contents of those
elements, one can only resort to titration experiments
under defined culture conditions to see whether the
supply of those elements is limiting the cell growth.
A cell culture process can be described by its cell
growth curve, nutrient consumption curves, and
product concentration profile. Cell growth, in
general, is divided into different growth stages: lag
phase, exponential growth phase, stationary phase,
and death (or the decline phase). The exponential
growth phase is characterized by a linear increase
in cell concentration on a semi-logarithmic plot over
time and is easy to determine. In many cases, the
extent of cell expansion in a culture is small (only 3
5 fold increase in cell concentration). Also, the time
point for the transition from lag phase to exponential
phase and from the exponential phase to stationary
phase may not be clearly depicted in a growth curve.
Fig. 6.2: Growth phases in a batch culture
Table 7. Doubling Time of Some Culture Cells

Cell type/contitions
Doubling Time
Mouse embryonic stem cell
~12 hr
Human Diploid Fibroblast (10% FBS)
24 - 40 hr
Human Diploid Fibroblast (2% FBS)
40 - 60 hr
CHO K1 (5% FBS or in rich medium)
16 - 24 hr
NSO cell
16 - 24 hr
Some cultures may experience a slow, or no-growth,

period ranging anywhere from a few hours to a few
days. This may be caused by using the inoculum
from cultures that had already reached a stationary
or decline phase, or by inoculating cells into vastly
different media or culture conditions. Low inoculum
cell concentrations may also cause poor initial
growth, due to insufficient conditioning factors in the
culture medium. In general, anchorage-dependent
cells should be inoculated at a minimum of 105
cells/mL or 104 cells/cm2; while suspension cell
cultures are generally started at about 105 cells/mL.
The exponential phase is marked by a constant cell

growth rate, or doubling time. Doubling times of
different cells span over a wide range, even under
favorable culture conditions. For instance, while
mouse embryonic stem cells divide every 11 12
hours, cells commonly used in bioproduct production
double every 15 30 hours, depending on medium
composition. Some human fibroblastic cells take
nearly two days to double their number, even under
optimal conditions with a high serum concentration.
After a period of rapid cell growth in the exponential

phase, the cell growth rate reduces and the culture
enters the stationary phase. Many reasons may
cause this transition, including the exhaustion or
suboptimal supplementation of key nutrients in
the culture, and/or the accumulation of growthinhibitory metabolites. A flat cell growth curve in
the stationary phase may indicate that a culture
that has ceased to proliferate. It may also reflect
a balance between cell growth and death. In the
latter case, the growth curve is characterized
by a constant viable cell concentration, along
with an increasing concentration of dead cells.
In the decline, or death phase, viable and total cell

concentrations decline due to an exhaustion of key
nutrients, the accumulation of metabolites to an
inhibitory level, or adverse culture conditions (such
as high osmolality). The growth behavior at a late
stage of culture often varies, depending on whether
cells are adherent or in suspension. Many anchoragedependent cells grow substantially slower once
cell density on the surface approaches confluence.
They can also sustain a confluent cell density, over
a period of days, without entering the death phase.
Conversely, cells grown in suspension, especially
those with an inherently strong receptor-mediated
apoptotic mechanism, often enter the death phase
soon after the viable cell concentration peaks.
In a production process, the rate of product
formation is at its highest when cell concentration is
maximal. A good production practice is to reach the
target cell concentration quickly, while sustaining
the culture as long as possible at that point.
Quantitative Description of Cell Growth & Product Formation

Specific Rates
Growth rate (G) (change in cell concentration per unit
time, number of cells/L-hr; or gm of cells/L-hr)
(Eq. 1)
G = dx
dt
x: viable cell concentration
Specific Growth Rate () (number of cells/cell-hr or

gm cell/gm cell-hr)
dx = nx
dt
(Eq. 2)
n = 1 dx
x dt
Doubling Time
ln x2 = n (t2 - t1)
x1
(Eq. 3)
td = ln 2 = 0.693
n
n
Specific Nutrient Consumption Rate (gm nutrient/gm
cell-hr)
(Eq. 4)
qs = - 1 ds
x dt
s: substrate (nutrient)
concentration
Specific Product Formation Rate

dp
qp = 1
x dt
(Eq. 5)
p: product concentration
A cell culture process can be characterized by its

cell growth, nutrient consumption, and product
accumulation profiles. To quantitatively describe cell
culture kinetics, three classes of quantities are used:
concentrations of components (e.g., cells, nutrients,
metabolites, and product), activity parameters (e.g.,
specific rates of growth, nutrient consumption,
and product formation), and stoichiometric ratios.
After obtaining the concentration profiles of key

process variables, the next step is to calculate
the rate of change of cell, nutrient, and product
concentrations. These are referred to as volumetric
rates because they are normalized to culture
volume. A quickly-changing culture has a high
volumetric rate, which may be the result of having
more cells, or by having cells that are more active.
Usually, cell volume constitutes only a very small

fraction of the volume in a suspension cell culture.
The volumetric rate is essentially based on the liquid
volume. In some cases, such as with high-density
solid microcarrier culture, the volume occupied by
solid beads is large; thus liquid volume and culture
volume are not equal. Usually the concentrations of
nutrients, metabolites, and product are measured
based on liquid volume, not on culture or reactor
volume. In such cases, the volume used for
different kinetic parameters must be clearly stated.
Another set of descriptors, called the specific

rate, is normalized to a per cell basis. These
activity parameters describe how active each
unit of the cell is for activities like making new
biomass (specific growth rate), consuming
glucose (specific glucose consumption rate), or
producing lactate (specific lactate production rate).
Cell growth is autocatalytic, meaning the rate is

dependent on the cell concentration. Because
cell concentration may be measured in different
ways, such cell mass, cell number, or even cellular
DNA or protein, the specific growth rate may
also be based on different cell measurements.
A Model Considering Cell Death

dxv = nx - dx
v
v
dt
(Eq. 6)
dxd = dx
v
dt
(Eq. 7)
xt = xv + xd
(Eq. 8)
qs = - 1 ds
xv dt
(Eq. 9)
n = 1 dxt
xv dt
(Eq. 10)
d = 1 dxd
xv dt
(Eq. 11)
With Cell Lysis

(Eq. 12)
dxv = nx - dxv - m x
v
v v
dt
dxd = dx - m x
v
v d
dt
(Eq. 13)
v: Dead Cell Lysis Rate
Different measurements may give somewhat

different specific growth rates, because cell size,
cell mass, and cellular content are not necessarily
proportional to each other at all culture stages.
It is important to note the difference between

volumetric cell growth rate (the number of
cells/L/hr) and the specific cell growth rate (,
number of cells/cell/hr or hr-1). The two are
related quantities; one describes how fast the
culture is increasing in cell concentration, the
other describes how active cells are proliferating.
The cell doubling time is easily obtained from

experimentally-determined growth curves. By
separating the variables x and t into two sides of the
differential equation and integrating with respect
to time and cell concentration, respectively, one
obtains the relationship between cell concentration
and time, given a constant specific growth rate (Eq.
3). The doubling time is the time period it takes to
develop a two-fold increase in cell concentration.
In the case that non-viable cells are present in a
significant portion of the cell concentration, a specific
death rate, , can be incorporated into the equation
for cell balance, with xd representing the dead cell
concentration (Eqs. 6, 7). In some rare cases, cell
lysis occurs in culture. Lysed cells are not observable
by cell counting. To describe the growth kinetics of
such a culture, one can also include a cell lysis term.
The specific nutrient consumption rate is obtained by

dividing the volumetric nutrient consumption rate
(ds/dt) by the cell concentration. The specific product
formation rate is defined, similarly. In all cases, the
cell concentration used to calculate the specific rate
is the viable cell concentration. In other words, one
assumes that dead cells are metabolically inactive.
Experimentally, the specific rates can be calculated

from the changes in concentration over time.
These specific rates are important for describing
the dynamics of various activities in cultures.
Stoichiometric Ratio and Yield

Coefficient
Yield of Biomass on Substrate

Y = Dx , dx
Ds ds
x
(Eq. 14)
Yield of Product on Substrate

Y =
p
Dp dp
,
Ds ds
(Eq. 15)
Stoichiometric ratio of lactate to glucose

a = DL , dL
DG dG
(Eq. 16)
Stoichiometric ratio of product to substrate

(Eq. 17)
Description of the State of the Cultures

Physiological state
There is no unique or universal definition for
physiological state. In general, the kinetic
parameters described above are sufficient to
describe the cells physiological state
Metabolic state
The stoichiometric ratios can be used to describe metabolic state
In evaluating a process, we need to assess the

efficiency of material conversion; that is, how
much of a raw material is actually converted to
the product or to cells. If we know a theoretical
maximum of the conversion efficiency, or know
its historical value, we can then assess how much
the current process can be further optimized.
A yield coefficient is the ratio of the quantity
of the product or cell produced, to that of the
raw materials used.
A yield coefficient for a
cell can be based on different input materials,
e.g., glucose, ammonium, or other nutrients.
In microbial processes, yield coefficient is

frequently used to evaluate material utilization
efficiency. The yield coefficient is given a symbol
Yx/s or Yp/s, for cell mass or product, respectively,
and is based on the consumption of a particular
substrate. These variables are key indicators
of process efficiency, because substrate cost is
often a major portion of the total product cost.
The efficiency of substrate utilization is essential.
Yield coefficient is rarely used in animal cell culture

processes. Cell concentration in a typical animal
cell culture process is rather low and is seldom
measured in mass. Furthermore, bulk materials
that are used by cells, i.e., glucose and amino acids,
are not the major contributors to the cost of goods.
The stoichiometric ratio of various nutrients and

metabolic products is more frequently used in
cell culture processes. Under different metabolic
conditions or in different growth stages, cells utilize
various nutrients differently and change the amount
of different metabolites produced. The stoichiometric
ratios of various nutrients and metabolites are
indicative of such alterations in metabolism.
For example, the stoichiometric ratio of lactate
to glucose, e.g., the ratio of the amount of lactate
produced to that of glucose consumed, is a
strong indication of energy metabolism being
at a highly glycolytic or highly oxidative state. If
most glucose is channeled through glycolysis to
lactate, the ratio is close to two moles of lactate

per mole of glucose. Conversely, if most glucose
is directed toward the TCA cycle for aerobic
oxidation, the ratio will be close to zero, while the
stoichiometric ratio of oxygen to glucose will be
closer to six moles of oxygen per mole of glucose.
The stoichiometric ratio and yield coefficient,

based on a given pair of compounds, can be
expressed in different units, e.g., mol/mol or g/g.
Integral Cell Concentration
Product accumulation rate in a culture can be

described by multiplying the specific product
formation rate by cell concentration (Eq. 18).
Integrate over the culture period from t0 to

tf, to obtain the product concentration at tf .
If qp is constant, one can take it out of integral.

One can see that the final product concentration is
proportional to the integral of cell concentration.
In a plot of cell concentration (x) vs. time, the
integral is the area under the curve of the x curve.
This is often called the integral cell concentration.
Integral Cell Number is the Area Under the Growth Curve
With the assumption that qp is constant, integral

cell concentration is thus proportional to
product concentration, and can be used as a first
approximation estimate of product concentration.
(Eq. 18)
(Eq. 19)
Example of Experimental Data Processing and Plotting of a Fedbatch Culture

Typical process data include those from on-line and off-line measurements. On-line data are often continuously
recorded except that some control actions (e.g. turning on base pump or oxygen valve) may be discrete in time.
Off-line measurements are invariable only on discrete time points. The measured data should be plotted to discern
inconsistency and outliers. Then the calculated (or derived) data are also displayed. The example data shows that the
oxygen uptake rate (OUR) peaks slightly before cell concentration, since OUR is a more sensitive indicator of cells
activity. O2 flow rate often reflects OUR. We also see that even though specific rates may change over time, some
stoichiometric ratios remain constant.
Fig. 6.4: Plots of process data of a typical cell culture
Example of Stoichiometric Ratios as Indicators of Metabolic States

Myeloma cells were grown in batch or fedbatch cultures. In the fedbatch cultures, glucose and glutamine
were at lower concentrations, initially. Once the concentration reached the set point, concentrated
glucose or glucose/glutamine was fed continuously via computer control to maintain the concentration(s)
around the set point. The final product concentration, the amount of glucose, glutamine, and oxygen
consumed, and the total amount of lactate and ammonium produced were used to calculate the
stoichiometric ratios. The duration of fedbatch culture was substantially longer than the batch culture.
It is notable that more glucose, glutamine, and oxygen were consumed in the fedbatch culture, because the
culture lasted longer and the cell concentration was higher. However, from the stoichiometric ratio, one can
see that the lactate to glucose ratio decreased from the initial value, similar to the batch culture, indicating a
metabolic change by limiting glucose at a low level. This also indicates that by controlling glutamine, one can
affect lactate production. This metabolic change is confirmed by the oxygen/glucose ratio. The lower lactate/
glucose ratio was accompanied by a higher oxygen/glucose ratio, suggesting that more glucose was channeled
into the TCA cycle for oxidation. Glutamine control also resulted in a lower production of ammonium.
Stoichiometric ratios change under different
metabolism
A combination of stoichiometric ratios can reflect

the metabolic states of cells.
In a fedbatch culture with glucose or glucoflutamine

level control, the stoichiometric ratio change can
reflect metabolic shift
Table 8. Typical stoichiometric ratios of hybridoma cells under different metabolism in batch and fed-batch cultures
Batch
Growth Data
Initial glucose conc. (mM)
17
1.4
1.4
Initial glutamine conc. (mM)
0.3
0.3
0.55
0.55
~0.5
0.2
2.4
12
10.5
Antibody conc. (mg / L)
60
200
Glucose consumption
12
46.5
27
Glutamine consumption
17
10.1
Oxygen consumption
24
143
122
Lactate production
37.4
17
Ammonium production
9.7
4.5
Lactate/glucose
1.5
1.60 0.16
0.90 0.05
Oxygen/Glucose
1.85
1.9 6.0
2.7 4.7
Ammonium/Glutamine
0.75
0.5
0.31 0.10
Alanine/Glutamine
0.35
0.34 1.35
0.08 0.42
Osmolality (mosm / kg)
~350
410
400
Glucose conc./set point (mM)

Glutamine conc./set point (mM)
Maximal viable cell conc. (10 cells / mL)
6
Consumption/
Production
(mmole/mmole)
Stoichiometric
Ratio (mmole/
mmole)
Fedbatch with
Fedbatch with Glucose
Glucose Control Glutamine Control
Kinetic Model of Cell Growth

Utility of Mathematical Models
1. Summarizing experimental data
2. Probing concepts and testing hypothesis
3. Predicting and optimizing processes
A
used
mathematical
for
model
different
can
be
purposes:
(1) To summarize a large volume of experimental

data. Data, or plots of data, are extremely difficult to
describe in words. If data are fit to a mathematical
model, regardless of whether the model is empirical
or mechanistic, the behavior of the data can then be
recreated, given the model and value of the parameters.
(2) To explore concepts and test hypotheses. When
we study a physical system and its behavior (such as
growing cells consuming glucose), we first develop
a verbal description of the system and the behavior.
We may then propose a hypothesis, also in verbal
form. The verbal description can be translated
into a mathematic form. The mathematical
model can then be used to explore the systems
possible behavior under different conditions.
(3) To predict the behavior of the systems, given a model

with sufficient complexity. The model can be used to
predict regions of parameter space that have not been
experimentally tested, previously. It can also be used
to optimize or control the dynamics of the system.
With an increasing emphasis on the notion
of Quality by Design, the application of the
mathematical model will gain importance. Proper
applications of well-posed models will help
explore system behaviors and define optimal
operating regions, in a potentially vast design space.
A Model Describing Growth and

Production
If we consider only a cell, a nutrient (e.g., glucose)

and a product (e.g., a recombinant antibody) as the
three most important variables in a culture system,
the mathematical model will consist of three material
balance equations for the concentrations of the three
species. The change of cell concentration will be due
to growth, which is described by the multiplication
of specific growth rate and cell concentration.
Similarly, the rate of change of substrate and product
concentrations is described by multiplication of
Balance Equations for a Batch Culture Growth Kinetics

Cell balance
(Eq. 20)
Glucose balance
(Eq. 21)
Product balance
(Eq. 18)
Lactate balance
(Eq. 22)
Models describing growth rate or production rate
dependence on environmental factors
Cell density dependence
(Eq. 23)
Nutrient dependence
(Eq. 24)
Inhibitor dependence
(Eq. 25)
What is needed to develop a mathematical

description of a cell culture processs?
A description for cell growth (and death), product
formation, nutrient utilization
Growth model - dependence of specific consumption
rate on the controlling variable (growth rate, nutrient
concentration, etc.)
Product formation - dependence of specific production
rate on controlling variable
Experimental data - the model will have some
parameters (such as half saturation constants). The
experimental data are used to determine the value of
those parameters
Material balance equations for state variables (the
concentrations of cell, nutrients, product, inhibitors,
etc.)
the respective specific rate and cell concentration.
These three equations describe only the balance

of those three species. To describe the dynamics
of the system, we will need to have description of
how the parameter (or the specific rate) changes for
important variable(s). For anchorage-dependent
cells, the growth rate is dependent on the extent of cell
confluence on the surface. Models for anchoragedependent cells, thus, attempt to describe the
dependence of the specific growth rate on cell density.
For process design or optimization purposes,
one first needs to identify the variable that most
affects the process outcome and then develop a
relationship between specific growth rate and this
variable. For example, if the glucose concentration
is an important variable, then one can develop a
model describing the relationship between growth
rate and glucose concentration. Then, as the glucose
concentration changes, so does the specific growth
rate. On the other hand, if the most important factor
affecting the cell is lactate concentration, then one
will need a balanced equation for the production
of lactate and a model relating cell growth (), and
possibly also specific glucose consumption and
product formation rates, to lactate concentration.
Once a model is available, one would still need
experimental data to identify the parameter values
in the equations. Then, with appropriate initial
conditions (i.e., the initial concentrations of cell
mass, glucose, lactate, and product), the system
behavior can be explored by simulation of the model.
A number of mathematical models have been

tested for describing animal cell growth in culture.
Most are based on Monod-type models that were
traditionally used to describe microbial growth.
Some employed more complicated structured or
segregated models. These models are empirical
in nature.
However, using these models to
predict growth conditions outside the range of
conditions under which the kinetic parameters
are obtained may not give reliable prediction.
Monod Model and its Derivatives

Modod Model
n=
nmax s
Ks + s
(Eq. 26)
Fig. 6.5: Growth rate dependence on rate limiting

substrate based on Monod model
If s Ks, then max

Two parameters, (max and ks), define the relationship
between and limiting substrate concentration s.
A Monod model employs two parameters, max

and ks, to define the relationship between the
specific growth rate, , and the limiting substrate
concentration, s. The specific growth rate is affected
not only by substrate concentration, but is also a
function of pH, temperature, the status of other
nutrients or growth supplements (e.g., serum), and
the presence of waste products. In applying a Monod
model, one assumes only substrate s is limiting and
that all factors are supplied in enough quantities.
The term-limiting substrates have been used

in two different contexts: 1) a stoichiometriclimiting nutrient refers to the nutrient that
is first depleted in a batch culture and whose
depletion causes the cessation of growth; or 2)
a rate-limiting nutrient is the nutrient whose
concentration restricts the growth rate of cells.
The equation gives a saturation type of kinetic

behavior; meaning increases with increasing
is also a function of pH, temperature, nutritional
status, (i.e. serum), waste products; it may also depend substrate concentration, until it reaches a maximal
upon cell density for anchorage-dependent cells which
value. At low concentrations, increasing the rate
are subject to contact inhibition.
limiting nutrient concentration increases the specific
Notes
growth rate, linearly. At very high concentrations,
These models are all empirical models that can be used the specific growth rate is constant at max.
for first order approximation of cell growth
Under most process conditions growth is not limited by A Monod model can be incorporated into the
nutrient availability
balance equations describing the batch culture
For stem cells and primary cells, growth factors have
more profound effects than nutrients
growth kinetics to provide a relationship between

the equations for cell growth and substrate. Indeed,
the resulting equation can be used to describe
batch growth when glucose is used as the limiting
substrate. Starting at a high concentration of
substrate, cells grow at max; as the substrate
concentration decreases so is the growth rate. The
simulated growth curve will show the entry into
stationary phase as the substrate decreases to a level
where is substantially reduced and, eventually,
growth ceases when the substrate is depleted.
However, animal cells require multiple nutrients for
proliferation. In addition to glucose, glutamine is
also a major nutrient. Many amino acids and other
Multiplicative Saturation Kinetics Model
n = nmax :
s2 : s2
K1 + s1 K2 + s2
(Eq. 27)
(Eq. 28)
nutrients are also required, although often most

are provided in excess and are not limiting. Monod
models are modified to describe the dependence of
growth rate on multiple nutrients. Many include
multiplicative terms to incorporate multiple
substrate utilization and product inhibition.
In the multiplicative model for cell growth, two

substrates, S1 and S2, are considered to be growth
rate limiting. Each substrate has its corresponding
half saturation constant for saturation kinetics. In
most common applications, the two substrates are
glucose and glutamine. The model can be extended
to consider metabolite inhibition. For example, in the
case where A and B are inhibitory metabolites. Most
models consider lactate and ammonia as inhibitors.
Environment, Kinetics and Stoichiometry
Specific Product Formation Dependence on Growth
+
q=
p
(Eq. 29)
We have discussed balance equations for cells,

substrates, products, and the model relating
growth rate to limiting substrate. Together, they
can be used to describe cell growth in culture. For
simple microbial systems, this is often sufficient.
However, for animal cells, the specific substrate
consumption and production rates are profoundly
affected by their environment. One, thus, also
seeks to describe the relationship between qx,
qp, and key variables affecting their behavior.
The kinetics of product formation in microbial

systems are frequently categorized, according to
their relationships to growth rate. qp is considered to
be influenced by two factors: a growth-associated
term, , which describes dependence on specific
growth rate; and a non-growth associated term,
. Depending on the relative magnitude of and
, the production can be growth associated, nongrowth associated, or mixed growth associated.
Such classifications of production kinetics are

useful for microbial fermentation of amino acids,
organic acids, and antibiotics. However, their
applicability to animal systems are limited. In
general, the specific production rates of animal cell
culture products are less sensitive to growth rate.
For some recombinant proteins, the productivity
is somewhat higher in the stationary phase of

fedbatch cultures. At that stage, many factors,
including osmolality, lactate, and CO2 concentration,
all have deviated from optimal growth conditions.
It is possible that some of those factors exerted a
stronger effect on productivity than growth rate.
Concluding Remarks
Stoichiometric and kinetic relationships among
process variables are important in describing
different cultures. From an experimental perspective,
these parameters and variables provide a basis
for quantitatively comparing different processes.
They also allow the simulation of a cultures kinetic
behavior under different growth conditions. A high
productivity process often includes a cell growth
phase and a prolonged stationary phase, in which
cells are kept at a high concentration and high
productivity state. Growing evidences suggest
that a switch of a cells metabolism from high
Another complication in applying a Monod-type of

model is that cell growth is rarely limited by substrate
concentration in cell culture bioprocess. Cells enter
the stationary phase often due to the accumulation
of metabolites (e.g., lactate and ammonium) or the
accumulation of salts (Na), due to base addition or CO2.
Lactate and ammonium are produced from glucose
and amino acid (primarily glutamine) metabolism,
while Na accumulation arises from base addition
to neutralize lactate and maintain pH. CO2 comes
from both cell metabolism and pH control actions.
glucose consumption and high lactate production
to low glucose and lactate consumption is the
key to a high product accumulation. The simple
models described in this chapter can be made
more effective models by including a mechanistic
description of how a cells metabolic shift occurs in
response to environmental factors. Incorporation
of a mechanistic model for describing cell culture
bioprocesses will enhance our physiological
understanding of cell metabolism and increase
the utility of models for optimizing the process.
Cell Culture Data Analysis

Process Data Analysis and PAT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Data Processing Pipeline. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Standardized Templates for Data Logging and Processing . . . . . . . . . . . . . . . . .
Cell Culture Data Processing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
A Typical Spreadsheet for Analysis of Cell Culture Data. . . . . . . . . . . . . . . . . . .
Data Visualization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Mapping Data to Pathways. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
167
168
168
169
170
172
173
174
Process Data Analysis and PAT

Process analytical technology (PAT) is gaining
greater attention for its potential role in enhancing
bioprocess robustness.
PAT encompasses: 1)
the acquisition of data pertaining to process and
product attributes, related to both productivity
and product quality, through on-line and offline measurements; 2) the analysis of process
data in each run; 3) developing and employing
mathematical models to relate process variables
to process outcome for a better control; and 4) the
data mining and recognition of hidden patterns
of behavior in historical data for further process
enhancement. Every element of PAT has been
used in process development for decades. The new
emphasis is on integrating each individual element
to create a more comprehensive understanding of
the process and to generate useful information.
Modern production plants are thoroughly
electronically monitored and controlled. They
CELL CULTURE DATA ANALYSIS |167
employ instrumentation in the reactor for

process monitoring, as well as to measure raw
material and product quality. A significant
challenge in the application of PAT is the
accumulation of the vast amount of data, which
can potentially impede process understanding.
The key to successfully implementing PAT
is, thus, to produce a better pipeline for data
processing and to have a better appreciation
of methodologies for bioprocess data analysis.
Data Processing Pipeline

Standardized Templates for Data
Logging and Processing
Use Standardized Template in Data Processing

Speed up routine data analysis
Automatically perform calculations
Automatically generate standardized plots
Automate data regression and calculation of
estimated values.

Ensure consistency and accuracy
Uniform terminology, units, graphs

Built-in formulas minimizes mistakes in calculations
Up-to-date set of key plots eases communication
and data interpretation within teams.
Facilitate data archiving
Develop a format which allows easy transfer

of data to other programs for analysis and
visualization
Incorporate other important parameters, such
as passage number, medium composition, and
operating information.
Additional Consideration of Spreadsheet
Templates
Metabolic flux analysis, visualization process

modeling or modeling.
Data archiving:
Searching
Experimental / run details
Standardized upload / download platforms
The importance of having a standardized data format

for data collection and archiving cannot be overstated.
After collection, data is processed to remove faulty
values (such as data resulting from wrong entry
or sensor failure). The raw data obtained from
different sources, such as from different analytical
instruments, may have different units or be formatted
differently and are homogenized. The data are then
further used to calculate derived variables and
are plotted for visualization. Finally, the data are
archived and sometimes employed in data mining,
for generating more process insights. Throughout
this transformation, the data are transferred
between many different software platforms.
Many of those data processing steps are often
repeated over and over; not only in manufacturing
settings but also in research environments. They
are often performed by different workers at
different times and at different manufacturing
sites. Thus, setting a standardized data format
and uniform data processing protocol is critical
to increase efficiency and minimize frustration.
What exactly should be standardized? Highest on
the list are the terminology, nomenclature, units
of quantities, and data format. The same object
should be named the same way, consistently across
all data sets, and should use the same symbol, units,
and equal numbers of significant figures in values.
When treating data, the formula must be uniformly

applied to different data sets. However, inadvertent
mistakes may occur through incidents, such
as errors from formula entry. To facilitate the
workflow, the use of a uniform data format and the
minimization of inadvertent errors are advisable to
set up data entry and process templates for various
programs. The same templates should be used by
all colleagues performing similar experiments.
Cell Culture Data Processing

Three Types of Data
Measurement Data
Cell concentration
Nutrient and metabolite concentrations
Process parameters: pH, DO, temperature, etc.
Calculated Integrated Data

Cumulative nutrient consumption and metabolite
production
Integral viable cell concentration (IVCC)
Calculated Differentiated Data

Specific rates
Stoichiometric ratios
Two Types of Calculations

In-process calculations
Process monitoring
Troubleshooting and diagnosis
Post-process calculations
Data smoothing for analysis
By setting up a template, various tasks can be

automated.
For example, upon the entry of
experimental data into the spreadsheet template,
the related specific rate and stoichiometric
ratios can be automatically calculated, and their
profiles over time can be plotted instantaneously.
With templates for different software programs
in place, even data transfer from one software
program to another can be automated.
The first level of bioprocess data is raw data that
has been acquired from various measurements,
including the concentrations of cells, nutrients
and metabolites, pH, and oxygen. Subsequently,
stoichiometric ratios and specific rates are calculated
over a time course to discern the trend of metabolic
and productivity changes in the entire culture. The
stoichiometric ratio can then be determined from
the amount of nutrients consumed over a time
period, and is an integral quantity in nature. Specific
rate is the rate of change of the quantity, further
divided by cell concentration. It is determined from
the slope of the curve of the quantity produced
or consumed over time. Specific rates are,
therefore, quantities determined by differentiation.
Except in simple batch cultures, the concentration

profiles of nutrients and products are often not
monotonic functions. Even culture volume may not
be constant. Feed medium is added occasionally,
resulting in step changes in nutrient and metabolite
levels. Typical culture profiles, thus, often entail
a step increase in nutrient levels after feeding
and a step decrease in metabolite levels, due
to a dilution caused by the volume of the feed.
The first step in data analysis is the compilation of

cumulative data pertaining to nutrient consumption
and
metabolite
production/accumulation.
Instead of using concentration profiles to perform
calculations, the cumulative amounts of nutrients
consumed and the amount of product produced
are calculated. The cumulative curves are mostly
monotonic, except for those few nutrients or
metabolites that are consumed at one time and
produced at another. A concentration profile with
many step changes cannot be easily subjected to
regression. Cumulative curves have no step changes,
thus allowing a regression to be performed on the
entire profile. The regression equation is then
used to calculate the slope for the determination
of stoichiometric ratios and specific rates.
Cell culture processes often extend over a long
duration, from 5 to 15 days up to a couple
of months. Data processing starts while the
process is still in progress and continues after
its completion. In-process data analysis allows
for potential outliers and faulty conditions to be
detected and corrected. Post-process analysis often
involves additional chemical analyses to provide
more process insight. A template spreadsheet
can be used for the first stage of data analysis.
A Typical Spreadsheet for Analysis of

Cell Culture Data
A spreadsheet template shall contain columns of

raw data entry followed by columns of cumulative
data for all measurement data for nutrients which
are consumed and products. Any time point of row
that incurs feed addition will incur a total mass
balance that takes volume change into consideration.
A simple way to account for volume change is

to perform calculations in total mass balance
(concentration of nutrient: x volume) instead of
on concentrations alone. In this case the amount
consumed between two time points is simply
StVt- St+1Vt+1)=St cumulative consumption is then
the sum of St over time. Forgetting to account
for volume change in material balance is a
Cumulative Data
t
# \ $ V $ dt . IVC
common mistake in fedbatch culture data analysis.
The calculation of cumulative data can be automated

once the measurement data are entered. The next
0
t
k
set of columns are specific rates. Specific rates are
Si,t = qi,t $ x $ V $ dt = Vt0 $ si,t0 - Vt $ si,t + Vfk $ sbest
fk
calculated for cumulative data by regression.
0
0
The regression of cumulative data can be automated.
Si,t: Cumulative amount (mMole or g) of nutrient i
Usually a third order polynomial fitting works well
consumed or produced at time t
for most data. However, inspection is necessary
V: Volume of culture
to ensure a good fit. The measurement data,
si: Concentration of component i in the culture broth
Vf: Volume of feed medium added
cumulative consumption/production and specific
sf: Concentration of component i in the feed medium
rate data shall be all automatically for visualization.
k: Total number of feed medium additions up until time t
IVC: Integral Viable Cell number
Upon the calculation of cumulative data
IVCt =
t-1
+ \t $ vt - \t - 1 $ vt - 1
stoichiometric ratios are also automatically plotted.

This allows for detection of metabolic changes in
Si,t - Si,t
S
T
i
m
=c
culture. If a stoichiometric ratio deviates significantly
TS j t
S j,t - S j,t
from historical data, it may also serve as a diagnosis
ai,j: Stoichiometric ratio for nutrient i with respect to nutrient alert for checking pasable process abnormality.
Stoichiometric Ratio
2
j at time t
An add-on to the spreadsheet template is an algorithm

for metabolic flux analysis. If the measurements
include all the major carbon compounds, glucose,
lactate, glutamine and other amino acids (if not all, the
majority), ammonium, then material balance can be
performed on the nitrogen balance. Carbon balance
will require the measurement of CO2 produced in
metabolism and is not easily done without isotope
labeling. If oxygen consumption data is available,
one can assume that R.Q. being 1.0 and set CO2
production to be same as oxygen consumption.
Fig. 7.1: Plots of cumulative consumption data
Specific Rates
Two-point specific growth rate calculations and
specific nutrient calculation for fedbatch culture
qs =
From the extent of carbon, nitrogen balance one

can assess the reliability of some stoichiometric
ratio data. If the carbon and nitrogen is reasonably
closed the data can then be further subjected to
metabolic flux analysis.MFA algorithm is typically
in MatLab or other mathematical solvers. The Excel
template can build in an exportable table for ready
transfer of the specific rate data to those programs.
S - S1
1
dS
1
$
.
$ 2
x $ V dt
+
x2 $ V2 x1 $ V1 t2 - t1
2
Slope calculation from curve of regression data
Data Visualization
Multidimensional/Interactive Data
Exploration
Process data are intrinsically multidimensional and
should be examined in multiple dimensions (e.g. time
course and stoichiometric ratios) to provide different
insights
Data from multiple cultures can be consolidated and
examined for trends
Software for visualization interactive for multiple
dimensional viewing analysis E.g.: Spotfire
DecisionSite.
Visualizing the data is critical for developing a

deeper understanding of the effect of various
parameters on process performance. Each cell
culture run usually entails many measurements
over multiple time points. Instead of browsing
data through tables, we plotted each quantity
as a concentration profile over time. We also
plotted data of one variable against another
variable, to specifically examine the ratio of
specific rates or the stoichiometric ratios.
As data accumulate over time, it is even more
important to plot data of multiple runs together,
so that different runs under the same or different
experimental conditions can be compared
easily. In such analyses, mathematical and
statistical tools are important; however, the
importance of data visualization in the analysis
of multiple runs cannot be overemphasized.
When working with a large set of data from

multiple runs, visualization software is very useful.
Quick access to data, the rearrangement of data
into different combinations of dimensions, or the
filtering of data by different process performance
or other criteria can greatly facilitate deeper
insight. In the plot shown, lactate concentration
profiles from over 250 runs are colored by the
final product titer. One can see that high-titer runs
mostly consume lactate in the late stage of culture,
while the low-titer runs produce lactate. The trend
is easily seen when all data are plotted together.
Fig. 7.2: Plots of archived historical data for discerning

process patterns
We then plot the specific rates of glucose

consumption and lactate production at different
time points, for all runs. It can be seen that lactate
consumption (negative values) occurs only when
glucose consumption is also low. One can further
see that even when the cells are consuming lactate,
the glucose consumption rate is still significant.
With the aid of a visualization tool, data

from all of these runs can be easily plotted in
different ways. To take advantage of the data
trove from large number of runs, a means
for quick visualization is very important.
Mapping Data to Pathways

Pathway related data is another type that its also exhibit three different glucose consumption
analysis can greatly benefit from a visualization rates. By plotting fluxes on to a metabolic map,
tool. An example is the fluxes through different it can be seen that high glucose consumption
reaction steps in various pathways obtained from rate leads to high lactate production, high
metabolic flux analysis (MFA). Another is the gene glutamine consumption and high TCA cycle flux.
expression transcript level data from microarray Such visualization will become almost essential
analysis. Presentation of the flux data on a when dealing with very complex pathways
pathway map makes it easy to compare metabolic like glycan biosynthesis on proteins passing
change over time or under different conditions. through Golgi apparatus before being secreted.
An example below shows flux distribution in
cells grown under three different conditions that
Metabolic Flux Analysis/Pathway Mapping
Metabolic flux analysis

uses culture data to
estimate fluxes of
intracellular species
through metabolic
pathways. It is useful to
transfer these numbers
onto a map of central
metabolism to visualize
the differences among
calculated values.
Mapping of data into
formats that incorporate
physiological information
facilitates interpretation
of complex datasets.
This example compares
fluxes of lactate
production state and
lactate consumption
state.
Fig. 7.3: Metabolic chart for visualizing metabolic fluxes
Concluding Remarks
Data analysis holds the key to understand and
improve the process. A large amount of software
for data processing, analysis and visualization is
available. Setting up a consistent and efficient
way of processing and plotting the data can
be hugely beneficial. A set of templates are
included in the template folder, along with some
sample data to demonstrate the calculation and

plotting. They can be used as a starting point for
modification to meet individual needs. The use of
template will also make it easier to compile and
analyze historical data accumulated over time. A
good practice of data processing and analysis is
fundamental to process analytical technology.
Visualization of Glycan Profiles
The Pathways leading to N-glycans form a

complex network
A small number of enzymes are involved to
form a large number of glycans
Most enzymes are used multiple times
Most glycans, intermediate glycans and late
glycans may all appear in the final product
Visualizing the distribution of glycans can
help deduce the plausible paths taken to
from each species
Comparison of Intracellar and

Culture Medium N-Glycan
Fig. 7.4: Visualization of glycosylation fluxes
Metabolic Flux Analysis

in Cell Culture Systems
Overall Material Balance for Reaction Systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Chemical Reaction Systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
MFA on a Cellular System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Utility and Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
General Approach. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
175
176
181
181
183
187
Overall Material Balance for Reaction Systems

In cultivating cells, it is sometimes necessary to
know where and through which pathways the
absorbed nutrients have traveled. This allows
one to manipulate the distribution of nutrients
and to optimize the process. One useful tool
for performing such tasks is material balance
analysis. When such balance is used to analyze the
distribution of materials in biochemical systems
(either on whole cells or on specific pathways),
it is called metabolic flux analysis (MFA).
Metabolic reactions are chemical reactions.

They are governed by stoichiometric principles.
However, there are special characteristics in cellular
systems that set them apart from chemical reaction
systems. In a test tube or in a chemical reactor,
the reactions (i.e., the conversion of reactants to
products) occur directly in the fluid phase. For cells
in culture or in tissue, the reactants (i.e., nutrients)
are delivered to individual cells. Products are also
METABOLIC FLUX ANALYSIS|175
formed in and excreted by individual cells into

the fluid phase. The reactions take place not in
a continuous phase, but in many discrete cells.
By applying MFA to cells in culture, one could

theoretically combine all of the cells mass into one
entity and could consider all cells as one biotic phase
and the fluid as one abiotic phase. All of the nutrients
taken up by cells (not just the nutrients added to the
culture) and products excreted into the fluid are
considered to be inputs and outputs, respectively.
Sometimes, MFA is applied to a pathway or a group

of pathways. In this case, the balance is applied to a
system, which does not have a physical boundary, like
a cell or even a group of cells. The pathway of interest
(for example, glycolysis or mitochondrial reactions)
occurs in many different locations within the cell
mass. And yet, we would treat all reactions of interest
in ALL cells as a system and perform a balance on it.
Another unique characteristic of the cell system is
that cells are growing and expanding in biomass, as
well as in cell volume. The output of cellular reactions,
thus, includes cell biomass. Furthermore, because
the volume of cells is expanding, it has a dilution
effect on the concentration of cellular materials.
Chemical Reaction Systems
In this chapter, we will first use a familiar

chemical reaction system to emphasize the
notion that the fundamental concept of MFA
is material balance. This will then be followed
by a discussion of the basic steps in MFA.
Consider a chemical reaction system, in which

multiple reactions are taking place simultaneously
and we want to determine the extent, or the flux, of each
reaction. We can solve this problem by performing
measurements on the concentrations of compounds
in the inflow and outflow streams of the system
and their relative flow rates. Then, we can apply
material balance on the stoichiometric equations
for all reactions known to occur in the system.
In the example shown below, we aim to determine
METABOLIC FLUX ANALYSIS |176
Material Balance on Reaction Systems

Example: Incomplete combustion of CH4 to CO2, CO and H2O in a batch reactor. Two reactions can possibly
occur. We dont directly measure the extent of each reaction, but we can measure the inputs and outputs at
the beginning and the end of reaction, and see how much has been consumed (i.e. the amount of reactants
left at the end subtracted from the amount put in initially), and how much has been produced. From these
balances we can calculate the flux of each reaction.
Known Reactions
Output
CO2 + 2 H 2O CO + CO + H O
CH 4 + O2 CH 4 + 2O2
1
2CH 4 + 3O2
2CO + 4 H 2O
If all the inputs and outputs (i.e., the amount of CH4, O2 consumed and that of CO2, CO, H2O produced) are
completely balanced, the fraction of CH4 going to each reaction can be determined. On the other hand, if the
material balance is not closed, there will be uncertainty about the solution, and the distribution of materials
can only be estimated.
Case I. 4 moles of CH4 and 7 moles of O2 are combusted to produce 2 moles each of CO2 and CO and 8
moles of H2O.
In this case, all three elements involvedC, H and Oare completely balanced.
The only
unknown can be easily calculated using a balance on any element, e.g., use C balance:
mole C
mole C = 2 mole CO2

1
4 mole CH 4 1
mole CO 2
mole CH 4
= 0.5
In Case I, in principal, one does not need to know the quantities of all inputs and outputs to find the solution.
From three elemental balances (C, H, O), we can have three linear equations. As long as there are only three
unknowns, we can also solve for . For example, even if the quantities of CO and H2O are not measured, the
solution will be the same.However, in the real world, there is always a high degree of uncertainty about the
accuracy of measurement; the overall balance is always an important check of the validity of the results.
Case II: 4 moles of CH4 and 7 moles of O2 are combusted to form 1.5 moles of CO2 and 1.6 moles of
CO2. The amount of H2O produced is not measured. Determine the fraction of methane completely
combusted.
In this case, the input and output materials are not balanced. There are a total of 4 moles of C combusted
but only 3.1 moles are accounted for in the products. This presents a problem regardless of whether H2O is
measured.
The cause of input and output materials not being balanced is not always known. It may be due to measurement
error, or possibly other reactions have occurred but are not accounted for are not accounted for. In this case,
we can only get a plausible answer of how much CH4 goes to complete and incomplete combustion. If we
know the extent of various measurement errors and the amount of H2O, we can get a more reliable estimate.
the fluxes of methane for two reactions, which occur

simultaneously in the furnace, using measurements
of compounds in the inlet and outlet.
The
concentration differences between inlet and outlet
of the furnace give the value of net consumption and
net production. They are the inputs and outputs into
the reaction system. In case 1, the carbon, hydrogen
and oxygen entering the system are all accounted
for in the products. One can easily determine how
methane distributes itself between the two reactions.
On the other hand, as in many cases, the material
balance from the measurement is not closed. This
happens frequently because of measurement
errors. Sometimes, it can be caused by compounds
in inputs or outputs that are not accounted for. For
instance, there might have been some unknown
reaction that occurred, leading to the emergence
of products that are not measured and included
in the output. Under such a situation, uncertainty
on the outcome of material balance inevitably
occurs. One then has to rely on the knowledge
of the reaction system to make appropriate
assumptions, or to perform further measurements.
A Systematic Way to Solve Material

Balance Problems
Setting Up Material Balance Equations
We will use the methane combustion in a furnace

as an example to illustrate how to set up equations
for MFA. Although the problem is so simple
that it can be solved using a piece of paper and a
pencil, the approach shown is more systematic
and will be suitable for calculating fluxes for a
large reaction system, such cellular metabolism.
We will perform material balance on the materials

consumed and produced, in order to determine the
fluxes for the reactions involved. First, we define
the system, including its inputs and outputs (i.e., net
consumption and production), as well as the chemical
(or metabolic) reactions involved. In the example
shown, two reactions are listed, for which the chemical
formula of the compounds and the stoichiometric
coefficients of the reactions are all known.
The rate of consumption and production is given
Step 1.
Write material balance equations for every species
the symbol q. In MFA, qs are always the specific

rates. Consumption is designated as a negative
value and production as a positive value. We
express the flux of each reaction as J. By denoting
the flux of reaction 1 as J1, the fluxes of methane,
oxygen, carbon monoxide, and water reacted in
reaction 1 are -2J1, -3J1, +2J1 and 4J1 respectively.
Their fluxes are related by the stoichiometric
coefficients (-2, -3, 2, and 4) of the reaction.
Next, we set up a material balance equation for

each species, which is present in the system.
Each compound involved in the system is given
a differential equation describing the change of
its concentration in the reactor as the balance of
its consumption, production, and other input and
output, if any. The balance equation for CH4, thus,
reflects that the rate of change of CH4 is the sum
of its input (qcH4) and its reaction fluxes (-2J1).
Quasi-Steady State
Step 2.
Pseudo-Steady State Assumption
Stoichiometric Matrix, Flux Vectors and

Solution
The equations set up above are differential equations.

For a short duration of time, the concentration of
all species in the reactor (or in the cell) may not be
changing very rapidly. In that case, the left-hand
side of the differential equations can be assumed
to be negligible (e.g., the system is assumed to be
at a pseudo-steady state). With this assumption,
the left-hand side of all differential equations
becomes zero. The equations, thus, become a
simple system of linear algebraic equations.
To solve a large set of equations, it is more efficient
to perform matrix operations. In these cases,
the matrix is called the stoichiometric matrix.
The stoichiometric matrix is the collection of
stoichiometric coefficients organized by reactions in
one dimension, and by species involved in reactions
in the other dimension. In the example shown,
there are two vectors for J1 and J2, respectively.
The two columns in the matrix correspond to the
stoichiometric coefficients of reactions 1 and 2. Each
row in the matrix represents the stoichiometric
Step 3.
Write the equations in the matrix form
Matrix Form
Specific rates vector r:
Positive : production
Negative : uptake
Stochiometric Matrix A:
Each column is a reaction

Each row is a species
Solve simultaneous equations AX=Q

Over specified system: Least-squares method
coefficients in the balance of a compound in the

system. For example, the balance of CH4 is the net of
-2J1 (+0J2) and the output from the furnace -qCH4. If a
compound exists only in the reactor but is not part
of the input or output, then its value in the vector
on the right hand side is zero. In applying MFA to
cell systems, many reaction intermediates exist
only intra-cellularly (such as glucose-6-phosphate
and fructose-6-phosphate), so their q is zero.
Under most conditions, the fluxes are unknown. The

inputs, outputs, and qs can usually be measured.
In setting up flux analyses, one typically considers
only the branching points (called nodes) of the
reactions, in which one reactant is split into two or
more reactions, or two reactions lead to a common
product. In a linear pathway, the material flow
is easily related by a stoichiometric relationship
along the path. For example, the flux of glycolysis
is a linear pathway, but only if the small diversion
of glyceraldehyde-3-phosphate to glycerol-3phosphate for lipid biosynthesis is ignored. At steady
state, hexokinase flux is half of that of pyruvate
kinase flux and, on a molar basis, is defined by the
1:2 proportion of their stoichiometric coefficients.
There is no need to set up balances for reactions
along the pathway between the two reactions.
Next, the equations that are not independent are

removed from the system of equations. The resulting
system of equations may be overspecified (in the
case of having more independent equations than
unknowns), underspecified (in the case of having
more unknowns than independent equations), or
have a unique solution (in the case of having the same
number of independent equations as unknowns). For
underspecified systems, one seeks to perform more
measurements. For overspecified systems, one has
to use his/her knowledge to remove a less significant
reaction or find a set of solutions that gives rise to
minimal errors. In general, for biological metabolic
reactions, the system is always overspecified.
MFA on a Cellular System

Utility and Limitations
Metabolic flux analysis is the application of material

balances on metabolic systems. The principle and
approach are not different from the example shown
for a chemical reaction system. When MFA is
applied to a cellular system, usually the inputs and
outputs into cells are measured and known. One
needs to enlist all of the reactions involved and set
up material balance equations for all intracellular
components that are considered, and then proceed
to solving the resulting system of equations. In
most cases, the effect of cell expansion and the
effect of dilution due to volume enlargement are
ignored. This is regarded as acceptable because
the timeframe considered in the analysis is usually
much shorter than the doubling time of cells.
MFA provides an additional dimension to analyze

experimental data. Instead of merely examining
various specific rates, MFA provides a bird eyes
view of the distribution of nutrients to different
metabolic pathways. This insight is especially
useful when comparing different metabolic
behaviors under different culture conditions
or growth stages.
By combining MFA with
transcriptome data, which provides a global view of
the changes of transcripts in cells, one can possibly
gain much more insight into the dynamics of cell
metabolism, which is not easily seen otherwise.
However, like material balance in chemical
reactions systems, the accuracy of MFA is limited by
the knowledge of reactions involved in the system
and the accounting of material balance. Further
complicating the analysis is the fact that cellular
metabolism is compartmentalized in organelles
and cytosol. This poses additional constraints
on the balancing of many molecules playing
key roles in inter-organelle communications.
Experimentally, a complete accounting of materials
in a cell culture system is challenging. Material
balance in a reaction system should be balanced on
every elemental species, or at least for the major ones
A Generalized Approach to MFA
Measure Specific Rates

Check Carbon/Nitrogen Balance
Stoichiometric Matrix
Calculate Fluxes J
Visualize Fluxes on Metabolic Map
Examine Calculated Specific Rates
Examine Calculated Carbon/

Nitrogen Balance
Fig. 8.1: A flow chart for metabolic flux analysis
(C, N, H and O). The consumption of major nutrients,

including glucose and all amino acids, can be easily
accomplished. Oxygen consumption can also be
measured. The products, including lactate, some
non-essential amino acids, and the protein product,
can also be quantified. However, a major metabolic
product, carbon dioxide, is not easily measured.
A typical cell culture medium contains a rather

high concentration of bicarbonate, as pH buffer,
making it very difficult to quantify CO2 produced by
cells, unless isotope is used. One way to estimate
carbon dioxide production takes into account the
fact that the respiratory quotient of most cells
under typical culture conditions is about 1.0. By
measuring oxygen consumption rate and assuming
RQ is 1.0, one can obtain the CO2 production rate
and use the value for closing the carbon balance.
Nitrogen balance can be obtained by measuring
the consumption of amino acids and accounting
for the production of cellular protein and product.
However, with measurement errors and uncertainty
in estimating biomass composition, it is common to
see the total nitrogen in inputs and outputs differ by
more than 10%. Still, this is often better than the
extent of closure one often obtains for carbon balance.
Some cell culture media contain complex

components, such as serum or plant hydrolysate.
Those components provide peptides, lipids,
and fatty acids for growth. The extent of their
consumption is hardly measured, making it
even more difficult to close the material balance.
MFA can give much insight into the metabolic

characteristics of cell culture processes, but
one should be extra diligent in the execution of
experiments and collection of measurements,
to ensure the results of MFA are reliable.
General Approach
Selecting Reactions for Analysis
A typical mammalian cell expresses about 15,000

genes at a given time. Only a fraction of the genes
are actually involved in material flow (or fluxes);
others are involved in cellular structure or in signal
flow. MFA primarily concerns the flow of materials.
For all practical purposes, we can account for
only carbon- and nitrogen-containing species.
The flow of ionic species (e.g., sodium, potassium,
phosphate, etc) are not considered in MFA.
Even considering only the flow of carbon- and

nitrogen-containing compounds, the number of
reactions is over a thousand. However, only a
portion of these reactions has a significant flux.
Overall, the glycolysis pathway has the highest
flux for cultured cells. There are only a hundred,
or so, reactions whose flux is >5% of that of
glycolysis. The vast majority of the reactions have
a flux in the order of 1% of glycolysis, or lower.
Fig. 8.2: Simplification of metabolic pathways for metabolic flux

analysis.
Applying MFA to entire cell metabolism

a vastly large number of reactions involving material
flux
most of them have very small fluxxes, much smaller
than errors in closing balance of materials
select the reactions and pathways which are relevant
and sufficiently large
to zoom in on pathways with very small fluxxes, use a
root type tracer to examine fluxxes of local reactions
Considering the errors in carbon and nitrogen

balances, MFA is best applied to analyze the flux of
major arteries in carbon flow. It does not provide
accurate estimates for minor reactions, such as
glycan distribution in the product protein.
To
apply MFA for the reactions with minor fluxes, an
accurate quantification of the specific formation
rate of key reaction intermediates is necessary.
The first step in applying MFA to a cell system

is to reduce whole cell reaction networks to a
manageable and meaningful subset of pathways. In
general, those pathways include glucose and amino
acid metabolism, and the biosynthesis of building
blocks for biomass and product formation. In the
course of simplifying the reaction network, different
assumptions are made and often a different set of
reactions are included in the analysis. The selection
of pathways will impact the flux distribution.
Compartmentalization
Carbohydrate metabolism takes place in cytosol and
mitochondria
lipid metabolism involves even more compartments
intercompartmental traffic of pyruvate and malateaspartate shuttle also involve amino acids and TCA
cycle intermediates
flux balance must be met for the intercompartmental
traffic
Cells are compartmentalized to segregate various

reactions; genome replication and RNA synthesis
occurs in the nucleus and glycolysis takes place in
cytosol. Among the major metabolic reactions, TCA
cycle, oxidative phosphorylation, and fatty acid
oxidation occur in the mitochondria. Across the
boundary of cytosol and mitochondria, pyruvate,
glutamine, and components of the malate-aspartate
shuttle pass at high fluxes. The flux of the malateaspartate shuttle transports the reducing equivalent
of NADH into the mitochondria, at the same level as
pyruvate. The flux of the citrate shuttle generates
acetyl CoA in the cytosol and is typically lower, but
under some growth conditions it can be rather large.
These shuttles and transfers across the

mitochondrial membrane pose further constraints
on material flow. Imposing those constrains on MFA
is important for obtaining a good estimate of fluxes
of energy metabolism. The regulation of fluxes
across the cytosol and the mitochondria may play a
role in shift of metabolism. Therefore, building a flux
analysis model based on two compartments of the
cytosol and mitochondrion is a worthwhile effort.
Biomass Equations
biomass equation is very difficult to construct and is
subject to errors
under fast growing and lactate production conditions,
biomass formation constitutes only a small fraction of
total carbon
With the exception of cell mass, all outputs of a

cell system have a defined chemical composition.
Their production rates are measured and readily
used in MFA. In contrast, the cell concentration
is often measured by the cell number, and
not biomass. Furthermore, the composition
of the cell mass is seldom characterized.
Under most culture conditions, the vast majority of
carbon taken up by cells as nutrients is converted to
lactate and carbon dioxide. Only a small fraction is
actually incorporated into new cell mass. Because
this amount is such a small fraction of the total
carbon consumed, the error in the estimation of
biomass does not significantly affect the MFA results.
However, under an oxidative metabolic state

(characterized by reduced specific glucose
consumption and very little lactate production or
consumption), the amount of carbon and nitrogen
channeled to biomass becomes a significant
Biomass Equation Derived

from Cellular Components
Proteins
DNA/RNA
Glucose
Amino acids
Macromolecule
BIOMASS
Lipids/Carbohydrates, etc.
pg per cell
Protein
DNA/RNA
Lipids/Carbohydrates
Total dry weight
300
15/30
55
400
Elemental composition of cell

C
1
0.2605 1.975
O
0.489
Fig. 8.3: Determining elemental composition for biomass

stoichiometric formula
Solution and Analysis
portion of the overall material flow. Since the

components of the biomass are drawn from
fluxes leading to building blocks, the composition
of cell mass affects the results of flux analysis.
Despite their potential influence on the estimation

of fluxes, the stoichiometric equations of biomass
formation are not often addressed. Based on
elemental analysis of C, N, O, and H of a mouse
hybridoma cell line, a general compositional
formula was given as: CH1.975N0.2605O0.489 . The
general range of cellular composition of lipids,
proteins, nucleic acids, and polysaccharides is also
available. The amino acid composition of cellular
proteins has been reported for a number of cell
lines. Overall, the available data is very limited.
One should be aware of this potential error when
applying the literature to the biomass equation.
A simplified cellular reaction system may consider
only the metabolism of glucose and amino
acids, while simultaneously lumping lipid and
nucleotide synthesis into a biomass formation
equation. The resulting reaction network consists
of about fifty fluxes involving a similar number of
compounds. Such a system can be solved using
software such as Mathmatica and MatLab, using
the least square method to minimize residue.
The solution gives a set of values to unknown fluxes,
as well as to the specific rates whose measured
values are already known. One should compare
the values given by the solution and the measured
values. It is also prudent to check material balance
based on the solution values. If major deviations
are seen, the results need to be reevaluated.
The results of MFA involve tens of variables

depicting the fluxes in the reaction network.
Such data are difficult to comprehend, without a
proper graphic presentation. To gain insight, it is
helpful to link the results of MFA to a metabolic
map for visualization. Such visual images greatly
enhance our ability to discern shifts in metabolism.
An Example
For an example of MFA, a MatLab algorithm is

presented and its solution is included. The reactions
considered in the reaction network are listed in
the accompanying table. These include reactions
for glycolysis, TCA cycle, amino acid degradation
pathways, biomass, and antibody synthesis. The
metabolic network is compartmentalized into the
mitochondria and the cytosol. Reactions in the
malate-aspartate shuttle (also known as the NADH
shuttle) are included to account for the transfer
of the reduction potential of NADH generated in
cytosol into the mitochondria, thereby regenerating
the levels of cytosolic NAD.
Further, three
enzymes, including mitochondrial malic enzyme,
cytosolic malic enzyme, and pyruvate carboxylase,
are also included in the reaction network. The
respiratory quotient is assumed to be 1.0; in other
words, the carbon dioxide production rate is
assumed to be the same as the oxygen uptake rate.
The specific rates determined for two metabolic

states of NS0 cells: one produces lactate in the
exponential growth phase and the other consumes
lactate in late growth stage. These rates are used to
solve the fluxes in each situation. Upon the solution,
the results are displayed in a metabolic chart.
The contrast of the two metabolic states is seen,
not only in the specific rates of glucose and
glutamine, but also in the internal distribution.
Table 1. List of Energy Catabolism Reactions for MFA

#
Reaction
Compartment
Pathway
GLCc + 2NADc 2PYRc + 2NADHc + H2O
Cytosol
Glycolysis
PYRc + NADHc LACc + NADc
Cytosol
Lactate Dehydrogenase Reaction
PYRm + NADm AcCoAm + NADHm + CO2
Mitochondria
Krebs Cycle
OAAm + AcCoAm CITm
Mitochondria
Krebs Cycle
CITm + NADm + H2O AKGm + NADHm + CO2
Mitochondria
Krebs Cycle
AKGm + NADm SUCCoAm + NADHm + CO2
Mitochondria
Krebs Cycle
SUCCoAm + H2O FUMm
Mitochondria
Krebs Cycle
FUMm + H2O MALm
Mitochondria
Krebs Cycle
MALm + NADm OAAm + NADHm
Mitochondria
Krebs Cycle
10
GLNc GLUc + NH3
Cytosol
Glutaminolysis
11
GLUm AKGm + NH3
Mitochondria
Glutaminolysis
12
PYRc + GLUc ALAc + AKGc
Cytosol
Alanine Synthesis
13
SERc PYRc + NH3
Cytosol
14
GLYc SERc
Cytosol
15
CYSm PYRm + NH3
Mitochondria
16
ASNc ASPc + NH3
Cytosol
17
HISm GLUm + NH3
Mitochondria
18
ARGm + AKGm 2GLUm
Mitochondria
19
PROm GLUm
Mitochondria
20
ILEm + AKGm SucCoAm + AcCoAm + GLUm
Mitochondria
21
VALm + AKGm GLUm + CO2 + SucCoAm
Mitochondria
22
METm SucCoAm
Mitochondria
23
THRm SucCoAm + NH3
Mitochondria
24
PHEm TYRm
Mitochondria
25
TYRm + AKGm GLUm + FUMm + 2AcCoAm
Mitochondria
LYSm + 2AKG 2GLUm + 2 CO2 + 2AcCoAm
Mitochondria
27
LEUm + AKGm GLUm + 3AcCoAm
Mitochondria
28
0.0104GLNc + 0.0110ALAc + 0.0050ARGm + 0.0072ASNc + 0.0082ASPc

+ 0.005CYSm + 0.0107GLUc + 0.0145GLYc + 0.0035HISm + 0.0050ILEm +
0.0142LEUm + 0.0145LYSm + 0.0028METm + 0.0072PHEm + 0.0148PROm +
0.0267SERc + 0.0160THRm + 0.0085TYRm + 0.0189VALm CH1.539N0.2645O0.314
Cytosol/Mitochondria
Antibody Synthesis
29
0.208GLCc + 0.0377GLNc + 0.0133ALAc + 0.0070ARGm + 0.0ASNc + 0.0261ASPc

+ 0.0004CYSm + 0.0006GLUc + 0.0165GLYc + 0.0033HISm + 0.0084ILEm +
0.0133LEUm + 0.0101LYSm + 0.0033METm + 0.005PHEm + 0.0081PROm +
0.0099SERc + 0.0080THRm + 0.0040TYRm + 0.0096VALm CH1.975N0.2605O0.489
Biomass Synthesis
30
CITm + MALc CITc + MALm
Fatty Acid Synthesis
31
CITc AcCoAc + OAAc
Cytosol
Fatty Acid Synthesis
32
MALc MALm
Glutaminolysis
33
GLUc GLUm
Glutaminolysis
34
OAAm + GLUm AKGm + ASPm
Mitochondria
Malate Aspartate Shuttle
35
OAAc + NADHc MALc +NADc
Cytosol
36
AKGc + ASPc OAAc + GLUc
Cytosol
37
ASPm + GLUc ASPc + GLUm
38
MALc + AKGm MALm + AKGc
39
MALm + NADm PYRm + CO2 + NADHm
Mitochondria
Malate Decarboxylation (Malic Enzyme)
40
MALc PYRc + CO2
Cytosol
Malate Decarboxylation (Malic Enzyme)
41
2NADHm + O2 2NADm
Mitochondria
42
2FADH2 + O2 2FAD
Mitochondria
43
PYRm + CO2 OAAm
Mitochondria
Pyuvate Caboxylation (Pyruvate Carboxylase)
26
Table 2. Abbreviation of Intermediates of Reaction Modes

Abbreviation
AB
AcCoA
AKG
ALA
ARG
ASN
ASP
BIOMASS
CYS
CO2
FUM
GLC
GLN
GLU
GLY
HIS
ILE
LAC
LEU
LYS
MAL
MET
NH3
OAA
PHE
PRO
PYR
SER
SucCoA
THR
TYR
VAL
O2
NADH
FADH2
FAD
Name
Antibody
Acetyl Coenzyme A
A-Ketoglutarate
Alanine
Arginine
Asparagine
Aspartate
Biomass
Cysteine
Carbon Dioxide
Fumarate
Glucose
Glutamine
Glutamate
Glycine
Histidine
Isoleucine
Lactate
Leucine
Lysine
Malate
Methionine
Ammonia
Oxaloacetate
Phenylalanine
Proline
Pyruvate
Serine
Succinate Coenzyme A
Theonine
Tyrosine
Valine
Oxygen
Nicotinamide Adenine Dinucleotide
(Reduced)
Nicotinamide Adenine Dinucleotide
(Oxidized)
Flavin Adenine Dinucleotide (Reduced)
Flavin Adenine Dinucleotide (Oxidized)
Subscript
c
m
Compartment
Cytosol
Mitochondria
NAD
Concluding Remarks
MFA is an important analytic technique of
quantitative physiology. It can provide insight into
process optimization and metabolic engineering.
A flux balance can be written for each metabolite,
within a cellular or metabolic system, to yield the
dynamic mass balance equations that interconnect
various metabolites.
With the knowledge of
stoichiometry and steady state assumption, one can
obtain the flux estimate for individual reactions.
MFA is a powerful tool, but it is invariably

based on a simplified reaction network. The
assumptions made in simplifying the reaction
network will affect the results of the analysis. It
is best used for a first approximation to obtain
some insights for further exploration of ideas.
Cell Culture Bioreactors

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
Basic Types of Bioreactors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
Stirred Tank (Well Mixed) Vs. Tubular Reactor (Plug Flow) . . . . . . . . . . . . . . . 193
CSTR and PFR with Reaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
Implication When Growth or Reaction Occurs in the Reactor . . . . . . . . . . . . . 195
Heterogeneous Reactor- High Solid Content . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
Operating Mode of Bioreactors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
Batch and Continuous Processes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
Batch Cultures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
Fedbatch Cultures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
Tissue Culture and Disposable Cell Culture Systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
Disposable Culture Systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
Cell Support Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
Suspension Culture vs. Adherent Cultures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
Microcarriers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
Cell Culture Bioreactors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .206
Simple Stirred Tank Bioreactor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
Airlift Bioreactor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
Fluidized Bed Bioreactor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
Membrane Bioreactor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
Multiple Membrane Plate Bioreactor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
Other Bioreactor Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
Microsphere Induced Cell Aggregates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
Agarose Cell Immobilization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
Microencapsulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
Membrane Stirred Tank . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
Spin Filter Stirred Tank. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
Vibromixer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
Concluding Remarks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
CELL CULTURE BIOREACTORS |191
Introduction
In the past decade, industrial mammalian cell

culture has become the production vehicle of choice
in the fast growing field of medical biologics. These
products now extend from therapeutic proteins
and viral vaccines to cells and viruses for cellular
and gene therapy. Before the arrival of recombinant
DNA technology, industrial mammalian cell
culture was used only in the production of viral
vaccines, and was confined to relatively small
scale operations, involving the culture of adherent
cells in tissue culture flasks. The arrival of rDNA
technology led to significant changes. Now, the host
mammalian cells are invariably continuous cell
lines, and the scale of operations has skyrocketed.
The transition from cultivation of adherent cells in

flasks to a truly industrial scale has transformed
bioprocess. The most dramatic changes have been
in the process, rather than in the basic design
of the reactor. In the 1980s and early 1990s,
research and commercial development focused
on exploring possibilities for new bioreactors.
Researchers pushed to overcome a number
of real and perceived hurdles, e.g., the notion
that mechanical agitation and air sparging had
detrimental effects on mammalian cells, the fact
that the culture systems supported only low cell
numbers and resultant low product concentrations,
and that the accumulation of metabolites caused
growth inhibition limiting potential production.
What was unexpected was the extraordinary

capability of cells to adapt to suspension
growth, and to adapt to serum-free, and
even protein-free, conditions. Over time, the
bioreactors used for industrial cell culture
have evolved into a few basic configurations.
This chapter will discuss the fundamentals of
bioreactors and highlight specific bioreactors
developed in the 1980s. These examples are
useful for their conceptual novelty and help to
illustrate the limitations on their use as industrial
bioreactors. Although these reactors are not
commonly used in current processes, they might
find specialized applications in other areas, such

as tissue engineering, or gene and cellular therapy.
Basic Types of Bioreactors
Bioreactors can be generally categorized according

to their mixing characteristics. The two extremes
of mixing are, at one pole complete, instantaneous
mixing and, at the opposite pole, a solid-like
complete absence of mixing. The differences are
best illustrated using a continuous flow reactor,
although many bioreactors are used in batch fashion.
Tubular Bioreactor
Fig. 9.1: Schematic of a tubular reactor
Stirred Tank Bioreactor

Fig. 9.2: Schematic of a well-mixed stirred tank reactor.
Idealized Bioreactor
Plug flow reactor (no mixing)
Well mixed reactor (instantaneously perfectly mixed)
Stirred Tank (Well Mixed) Vs. Tubular

Reactor (Plug Flow)
F
Co
Continuous Stirred Tank Bioreactor

Fig. 9.3: A continuous flow stirred tank bioreactor with a
switch for tracer injection.
The two extreme mixing patterns characterize

two types of idealized continuous reactors:
well-mixed stirred tank and plug-flow (tubular)
reactor. In an ideal well-mixed bioreactor, the
mixing is assumed to be intensive such that the
fluid is homogeneous throughout the reactor.
When a new solution is added to the reactor, the
solute is instantaneously, uniformly distributed.
A tubular reactor is at the other extreme of an
idealized bioreactor; it has absolutely no mixing.
It is also called a plug-flow reactor, or piston-flow
bioreactor. As its name implies, a solution entering
the bioreactor will move downstream like a wall or
a plug, and steadily forward until it reaches the exit.
The distinction between a well-mixed continuous
stirred tank reactor (CSTR) and plug flow reactor
(PFR) is best illustrated by a comparison of their
behavior after a step change in feed concentration.
Consider a continuous reactor that has an inlet
stream (feed) and an outlet stream that are equal in
volumetric flow rate; thus, the volume of the reactor
is constant. Initially both the fluid in the reactor and
the feed stream are colorless. At time=0, the feed
stream is switched to a fluid containing a red dye
at a concentration of C0. We then observe how the
concentration of the dye has changed at the outlet.
In the case that the reactor is well mixed (i.e., a
CSTR) as soon as the feed stream is switched,
the color is distributed uniformly in the reactor.
Because the dye is uniformly distributed, regardless
Fig. 9.4: The concentration profile of the tracer after

injection into a CSTR
Fig. 9.5: A plug-flow reactor with a switch for tracer injection.
Fig. 9.6: Concentration of tracer detected at the end of a PFR

after tracer injection.
Initial condition: t = 0, C = 0
CSTR and PFR with Reaction

The same reaction carried out in PFR or CSTR gives
different kinetic behavior:
In PRF, as the feed moves downstream, the
reactant(s) is consumed and the product(s) is
produced. Different locations in the reactor have
different concentrations.
In CSTR, the concentration of reactant(s) and
product(s) is homogeneous. The concentration at
the exit is the same as in the fluid in the reactor.
of where the effluent stream is drawn from, it will

also acquire the red dye at the same concentration
as in the reactor. Of course, for the purpose of our
discussion, we neglect the time delay caused by
flow in the pipe lead into and out of the reactor.
As more colored feed comes into the reactor, the
concentration of the color seen in the outlet will
gradually increase. The balance equation for the
concentration of the dye that gives the color is
shown, where F is the flow rate, V the reactor volume,
C the dye concentration, and subscript o denotes
the feed. Note that in this case, the concentration
in the outlet flow is the same as in the reactor,
because of the assumption of being well mixed.
By setting initial condition C=0, one can solve the

equation, and plot the profile of the normalized
concentration of the dye (C/Ci) over time. V/F has units
of time and represents the holding time (), the time
it takes one reactor volume of feed to pass through
the reactor. One can see that after a holding time, the
concentration of the dye in the reactor is 0.63 of that
in the feed. It takes three holding times (3) for the
concentration to approach (~98%) that in the feed.
In a PFR, the red colored dye moves downstream

like a sharp band, since there is no backmixing
or diffusion to blur the sharp boundary between
the color and colorless streams. After switching
the feed stream to the red color, it takes one
holding time for the red color to appear at the
exit. When the color appears at the exit, its
concentration is identical to the feed concentration.
The discussion above considers only mixing

behavior in the reactor, essentially just mixing
tanks. In a bioreactor, consumption (of nutrients)
and generation (product formation and growth)
take place; these processes (reaction) will cause
the concentration profile to differ from that in a
simple mixing tank. For CSTR, the assumption that
the concentration at the outlet is identical to that
in the reactor holds. The effect of the reaction is
dealt with by adding a term to the material balance
equation to account for the reaction. One equation
is written for each of the reactants and products.
The equation describing the overall balance of the

dye in a PFR differs from that for a CSTR. In a PFR, the
concentration is no longer constant throughout the
reactor, but is dependent on the position within the
reactor. Because the concentration of both reactants
and products changes along the flow direction in the
reactor, the size (i.e., the length) of the reactor, or the
holding time in the reactor, affects their values. To
account for the effect of position, we do a material
balance on the dye over a very small section along the
reactor. The equation has both time and position as
independent variables. Once a steady state is reached,
the dependence of time can be eliminated, and the
equation becomes an ordinary differential equation
with the position as the only independent variable.
Implication When Growth or Reaction

Occurs in the Reactor
It can be seen that in the presence of reaction,

the concentration of the reactant decreases as
the feed stream moves downstream, while the
product concentration increases. For bioprocess
applications using a PFR, the nutrient level
will decrease while metabolite concentration
increases along the bioreactor. At some position,
the nutrients will be depleted, and a reactor
longer than that length becomes unproductive.
Plug-flow bioreactors are intrinsically more difficult

to scale up than mixing vessels, as the concentration
gradient of essential nutrients, oxygen in particular,
will inevitably become limiting in the downstream
region of the reactor. One way to overcome the
limitation is to increase the nutrient supply rate
by using a higher concentration in the feed, or by
operating at a higher flow rate. There are, however,
practical limits on both nutrient concentration and
flow rate. For example, nutrient concentration is
limited by its solubility, and a high flow rate will
require a higher capacity pump to overcome a higher
pressure drop across the reactor. These kinds of
limitations are especially true for growth of aerobic
organisms, because oxygen solubility in water is
very low and is quickly depleted unless supplied
continuously. Thus, the size and scalability of a PFR
Heterogeneous Reactor- High Solid

Content
Fig. 9.7: Schematic of a heterogeneous reactor
microcarrier
Heterogenous Stirred Tank Bioreactor

Fig. 9.8: A reactor supporting cell growth on microcarriers is a
heterogeneous reactor
Operating Mode of Bioreactors

Batch and Continuous Processes
reactor is rather limited. In a CSTR model, all cells

in the reactor encounter the same environment. The
nutrients that feed into the reactor are distributed
uniformly, albeit at abundant or suboptimal levels.
A bioreactor typically encompasses three phases:

a liquid phase, a solid phase (cells can be regarded
as solid), and a gas phase (bubbles for aeration).
Often, cell mass represents a very small fraction
of the culture volume; thus, the content of a cell
culture bioreactor is often treated as homogeneous,
as if having a single liquid phase, and cells, like
nutrients, are treated as part of the liquid phase;
no special consideration of the volume taken up by
cells in establishing material balance equations .
However, in microbial fermentation and in plant
cell culture, cells may make up a large fraction
(up to 50%) of total reactor volume. The reactor
volume, thus, is partitioned into cell volume and
culture medium volume. The cell and nutrient
concentrations calculated from medium volume
or total reactor volume have different values.
Some so called heterogeneous cell culture

bioreactors contain a significant volume fraction
of solid phase such as microcarrier beads. In a
microcarrier bioreactor for example, microcarrier
beads often constitute 10-30% of the culture
volume. In this circumstance, even cell concentration
needs to be well defined. For example, it must be
specified whether 107 cells per milliliter is with
respect to total culture volume or to liquid volume.
The operation of a bioreactor is generally classified as

batch or continuous mode. In a continuous process,
the feed is continuously being introduced to the
reactor, and product stream is continuously being
withdrawn from the reactor. In a batch process, the
medium, including all nutrients (except oxygen), is
added at the beginning, and no other nutrient addition
occurs until the end of cultivation. Continuous
culture is not commonly practiced in the cultivation
of mammalian cells unless in conjunction with cell

recycle. This topic will be discussed in a later chapter.
Batch cultures are often limited by the fact that

in many cases the product concentration is
not sufficiently high. To achieve a high product
concentration it is necessary to have a high cell
concentration; for that the nutrient concentration
in the medium must also be high. In many cases
high concentrations of nutrients are not attainable
in batch culture because the initial concentration
at the start of culture is limited by the solubility of
some nutrients, or by the concentration at which
cells can be grown without growth inhibition. It
may be necessary to add additional substrate or
product precursor to sustained continued growth
to reach higher cell and product concentrations. In
other cases, such as intermittent fed-batch cultures,
instead of terminating a batch process at the end of
culture, a portion of the culture is kept, and fresh
medium is added to start the process again.
Batch Cultures
Fig. 9.9: Kinetics of growth and product formation in a batch

culture
Batch processes are simple and are widely used.

In fact, batch process culture is used much more
frequently than the fed-batch process. Before
reaching production scale, cell expansion is carried
out in a series of batch cultures with increasing
reactor volumes. This series of batch cultures
for cell expansion is often referred to as seed
train. During expansion in batch cultivations, the
culture is discontinued in the reactor and cells
are transferred to the next larger reactor while
still rapidly growing. If a culture is allowed to run
its course, cell growth will be inhibited by the
accumulation of metabolites, and the resulting
product concentration will be low. In general, batch
processes do not result in high productivity. Batch
cultures are used in viral vaccine production, or,
if used in recombinant protein manufacturing,
are largely limited to the cell expansion stage.
Fedbatch Cultures
Fed-batch processes do not differ significantly

from batch cultures. The simplest form of fedbatch culture involves intermittent harvest. At
a late exponential growth stage of the culture, a
portion of the cells and product are harvested, and
the culture is replenished with fresh medium. This
avoids metabolite inhibition of cell growth and
replenishes nutrients for continued cell growth.
Intermittent Harvest
Fig. 9.10: Kinetics of growth and product formation in a fedbatch culture with intermittent harvest
Fedbatch
Fig. 9.11: Kinetics of growth and product formation in a fedbatch culture with increasing culture volume
This process may be repeated several times.

This simple strategy is commonplace for the
production of viral vaccines produced by persistent
infection, as it allows for an extended production
period. It is also used in roller bottle processes
with adherent cells. By intermittent harvesting,
turn-around time involved in cleaning up and
restarting the process is saved. Furthermore, after
replenishing the medium, the culture reaches
peak cell concentration faster than if the culture
is started de novo with fresh inoculum, which
usually is at a significantly lower cell concentration.
For production of recombinant proteins and

antibodies, a more traditional fed-batch process
is typically used. A fed-batch culture is started
in a volume much lower than the full capacity of
the bioreactor (approximately 40% to 50% of
the maximum volume, or at a level sufficient to
allow the impeller to be submerged). Nutrients,
usually in a more concentrated form than in basal
medium, are added during the cultivation to allow
cell and product concentrations to reach much
higher levels than can be attained in batch culture.
Tissue Culture and Disposable Cell Culture Systems

In this section we will describe reactors used for
large-scale cultivation of mammalian cells. Arguably,
the oldest bioreactors are animals themselves or
their tissues. The application of virus vaccines dates
back two hundred years, when Edward Jenner used
cow pox to inoculate humans for protection against
small pox. Many viral vaccines developed in the
first half of the 20th century employed animal or
animal tissues as production vehicles. Use of animal
tissues (e.g., chicken eggs, rodent brains) for virus
production is still practiced today for some viruses,
although cell culture is considered the norm.
Disposable Culture Systems

Single Use Bioreactor
Simple, low-capital investment, fast to implement,
ease of equipment validation
A variety of systems with vastly different mixing
mechanisms
Less well-characterized in fluid dynamics, mass
transfer
Suitable for seed culture, scales up to hundreds of
liters
Possible faster turn around time
Lacking the strength of steel; some operations
routinely done on steel vessels, such as pressurized
liquid transfer, cannot be performed on single use
vessels
Roller Bottles
Many different types of disposable, single-use

plastic bioreactors including T-flasks, roller
bottles, and multiple flat panels are commonly
used for production of recombinant proteins and
viral vaccines, using both attachment-dependent
and suspension cells. The plastic surface may be
treated to facilitate attachment of adherent cells.
Suspension cells can be readily grown in a stirred
vessel. However, there is an increasing trend toward
the use of disposable (single-use) bioreactors at
small and moderate scales. Factors such as ease of
operation, shorter time to scale up, lower capital
investment, and reduction of process validation
burden have contributed to their wide adoption.
Roller bottles are cylindrical screw-capped bottles
with a total volume ranging from 1 to 1.5 liters, and
are suitable for a culture volume of 0.1 to 0.3 liters.
Stacks of bottles are placed on a rack and rotated at
1 to 4 rpm. For small-scale operations, roller bottles
provide many advantages for the cultivation of
adherent cells. The system is relatively inexpensive
to set up, and allows for rapid adjustment of
production throughput in response to changing
needs. Furthermore, complete replacement of
medium, e.g., from growth medium to product
medium, is rather straightforward. Roller bottles
are particularly useful in the case that the serumcontaining medium needs to be replaced by a serumfree or protein-free medium for the production
phase. The transparent glass or plastic wall allows
visual or microscopic examination of the culture
status. Microbial contaminated bottles can be readily
identified and discarded before mixing with others.
For large-scale production of biologics, however,

there are numerous drawbacks to roller bottles.
On-line environmental monitoring and control is
virtually impossible or at least impractical. Aseptic
bottle handling for inoculation, protease-treatment
for cell detachment and expansion, medium
exchange, and product harvest are labor intensive
and must be performed by a skilled technicians
to ensure a low failure rate. A batch of a size
suitable for manufacturing purposes may require
hundreds or even thousands of bottles, and the
large number of manual steps involved dramatically
increases the risk of microbial contamination.
Fig. 9.12: A roller bottle and roller bottles on racks for cell
culture
Despite these significant drawbacks, roller bottles

are widely used in the production of recombinant
proteins and viral vaccines. Often, they are used
primarily because the product involved has already
been approved by regulatory agencies and a process
change would incur significant costs for regulatory
approval and could even result in product
comparability issues. In some cases, roller bottles
were selected because the process was easy to
implement and the production capacity needed was
manageable. However, when faced with the need to
increase production, e.g., due to market expansion,
scale-up of a roller bottle process is challenging.
Notable examples of industrial roller bottle processes

include the production of erythropoietin (EPO)
using recombinant Chinese Hamster Ovary cells, the
production of live attenuated chickenpox (varicella)
vaccine, and herpes zoster (shingles) vaccine using
the adherent secondary human lung fibroblast cell
strain, MRC-5. In some manufacturing facilities a fully
automatic, robotic roller bottle handling system is
available. However, a robotic system cannot handle
a large number of bottles in parallel. Handling
a large number of roller bottles is a prolonged

process. For example, in a process entailing virus
inoculation or a culture condition switch to induce
product formation, inoculation and induction
time can vary very widely from the processing of
the first bottle to the processing of the last bottle.
Multiple Plate System
Nunc Cell Factories (NCFs) are widely used in

laboratories and in industrial production of viral
vectors, vaccines, and recombinant proteins.
Although designed to culture adherent cells,
NCFs can also be used for suspension cells. A 40tray NCF has a capacity of 25,280 cm2 and ~8 L
liquid volume. For large-scale manufacturing, a
mechanical handling system can be used. One
notable example is the production of multivalent
Rotavirus vaccine using Vero cells, a continuous
African green monkey kidney cell line. The vaccine
is a live, attenuated virus vaccine that contains
five different human-bovine virus reassortants,
each produced from a different process.
Fig. 9.13: A commercial multiple plate system for cell culture
Bags, Cylinders on Shakers, Movers
Another multiple plate system (CellCube Module)

entails nine-inch square polystyrene plates stacked
vertically with 1 mm spacing in a resin case. The
space between stacks is completely filled with
medium. Continuous medium circulation is used
for oxygen and nutrient supply. CellCube has
been used in the cultivation of MRC-5 cells for the
production of attenuated hepatitis A virus (HAV)
and for the production of VAQTA, an inactivated
HAV vaccine. These disposable bioreactors have
also been somewhat popular for intermediate levels
of production of cells and gene therapy vectors.
Blood bags have long been used for the cultivation

of cells for cell therapy. Small bags are placed in
incubators for temperature control. For larger single
use bag systems, mechanical mechanisms provide
mixing and oxygen transfer. Bags are especially
suitable for cell expansion before reaching final
production scale up. WaveTM consists of presterilized, disposable Cellbags and a special rocking
platform. The rocking motion of the platform
induces waves in the culture fluid to provide mixing

and oxygen/gas transfer.
Fig. 9.14: A single-use bag cell cultivation system
Cell Support Systems

Suspension Culture vs. Adherent
Cultures
In manufacturing settings, validating the equipment

for restart of a process contributes a significant
part of the overall operating costs. A more extreme
case of single use alternatives to a conventional
fermentor is a retrofit product to mimic an existing
stainless steel bioreactor vessel. It consists of a
reusable stainless steel outer support container and
a single use cell chamber with a working volume of
up to 1000 liters, which can be integrated into the
existing bioreactor control system. These modest
scale, single use reactors are not intended to replace
a conventional stirred tank made of stainless steel
with fixed piping, auxiliary equipment, etc. Because
of its single use construction, its operation is limited.
For example, a single use fermentor cannot be
pressurized to quickly transfer its contents to other
process units, or even while receiving inoculum.
Many cell lines used in the production of
vaccines and biologics grow in suspension cells.
Disposable systems are invariably limited in
their scale of operation. For processes whose
production scale is relatively small, a disposable
system provides some advantages. For protein
therapeutics needed in large quantities, the use
of disposable systems may be limited to inoculum
preparation. Large scale operations, thus, continue
to employ fermentors or other bioreactors.
The majority of cells used for protein production via

rDNA are suspension cells, which, as suggested by
the name, have no need for attachment to a surface
and are simply suspended in the medium in a
bioreactor. For growth of adherent cells, cell supports
are needed to provide adherent surface. The cell
support is placed in a bioreactor which provides a
mechanism to supply nutrients and oxygen to cells. In
some cases, cell support is also used for suspension
cells, allowing cells to be kept in the reactor while
the medium is being perfused or exchanged.
Microcarriers
Most normal diploid cell strains or primary cells are

anchorage-dependent. For large scale operations in a
stirred tank bioreactor, microcarriers are used to provide
adherent surfaces. Conventional solid or microporous
microcarriers are suitable for normal diploid cells
which require attachment and spreading. Macroporous
microcarriers also allow cells to grow within the interior,
and are used for a wide variety of continuous cell lines.
Solid and Microporous Microcarriers
The use of microcarriers for cell culture was first

demonstrated by Van Wezel (1967). The basic
concept is to allow cells to attach to the surface
of small suspended beads so that conventional
stirred tank bioreactors can be used for cell
cultivation. To facilitate suspension of these cellladen microcarriers, their diameter and density are
usually in the range of 100300 m and 1.021.05
g/cm3, respectively. This diameter range provides
good growth surface area per reactor volume. Even
at a moderate microcarrier concentration occupying
only 815% of the culture volume, a significantly
larger surface area per reactor volume can be
achieved than that possible with roller bottles.
Human fibroplast FS4 on cytodex 1 microcarriers
CHO on cytodex 3 microcarriers

Fig. 9.15: Cell cultivated on microcarriers
Most anchorage-dependent cells do not develop their

normal morphology, and some do not multiply well
on surfaces with an excessively high curvature. These
cells do not grow well on small microcarriers. On the
other hand, some cells, especially transformed cells,
multiply well even on small microcarriers. When
these cells are grown on very small microcarriers,
they agglomerate to form aggregates and continue
to grow to high density. The small microcarriers,
usually with a diameter of about 50 m, serve as
nuclei for the initiation of aggregate formation.
Microcarriers can be made of many different
materials including dextran, gelatin, polystyrene,
glass, and cellulose, not all of which are commercially
available. In general, microcarriers have a wettable
surface, and are sometimes coated with collagen or
other adhesion molecules to enhance cell adhesion
and spreading. The most widely used microcarriers,
which are based on dextran, are derivatized
Table 1. Desired Characteristics of Microcarriers

Density
~1.02 g / cm3
Only slightly higher than

water for easy suspension
and setting
Diameter
150 - 200 m
Should be the smallest

possible and yet allow for
cell spreading
Porosity
From solid to
nearly 90%
Prefer solid, for use as

inoculum bead to bead
transfer
Surface
Properties
ECM coating,
slightly
positively
charged
Positive charge enhances

initial attachment
Macroporous Microcarriers
Fig. 9.16: Scanning electron micrograph of a collagen based

macroporous microcarrier bead
Cofocal microscopic optical section of

HepG2 cells in collagen macroporous
microcarriers
Fig. 9.17: Cofocal optical section of cells grown in a

macroporous microcarrier
with charged molecules or denatured collagen.
An advantage of microcarrier culture is the ease

of separating cells from culture medium. Many
microcarrier cultures require medium exchanges
during cultivation to remove lactate, ammonia,
and other metabolites and to replenish nutrients.
This can be accomplished by simply allowing cellladen microcarriers to settle so that the spent
medium can be withdrawn and replenished.
In large-scale operations, continuous or semicontinuous perfusion is more frequently used.
This can be accomplished by withdrawing medium
through a coarse screen that retains microcarriers
in the reactor but allows medium to pass.
A wide variety of cell types have been grown on

microcarriers including fibroblasts, epithelial cells,
hepatocytes, neuroblastoma cells, and endothelial
cells from various species. Overall, microcarrier
culture is a convenient laboratory and research tool,
and has the advantage of being amenable to large-scale
production if a large quantity of product is needed.
Macroporous microcarriers contain large internal

pores. The void space inside allows cells to be
cultivated on internal and external surfaces. Cells
in the interior are less susceptible to mechanical
damage caused by agitation and gas sparging. On the
downside, cells in the interior of microcarriers are
more likely to be subject to oxygen limitation due to
the long diffusional distance, especially since most
macroporous microcarriers have a larger diameter
(500 m2 mm) than standard microcarriers.
Macroporous microcarriers are made of various

materials including gelatin, collagen, and plastic.
Many cell lines have been successfully grown on
macroporous microcarriers including Vero, HepG2,
CHO, and 293 cells. The final cell concentration
achieved tends to be higher than that obtained
with an equivalent volumetric concentration
of conventional microcarriers. In some cases,
however, the growth kinetics are slower because the
penetration of cells into the interior may be slowed
or even retarded by the restrictive opening of the
pores.
Table 2. Listing of Microcarriers

Different Types of Microcarriers
Dextran
(GE Healthcare)
Cross-linked dextran matrix with various surface modifications
Cytodex 1
Positively charged DEAE group
Cytodex 3
Denatured type I collagen coating
Polystyrene
(SoloHill Engineering)
Cross-linked polystyrene core material with a wide variety of

surface modifications
Plastic
No surface modification, or positively charged
ProNectin F
Recombinant fibronectin coating
FACT III
Type I porcine collagen coating
Glass
High silica glass coating
Hillex
Surface modified with cationic trimethyl ammonium
Cellulose
Microgranular DEAE-cellulose microcarrier
DE-52
Anion exchange capacity of 1 meq / g dry materials
DE-53
Anion exchange capacity of 2 meq / g dry materials
Alginate
(Hamilton)
Global Eukaryotic Microcarrier (GEM), composed of alginate

matrix with modification of surface
Glass (Biosilon)
Hollow glass
Macroporous Microcarriers
Gelatin
(Percell Biolytica)
CultiSpher-G and CultiSpher-S, cross-linked gelatin matrix with

pore size 10-20 m, CultiSpher-S has a higher thermal and
mechanical stability due to a different cross-linking procedure
Cellulose
(GE Healthcare)
Cytopore, cross-linked cellulose matrix with a pore size

averaging 30 m. Its surface is modified by the introduction of
DEAE group
Collagen
(MP Biomedicals)
Cellagen, prepared from bovine corium insoluble collagen by

pepsin treatment
Polyethylene
(GE Healthcare)
Cytoline, composed of high-density polyethylene weighted with

silica
Polypropylene
(New Brunswick)
Fibra-Cel disk, composed of polyester mesh with polypropylene

support
Cell Aggregates
Fig. 9.18: Confocal section of HEK293 cell aggregates.

Green: viable cells, red: dead cell nuclei

Simple Stirred Tank Bioreactor
Some transformed cells can grow as aggregates

when cultivated in shaker flasks or in stirred
vessels. Different methods have been used to induce
aggregate formation. Aggregation can be promoted
by manipulating the calcium concentration in
conjunction with the agitation rate. Aggregate
cultures have advantages similar to microcarrier
cultures. They can be cultivated in conventional
stirred tank reactors with environmental
control. They can be allowed to settle relatively
rapidly by stopping agitation, permitting
easy medium replenishment or perfusion.
Stirred tanks, or conventional fermentors, have been

widely used for culturing suspension cells since the
1960s. By using microcarriers, adherent cells can also
be cultured in a stirred tank. The basic configuration
of stirred tank bioreactors for mammalian cell
culture is similar to that of microbial fermentors.
A major difference is that the aspect ratio (ratio of
height to diameter) is usually smaller in mammalian
cell culture bioreactors. The power input per
unit volume of bioreactor is also substantially
lower in mammalian cell culture bioreactors.
While the Rushton type impeller is the norm in

microbial fermentors, mammalian cell culture
fermentors mostly employ axial flow type impellers.
The difference reflects the different purposes of
agitation in microbial fermentation and in cell
culture. In microbial fermentation, agitation is
needed at a higher power input to disperse air
bubbles and to increase oxygen transfer efficiency,
whereas in mammalian cell culture, the primary
purpose of agitation is to maintain relatively
uniform suspension of cells or microcarriers.
In general, the mixing time in a mammalian cell
culture bioreactor is substantially longer than
that in a microbial fermentor of similar scale. The
oxygen transfer capacity in a cell culture bioreactor
is also substantially lower than that in a microbial
fermentor. However, the typical oxygen demand in a

mammalian cell culture is 10 to 50 times lower than
that in microbial fermentation.
Airlift Bioreactor
A variant of the bubble column reactor with internal

circulation loops is used for the cultivation of
mammalian, insect, and plant cells. In these airlift
reactors, internal liquid circulation is achieved
by sparging through the internal draft tube. The
fluid in the draft tube has a lower effective density
than the bubble-free section, which, along with
the upward momentum generated by air flow,
induces liquid circulation. Medium flows upward
through the sparged section (riser) and downward
in the bubble-free section (downcomer). This
method of generating circulation has a low energy
requirement compared with stirred-tank reactors.
Air
Fig. 9.19: Schematic of an air-lift bioreactor
Fluidized Bed Bioreactor
Airlift bioreactors for cell cultivation are considered to

be low-shear devices because there is no mechanical
agitation. They have been used successfully
with suspension cultures of BHK 21, human
lymphoblastoid, CHO, hybridomas, and insect cells.
Fluidized bed reactors have long been used in

chemical catalysis. The fluid stream (often gas
in catalysis) enters through a flow distributor
at the bottom at high velocity to suspend the
solid catalyst particles. The reactants enter the
catalyst and products diffuse out into the fluid
to be carried out through the top of the reactor
where a separator prevents any particles from
being blown out. The main advantage of a fluidized
bed is the high heat transfer efficiency between
the high velocity fluid and the catalyst surface.
When applied to cells or microcarriers directly, the
density difference between solid phase and liquid
phase is small, making particle retention difficult.
In the 1980s, collagen macroporous beads were
used in commercial fluidized beds offered by Verax.
The carriers were weighted by inclusion of iron
particles to increase the density difference between
fluid and carrier so that the particle could be
retained in the reactor at the flow rates required for
supplying sufficient oxygen to support cell growth.
Membrane Bioreactor
The use of hollow fiber reactors for cultivation of
mammalian cells dates back to the early 1970s. A
hollow fiber system can be used for both anchoragedependent and suspension cells. It consists of a
bundle of capillary fibers sealed inside a cylindrical
tube. The basic configuration is rather similar to
the hollow fiber cartridge used in kidney dialysis.
Hollow Fiber Bioreactor
The hollow-fiber entails a porous polymeric

layer that provides mechanical support covered
by a thin membrane which provides selectivity
based on the size of molecules. In most cases, an
ultrafiltration membrane is used. The molecular
weight cut-off (MWCO) of the membrane differs
according to the specific application, ranging from
a few thousand to a hundred thousand daltons.
Fig. 9.20: Schematic of a hollow-fiber bioreactor
Multiple Membrane Plate Bioreactor
The culture media is pumped through the fiber

lumen, and cells grow either in the extracapillary
space, or on the shell side. Supply of low-molecular
weight nutrients to the cells and the removal of waste
products occur by diffusive transport across the
membrane between the lumen and the shell spaces.
The ultrafiltration membrane prevents free diffusion
of secreted product molecules from passing through
the membrane and allows them to accumulate in
the extracapillary space to a high concentration.
Although the use of microfiltration hollow fiber

membranes for cell culture is infrequent, it does
appear in various research applications for studying
metabolism and for producing small quantities of
materials for research or diagnostic applications.
Scaling up of a hollow-fiber system is limited by

the ability to extend the axial length of the fiber
due to oxygen transfer limitation. Use of a very
high flow rate to supply more oxygen is limited
by the capacity of the pump and the mechanical
strength of the membrane. Expanding the cartridge
diameter to increase the capacity eventually faces
the problem of uneven flow distribution among the
fibers. In principle, one can mix two different fibers
to supply oxygen and medium separately. However,
mixing different hollow fibers in a cartridge poses
Other Bioreactor Systems

Microsphere Induced Cell Aggregates
(a) CHO cells attaching to microspheres and agglomeration of

beads.
(b) Sections of HEK 293 cells
forming microspheres induced
aggregates.
(c) SEM micrograph of HEK 293
cells on microspheres
a major challenge in manufacturing. An alternative

configuration using multiple flat membranes has
been attempted. However, this seemingly versatile
reactor also suffers from practical manufacturing
complexity, and is not used in large scale operations.
Some cells do not form aggregates readily. When

suspended in culture the rate of aggregate formation
is slow and cells lose viability over time. One way to
induce aggregate formation is to add microspheres
to cell a suspension to promote agglomeration to the
microspheres. These agglomerated microspheres
become aggregates as cells grow. If the aggregate
diameters become too large, necrotic centers
can formed as a result of transport limitations.
Many cell lines including BHK, CHO, 293, and
swine testicular cells have been cultivated as
aggregates with sizes ranging between 90 and
400 m without the formation of necrotic centers.
Figure 9.21: Microspheres induced cell aggregates
Agarose Cell Immobilization
Agarose entrapment of cells is usually accomplished

by passing a cell/agarose suspension through a
small orifice into an air parallel jet stream. The
droplets of agarose containing entrapped cells
are collected in a chilled oil phase to allow the
agarose to gel. Alternatively, the cell-agarose
suspension may be allowed to drop onto the center
of a fast rotating disk. The centrifugal spinning
force causes droplets to form and be dispensed
outside the disk. Agarose beads tend to be rather
large, commonly, hundreds of micrometers in
diameter. The large size of agarose beads limits
oxygen transfer at high cell concentrations. In
addition, agarose beads lack sufficient mechanical
strength to sustain mechanical optimization even in
a moderately small scale (tens of liters) bioreactor.
Microencapsulation
Another method of cell entrapment entails entrapping

cells in a polymeric matrix to form spheres followed
by coating with a polymeric film to control diffusivity
of molecules. The idea is to allow fast nutrient
diffusion while retaining the product molecules.
One of the polymers most often used for cell

entrapment is calcium alginate. Cells are suspended
in sodium alginate and added dropwise into
a calcium chloride solution, which allows for
gelation. The alginate beads may be coated with
polylysine for increased mechanical and chemical
stability. The alginate gel inside the polylysine
coated bead can be liquefied through treatment
with a calcium chelator such as EDTA or citrate.
Fig. 9.22: Formation of microencapsulated cell beads
Membrane Stirred Tank
Large-scale
application
using
this
microencapsulation technique is not easy.
However,
the
microcapsule
provides
a
means of immunoisolation of transplanted
cells or tissues, and could be suitable for
some
tissue
engineering
applications.
The membrane stirred tank was developed by
Professor Jrgen Lehmann in the 1980s. This
bioreactor uses long microporous polypropylene
tubing wrapped around rotating rods. By adjusting
the air pressure in the polypropylene tubing, the
micropore expands to allow gas to be in direct
contact with medium, thereby providing bubble-free
aeration. The rotation of the tubing also provides
gentle agitation to microcarriers or suspended cells.
Fig. 9.23: A membrane stirred tank bioreactor
Spin Filter Stirred Tank
The centerpiece of a rotating wire cage bioreactor

is the wire cage, which is often mounted on the
agitation shaft. The bottom of the cage is solid,
while the side is made up of wire mesh with
openings ranging from 25 to 60 m. The average
diameter of cells is approximately 10 to 15 m.
Typically, fresh medium is added continuously
outside the draft tube and the culture fluid is
withdrawn from inside the cage at the same rate.
Under certain operating conditions, the cell

concentration inside the wire cage is lower than that
outside the cage, thus achieving cell retention in the
bioreactor. Under optimal conditions, the device
does not act as a filter, and no cake is formed. The
retention of the cells does not appear to be due to
centrifugal force exerted by the rotating motion of the
cage, because the terminal velocity of the cells due to
centrifugal force is two to three orders of magnitude
lower than that of the fluid velocity across the screen
due to perfusion. The electrostatic effect is also
unlikely to be responsible, since the ionic strength of
the culture fluid is relatively high, and the thickness
of the Debye layer is only in the order of 1 nm.
The rotating wire cage bioreactor has also been used
in aggregate and microcarrier cultures. In these
cases, the mechanism of cell retention is relatively
straight forward as the particle is much larger than
the openings on the cage.
Fig. 9.24: A spinner filter stirred tank bioreactor
The mechanism of retaining cells in suspension by

this device is still not clear. It is plausible that the
retention is caused by a fluid mechanical effect,
maybe one similar to the behavior of particles of
low Reynolds number near the wall in a Poiseuille
flow or laminar boundary layer flow along a flat
plate. Lacking a mechanistic understanding of cell
retention has made scale up of this device difficult.
Many recently installed spin filters are operated
at very high rotation rates. At such a high sped of
rotation rate centrifugal force plays a key role in cell
separation by dispelling cells away from the surface
of the rotating cage. Cell separation in these reactors
is certainly different from the low speed wire cage
reactors described in this section.
Vibromixer
Fig. 9.25: A vibro-mixer based bioreactor
high frequency
vibration
A vibromixer uses a perforated disk as the mixing

mechanism rather than a conventional impeller. The
disk vibrates in a vertical direction (perpendicular
to the plane of the disc) at high frequency causing
liquid to circulate through the perforated holes
and provide mixing. The vibromixer was used in
the 1960s for the cultivation of suspension cells
and virus production. Its use in cell cultivation has
diminished in recent decades. However, it is still used
in some cases to keep concentrated microcarriers
in suspension for cell detachment during the
trypsinization step. The fluid mixing provides gentle
shear to detach trypsinized cells from microcarriers.
liquid moved by the

vibrating plate as going
through the holes
Fig. 9.26: Movement of culture liquid around the vibromixing plate
Concluding Remarks
Mammalian cells are widely used in the production
of viral vaccines and therapeutic proteins. Most of
those processes employ cells growing in suspension;
however, some applications, particularly in vaccine
production, use adherent cells. Stirred tank
bioreactors are most widely used for mammalian cell
culture processes. When a stirred tank bioreactor
is used for growing adherent cells, a supporting
surface for cell attachment is provided through the
use of microcarriers or macroporous microcarriers.
Alternatively, adherent cells may be grown as
aggregates in suspension. Fed-batch culture is
commonly practiced since it facilitates reaching
high cell density and high product concentrations.
When a stirred tank bioreactor is operated in
a continuous mode, it is a common practice to
employ a cell retention device for perfusion in
order to increase cell and product concentrations.
As mammalian cell culture processes are increasingly

being used in manufacturing, one also witnesses an
increasing adoption of disposable (or single-use)
cell culture devices. Traditional roller bottles are still
often seen. Additionally, multiple flat plates or parallel
trays, along with bag type cell growth chambers and
even plastic stirred tanks, are used in moderate scale
production and in the seed culture propagation for
large scale bioreactors. These disposable devices
offer simplicity in operations, and ease process
validation in the production of biopharmaceuticals.
Some innovative bioreactors developed nearly two
decades ago are now finding new applications in
tissue engineering and cell therapy. As these new
technologies advance and the demand of those
specialized cells grows we will continue to see
innovative developments in bioreactor technology.
Oxygen Transfer in
Oxygen Transfer Through Gas-Liquid Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Oxygen and Carbon Dioxide Concentration in Medium . . . . . . . . . . . . . . . . . .
Oxygen Consumption and CO2 Production . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Method of Supplying Oxygen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Gas Sparging in Cell Culture Bioreactor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Bubble Size and Sparger Orifice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Experimental Measurement of KLa and OUR . . . . . . . . . . . . . . . . . . . . . . . . . . .
Effect of Hydrostatic Pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Damage to Cells by Gas Sparging. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Mass Transfer Resistance in Cell Immobilization Reactor . . . . . . . . . . . . . . . . . . . . . . .
Oxygen Transfer in Plug-Flow Bioreactors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
213
217
219
220
221
222
224
225
226
229
229
231
Oxygen Transfer Through Gas-Liquid Interface

In cell cultivation, oxygen is transferred from air
through the gas phase (via either gas bubbles or
the gas space in the flask) and the liquid phase
interface into the medium where it is consumed by
cells. Simultaneously, CO2 produced by the cells is
excreted into the medium and then diffuses into the
gas phase. Transfer of oxygen or CO2 molecules from
one phase to the other occurs only when oxygen
or CO2 in the two phases is out of equilibrium. If
oxygen in the two phases are in equilibrium (i.e.,
the medium is already saturated with oxygen),
there will not be any net transfer in either direction.
Fig. 10.1: Concentrations and partial pressure of oxygen across

an air bubble in culture fluid
If a liquid deprived of any gas molecule is in contact

with air composed of 79% N2 and 21% O2, both N2
and O2 will diffuse into the liquid and eventually
reach equilibrium between the two phases. (Of note,
because N2 has a very low solubility and is hardly
OXYGEN TRANSFER IN CELL CULTURE BIOREACTORS |213
Description of Dissolved Oxygen Concentration

Solubility in water at 37C with air of 1 atm, 21% O2
0.2 mmole / L
0.64 mg / L
Since air in equilibrium contains 159.6 mm Hg or 0.21
atm, sometimes solubility is expressed also in mm Hg,
especially for blood oxygen level
To transfer oxygen or CO2 between gas and liquid phases

through interface

The two phases must not be in equilibrium
consumed by cells, it will not be further considered

in this chapter.) At equilibrium, if the oxygen content
in the gas phase is increased to 40%, then oxygen
in the liquid phase will no longer be saturated
and, gradually, oxygen will diffuse into the liquid
phase until the dissolved oxygen concentration
becomes equilibrated in the gas phase. Conversely
if the oxygen level in the gas phase is decreased to
10%, then oxygen will escape from the liquid into
the gas phase until a new equilibrium is reached.
Thus, in order for mass to transfer between the two

phases, a deviation from equilibrium is essential. Such
a deviation is often referred to as the driving force
for mass transfer. Quantifying the driving force, or
the degree of deviation from equilibrium is the first
step toward calculating the rate of mass transfer.
Oxygen level in the gas phase is usually described

as mole percentage or mole fraction (yO2). It can
also be expressed as partial pressure (PO2), or
the mole fraction of oxygen multiplied by total
pressure. If the ambient pressure is 1 atm (or
760 mm Hg) and the oxygen mole fraction is
0.21 (yO2 = 0.21), then the partial pressure of
oxygen in the air is 0.21 atm (or 159.6 mm Hg).
Fig. 10.2: Driving force for oxygen transfer across air-liquid

interface
The oxygen concentration in a liquid phase (or

dissolved oxygen concentration) is often expressed
as mmole/L. When pure water is saturated with
oxygen under a pressure of 1 atm and 21% O2 at 37C,
its oxygen concentration is 2.2 mmole/L. Because
this concentration is the equilibrium value with an
oxygen partial pressure of 159.6 mm Hg, sometimes
it is expressed as 159.6 mm Hg. Expressing dissolved
oxygen concentration in terms of its gas phase partial
pressure is common in the medical profession.
Since oxygen is a critical nutrient in the living
world, one needs to be familiar with the multiple
descriptors used to describe its concentration.
For example, water which is at 50% of saturation

with air at 1 atm and at 37C has a dissolved oxygen
concentrationof0.11mmole/L, or 79.8mmHg.Ifwater
containing 0.11 mmole / L oxygen is in equilibrium
with a N2/O2 mixture, then that mixture would
have 79.8 mm Hg [or 10.5% (mol/mol)] of oxygen.
If water containing 0.11 mmol/L of oxygen is

put in contact with air (with 21% O2 at 1 atm at
37C), it is undersaturated and is 0.11 mmol/L
away from saturation. In other words, the water
could hold 0.11 mmol more oxygen per liter.
One could also say the water is 79.8 (159.6 79.8) mm Hg below oxygen saturation. Such
concentration difference from saturation is the
driving force for oxygen transfer. One can then
quantify the driving force using different units;
describing it in terms of liquid phase concentration
(mM), or gas phase concentration (mm Hg).
A concentration profile of oxygen across an interface

of gas phase and liquid phase is depicted in the
figure. In the bulk liquid (i.e., the liquid at some
distance away from the interface), the concentration
of oxygen is C. In the bulk gas phase, the oxygen
fraction is yO2, and its partial pressure is PO2.
To dissect the mechanism of oxygen transfer, we

model the interface as having a boundary layer or
an imaginary film separating the two sides. Let us
denote the oxygen concentration in the liquid as C* if
the system is at an equilibrium with gas phase (whose
oxygen partial pressure is PO2), and denote the gas
phase concentration as P*O2 when it is in equilibrium
with liquid phase (which is at a concentration of
C). The liquid phase system is therefore C*-C away
from equilibrium. In other words, the driving
force for the liquid phase to reach equilibrium is
C*-C. Recall that we can also use gas phase units
to describe the liquid phase concentration, so the
driving force can also be described as PO2-P*O2. The
two different descriptions of the same driving force
are related by a constant as will be seen again later.
Again the driving force across the boundary of the
two phases can be expressed in either gas phase
concentration or liquid phase concentration. Thus,
when the driving force is not zero, oxygen will diffuse
across the boundary layer to move the system toward
equilibrium. The higher the concentration gradient
across the interface is (i.e., a stronger driving
force), the faster the oxygen transfer rate will be.
The rate of oxygen transfer is dependent on three
Oxygen Transfer Rate (OTR) = KLa (C*-C) =KLa C

[mmole / l hr] = [cm / hr] [1 / cm] [mmole / L]
C*: solubility of oxygen,
C: oxygen concentration gradient (i.e., the driving force)
Three Factors Affecting OTR

Mass transfer coefficient (KL)
Specific transfer area (i.e., the interfacial area)
Driving force (i.e., the gradient across interface)
To Improve Oxygen Transfer

Increase KL (make the interface more turbulent)
Increase a (use smaller bubbles for sparging, or use
silicone tubing)
Increase C (use oxygen enriched air, or dont maintain
dissolved oxygen (C) at an unnecessarily high level.)
factors: 1) the magnitude of the difference between

C* and CL, 2) the area that is available for oxygen
transfer, and 3) the mass transfer coefficient.
The second factor, available surface area, is typically

expressed as interfacial area per volume of liquid.
It, thus, has a unit of inverse of length (L-1, like cm1). Think of water evaporation, or mass transfer
of water vapor, from a liquid water surface into
dry air. Therefore, if a given amount of water were
contained in a cup, it would take longer for it to
evaporate than if that same amount of water was
spread out in a flat dish because the surface area
is much larger in a dish. Another example of using
a large surface to increase oxygen transfer is our
lungs, which, when fully expanded, can yield an area
of nearly 150 m2. This is why using a lung to transfer
oxygen is far more effective than using our skin.
The third factor affecting the rate of oxygen

transference is the mass transfer coefficient.
This is a measurement of the resistance for mass
transfer at the interface. It is affected by the
specific molecules that are being transferred
and the physical and chemical properties of the
liquid. The mass transfer coefficient (KL) has
the same units as velocity (L/t, like cm/sec).
One can imagine the boundary layers in the interface

as a stagnant film surrounded by a bulk liquid,
which is under some mechanism of localized mixing.
The oxygen molecules have to diffuse across the
boundary layer to reach other phases. If the system
is under vigorous mixing, the boundary layer would
be thinner and the transfer would be faster. Again
considering water evaporation in a dish, if that
dish were being stirred and the air is being blown
across the dish, the mass transfer coefficient at
the interface would be higher and the evaporation
would be faster than when the dish is unperturbed.
Oxygen and Carbon Dioxide

Concentration in Medium
Calculation From Henrys Law
Henrys Law
xA =
PA
H
xA: mole solute A / mole

solution
PA: partial pressure of A
CA: mole A / L
H: Henrys law constant
Table 1. Henrys Law Constants for O2
(atm/ mole O2/mole H2O)
T, C
10-4 x H
5
2.91
20
4.01
25
4.38
30
4.75
35
5.07
37
5.18
Example: at 37 C, in 1 atm air (PO2 = 0.21 atm),

O2 concentration in H2O is:
(Eq. 1)
C* = (0.21atm) / (5.18x104atm / moleO2 /
moleH2O) x (55.5 mole H2O / L H2O)
= 0.22 mmole O2 / L H2O
In ambient air, (oxygen partial pressure = 0.21
atm), C* ~ 0.22 mmole / L in H20 at 37 C = 160
mmHg = 5.8 mg / L (5.8 ppm)
Oxygen solubility is not affected by other

dissolved species in medium. Its solubility is
virtually identical to PBS and water.
Because the rate of oxygen transference between a

gas phase and a liquid phase is dependent on how
far it is from equilibrium, it is important to first
determine its saturation level, i.e., concentration
at equilibrium, to accurately estimate the oxygen
transfer rate. Oxygen is sparingly soluble in water.
For such gas-liquid pairs, the solubility of a gas
solute in the liquid phase is described by Henrys
Law. Henrys Law states that, at equilibrium, the
concentration of a gas in the liquid phase (i.e., its
solubility) is proportional to the pressure of the gas.
The concentration in each phase is described in
different ways. The most commonly-used variations
are partial pressure for gas phase (which is the
mole fraction of the gas solute multiplied by the
total pressure) and the mole fraction for the liquid
phase. For process calculations, one may find it
convenient to express liquid phase concentration
in mmole/L.
In the medical profession, the
gas phase composition is often expressed in
mm Hg, with 760 mm Hg = 1 atm. The value of
Henrys law constant for a gas-liquid pair is, thus,
dependent on how the concentration is expressed.
Henrys law applies well to an ideal solution. Despite

all of its components, cell culture medium is still close
to an ideal solution so the saturation concentration
calculated using an aqueous solution applies to
cell culture media. However, cell culture media is
frequently used under a gas phase containing 1 - 10%
CO2. Also, in a bioreactor, the air pressure is usually
greater than 1 atm. Thus, oxygen solubility needs to
be corrected for the differences in pressure and CO2.
Solubility in medium should be adjusted by CO2 used

in gas phase (i.e., the fraction of O2 in gas phase is
usually not 0.21).
6CO2 (g) @ = H $ 6CO2 (aq) @

CCO2 =
The solubility of CO2 can similarly be calculated

using Henrys law. However, in an aqueous
solution, CO2 associates with water molecules
and then dissociates into HCO3- and H+. The value
calculated using Henrys law is the sum of CO2(aq)
and H2CO3 in pure water. Thus, the solubility of
CO2 will be affected by the pH of the solution.
(Eq. 2)
PCO2
H
(Eq. 3)
Table 2. Henrys Law Constants for CO2

(atm / mole CO2 / mole H2O)
T, C
10
15
20
25
30
35
40
10-3 x H 0.728 0.876 1.04 1.22 1.42 1.64 1.80 2.09 2.33
Example: Using 10% CO2 in air, CO2 concentration at

37C is
Solubility of CO2 affected by pH
CO2 (g) ? CO2 (aq)
(Eq. 4)
CO2 (aq) + H2 O ? H2 CO3 (aq)

H2 CO3 ? HCO3- + H+
Oxygen Consumption and CO2

Production
Optimal Oxygen Concentration
Most cells grow well at a dissolved oxygen level of 30%
of saturation with air
Oxygen Demand by Cells
On the average q02 1.0 x 10-10 mmole / cell-hr
CO2 Production by Cells
Respiratory quotient of cells is about 1.0, i.e., 1 mole
CO2 is produced for every mole of O2 consumed, or qCO2
1.0 x 10-10 mmole / cell-hr
A high CO2 concentration is inhibitory
Inhibitory level usually starts at 15% to 20% CO2
Stripping of CO2 by Aeration
CO2 is required for cell growth (e.g. in fatty acid
synthesis), in addition to providing pH buffer in vivo.
Cells can grow well even in sodium-bicarbonate-free
medium, especially in a stationary culture, because
CO2 produced by cells accumulates in medium to
condition their growth.
When cell concentration is low in a sodiumbicarbonate-free medium, excessive aeration may strip
off CO2 to inhibit cell growth, unless some CO2 (~0.5
1%) is in the gas.
The objective in oxygen transfer is to supply oxygen

at a rate that is enough to maintain the dissolved
oxygen at a set level (or optimal level) for cell
growth and production. For most mammalian cells,
a 30% saturation with ambient air is sufficient for
optimal growth. Given that the saturation level
at 1 atm of ambient air is about 0.2 mmole/L, the
optimal value for cell growth is about 0.06 mmole/L.
The amount of oxygen cells consume per culture
volume per unit of time is called the oxygen uptake
rate (OUR). As you may recall from the kinetics
chapter, OUR is dependent on the specific oxygen
consumption rate (qO2) and the cell concentration.
Most oxygen taken up by cells is used to oxidize

carbon sources. The most important carbon
sources for mammalian cells in culture are glucose
and glutamine. The oxidation reaction of both
compounds yields CO2 at a stoichiometric ratio
of 1.0 and 0.9, respectively. This CO2 to O2 ratio
is called the respiratory quotient (RQ). The CO2
produced by cells will need to be removed from
the culture to avoid excessive accumulation, which
would cause a drop in pH or an inhibition of growth.
In cell culture operation, aeration accomplishes two
goals: supply of oxygen and stripping carbon dioxide
out of culture medium. CO2, inhibitory at high
concentrations, is also an essential component of the
medium. A certain level of gaseous CO2 is necessary
to maintain the pH in a neutral range. CO2 is also
used in a number of biosynthetic reactions, notably
fatty acid synthesis. If CO2 is completely stripped
from the medium, cell growth may be inhibited.
Method of Supplying Oxygen

Method of Supplying Oxygen
Surface aeration
Silicon tubing/membrane
Sparging
Balance of Oxygen in a Reactor
dC
= OTR V OUR V
dt
= K L a V (C * C2 ) qO 2 x V
V
OTR: Oxygen Transfer Rate

OUR: Oxygen Uptake Rate
V: Culture volume
KL is about 3 cm / hr 6 cm / hr for surface aeration

in cell culture vessel
Example: What cell concentration can be reached in
a 400 ml spinner flask (diameter 8 cm)?
Estimate: KL = 5 cm / hr
Surface area A= (4 cm)2 = 50.2 cm2
a = 50.2 cm2 / 500 cm3 = 0.1 cm-1 (a is surface area per
culture volume)
Assume D.O. is almost zero at maximum cell
concentration (maximum driving force)
C = 0.18 mmole / l
qO2 = 1.0 x 10-10 mmole / cell hr
qO2 x = OTR = 5 cm / hr 0.1 / cm 0.18 mmole / l =
0.09 mmole / -hr
x = 0.09 mmole / l-hr 1 x 10-10 mmole / cell hr
= 9 x 108 / l = 9 x 105 / ml
To supply oxygen, one can introduce it through

a gas phase in contact with culture medium. In a
small-scale reactor or flask, a gas head space is
often sufficient in supplying oxygen. In a large-scale
reactor, air or oxygen-enriched air is introduced
as gas bubbles from the bottom. In some cases,
direct bubbling of gas phase into culture medium
is to be avoided to minimize foaming or other
adverse effects. Under such conditions, the gas
phase may be introduced through a membrane
with high oxygen permeability, such as silicone or
Teflon membranes. In other cases, the gas phase
is not introduced into the reactor but rather the
medium is pumped out and passed through an
oxygenator. The oxygen-enriched medium is then
recirculated back to the reactor. A hollow fiber
bioreactor is typically operated in this fashion.
To control oxygen transfer, one can manipulate any
of the three factors affecting oxygen transfer. One
way is to increase the driving force by enriching the
air with more oxygen. If surface aeration is used,
one can install a surface aerator to increase the mass
transfer coefficient on the liquid surface. If tubing or
a membrane is used, one can use a larger membrane
or more tubing to increase the surface area.
In small-scale reactor operations, direct sparging

(i.e., bubbling air) is more prone to problems. The
agitation rate in a cell culture reactor is comparatively
low compared to a microbial reactor. A microbial
fermenter employs turbine impellers at a very high
agitation rate to make gas bubbles entrenched in
the liquid. The bubbles take a torturous path to rise
and eventually burst from the liquid surface. The
holding time of a bubble in the liquid is relatively
long. In comparison, in a cell culture bioreactor, the
bubbles rise in a relatively straight, vertical direction
and emerge from the liquid phase quickly. Therefore,
the efficiency of oxygen transfer for a given air flow
rate in a bioreactor is low. But, as the scale increases
in cell culture, the holding time also increases with
the length of the traveling path of the bubbles,
thus increasing the efficiency of oxygen transfer.
Gas Sparging in Cell Culture Bioreactor

In a cell culture reactor, the most practical means of
supplying oxygen is by blowing air into the reactor
through a sparger. If one were to supply oxygen
through the medium flow, the medium would have
to be recirculated many times per hour because
it can only carry a small amount of oxygen. Thus,
air sparging is commonly used in bioreactors.
Fig. 10.3: Supply of oxygen through liquid circulation
v dc = F (Cin - C) - qo xv
dt
2
Example:
Medium flow rate required to supply oxygen without
gas phase aeration:-
qO $ x = 5 # 10- 10 mmole cell $ hr $ 2 # 10 9 cell l

= 1 mmole l $ hr
assume Cin = 0.2 mmole l (saturated by air)
F V = 5hr- 1 ` need to recirculate medium
5 times every hour
2
Sparging
Short bubble residence time in small bioreactors
Increasing the agitation rate increases oxygen transfer
through better surface aeration and longer bubble
residence time.
Direct sparging in microcarrier culture is possible but
more difficult
Can sparge in the "microcarrier free" zone (use a wire
cage)
Direct sparging in microcarrier culture poses a

number of challenges. Cells attached to microcarriers
render the microcarrier prone to stick to air bubbles.
As the bubbles rise to the liquid surface, carriers
may also rise as well. The associated fluid shear
caused by bubble rising may also cause cell damage.
If a foam layer is allowed to accumulate, one may
see microcarriers accumulate on top of the foam.
The use of antifoam may alleviate the problem,
or one may use a screen/sieve separator to allow
bubbles to be sparged in a microcarrier free zone.
To enhance the oxygen transfer rate in a bioreactor,

one may make changes to manipulate each of
the three factors affecting oxygen transfer rate
(OTR): KL, a, and C. Increasing the airflow rate
will increase the interfacial area, a. Increasing
oxygen concentration increases the driving force,
C. Increasing the agitation power input increases
KL and enhances bubble breakup to increase the
interfacial area, a. Smaller bubbles give a larger
interfacial area at a given air flow rate. In microbial
fermentation, a high agitation power input is used
to break up air bubbles. The agitation power input
for mammalian cell culture is relatively low. The
range of airflow rate used in cell culture is also
low, compared to microbial culture. A high air
sparge rate causes excessive foam formation, but an
extensive use of antifoam agent causes cell damage.
Therefore, enriching the oxygen content in the air
supply and using smaller bubbles are commonly
practiced, especially in small-scale cultures.
Bubble Size and Sparger Orifice

Effect of Orifice Diameter on Bubble Size
6d s
d p = 0
g
1/ 3
d0= orifice diameter

s = surface tension
= difference in density between
liquid and gas
Dispersions of gases in liquids are widely employed

in fermentation processes. The interfacial area
can be very large: 3-mm spherical bubbles with a
gas holdup of 26% (i.e., a liter culture containing
26% volume of gas bubbles and 74% medium)
provide 52cm2/cm3 a very large value.
Bubbles behave very much like oil drops, but their
buoyancy and rising velocity are greater. Mass
transfer within the bubble is rapid, because the
molecular diffusivity in gases is large. Normally
the resistance to mass transfer on the liquid
side of the interface is the dominant resistance.
In the case that a single bubble is slowly released

from a submerged orifice or small capillary, if the
gas flow is very slow, the bubble is released when
the buoyant force just overcomes the surface
tension, at which time the bubble diameter is
governed by the balance between buoyancy
and surface tension exerted on the bubble.
The force balance between buoyancy and surface

tension indicates a modest decrease of diameter,
dp, with an increase in the orifice diameter,
do. Inertial effects become dominant at larger
flow rates employed in industrial multi-orifice
spargers, thus rendering the initial bubble
size almost independent of the orifice size.
Using orifice size to control bubbles is thus not very
effective. In small-scale bioreactors, one may employ
spargers with a fine opening to generate very small
bubbles. The fine bubbles agglomerate as they rise
to the liquid surface and coalesce into large bubbles.
Thus, in large-scale reactors (on the order of
hundreds of liters), the effectiveness of fine bubbles
in an increasing interfacial area begins to diminish.
Bubble size may also affect bubble rising velocity

and the mass transfer coefficient. In principle,
larger bubbles have a higher terminal rising
velocity and a higher KL if bubbles are considered
to be rigid spheres. However, as bubble diameter
increases, the bubble tends to deform and become
flatter on top, thus decreasing the rise velocity.
Experimentally, it has been observed that both

rising velocity and KL only increase with bubble
diameter when the bubble size is small (<3
mm). Beyond that, they are relatively invariant.
Overall, manipulating bubble size to be very fine

is beneficial in small reactors as it increases the
gas transfer area for both oxygen and carbon
dioxide. This is important because the impeller
and agitation rate used is not effective for
breaking up gas bubbles. On a larger scale, though,
controlling bubble size becomes difficult because
of the high gas flow rate and bubble coalescence.
Experimental Measurement of KLa and

OUR
dC = KL a (C* - C)
dt
Initial Condition
t=0, C=CO
After Integration
Plot
ln (C* - C) = KL a $ t
ln(C* C) vs. t
Mass transfer coefficient can be determined

experimentally. Overall, KL is relatively insensitive
to bubble size, except for very small bubbles and
under simple flow patterns. The interfacial area, a,
is difficult to estimate or to measure in bioreactors.
As a result, most experimental measurements of
mass transfer characteristics lump KL and a together
and report the value as KLa. KLa is sometimes
called the volumetric mass transfer coefficient.
To measure KLa, one can aerate a reactor that initially

has a very low oxygen concentration. The dissolved
oxygen concentration will steadily increase until
reaching saturation. A semilog plot of C*-C vs. time
will give the slope as KLa. One can also perform the
experiment by stripping oxygen from medium that
was, initially, at a higher dissolved oxygen level. In
this case, the slope in the semilog plot will be -KLa.
In most operations, the aeration rate will change
over time. One can develop a correlation between
KLa and airflow rate for use during cultivation.
Fig. 10.4: Measurement of KLa by degassing, or stripping of

oxygen from the liquid phase
Measure OUR in laboratory
OUR or the specific oxygen consumption rate

can be measured by filling up a container with
a cell suspension without leaving room for a
gas phase. Since there is no oxygen supply, the
change in dissolved oxygen level will be only
from cell consumption. The slope of the linearly
decreasing dissolved oxygen level will yield the
oxygen uptake rate. Then, the specific oxygen
consumption rate can be obtained by dividing OUR
by the cell concentration in the cell suspension.
Figure 10.5: Measurement of oxygen uptake rate
Effect of Hydrostatic Pressure

Effect of Hydrostatic Pressure
10 m liquid height gives a 1 atm hydrostatic pressure at
the bottom of tank
Up to almost 9 atm has no adverse effect on hybridoma
cell growth and specific glucose consumption.
In very large bioreactors, circulating cells will

occasionally reach a region where the hydrostatic
pressure is high. This is particularly a concern
for air-lift bioreactors, in which the aspect
ratio (height over width) is much larger than
stirred-tank reactors.
Experimentally, it has
been shown that cells grow normally at up to
nine atm and exhibit little discernible metabolic
differences, compared to normal conditions.
Damage to Cells by Gas Sparging
Damage to cells by gas sparging

Cells and cell-laden microcarriers adhere to gas
bubbles
Energy dissipation in gas sparging
It is generally thought that bubble breakup at
liquid surface dissipates most energy and is most
damaging to cells
Protective agents are added to medium in sparged
(see media section)
In an aerated mixing vessel, forces exerted on

cells may cause cellular deformation in several
ways. Forces normal to the cell surface may
squeeze the cell, cells may impinge on moving
mechanical objects, and/or cells may be subjected
to a fluid shear field, such that one side of the cell
is subjected to a fast flower than the other side.
Despite the stress, cells have a substantial
capability to change their shape. The shape change
involves reorganizing the cellular cytoskeleton
and redistributing the lipid bilayer membrane
(both the cytoplasmic and organelles). If the rate
of deformation is fast and the extent is extensive,
the cytoskeletal network and the membrane
integrity will be compromised; therefore leading
to stressed conditions, or even cell death.
Vigorous sparging of air bubbles through

the culture can also cause cell damage. The
mechanism likely to cause most severe damage
is the fluid shear field caused by bubbles.
Several possible events may dissipate the

high intensity of mechanical energy when
bubbles traverse through cell suspension:
1. Release of the bubble from the orifice of the

sparger. As the bubble escapes from orifice, a
volume suddenly becomes void, causing the
cell suspension to rush in to fill the void space.
A velocity gradient, and thus a shear field, will
inevitably occur under such flow conditions.
Fig. 10.6: Deformation of a cell by compressing force
2. A rising bubble will cause velocity gradient

around itself and thereby change the liquid
velocity in proximity to the surface of the
bubble. Conversely, far from the bubble, the
liquid velocity is the same as bulk velocity.
3. Similarly, bubbles coalesce or breakup, also causing liquid velocity gradient.
4. Bubbles bursting from the liquid surface cause the

volume originally occupied by gas to be rapidly
replaced by liquid, thereby causing a shear field.
Cell
fast
moving
object
Liquid Liquid
Velocity difference
two sides of on
a celltwo
causes
shearof
stress
Fig. 10.7:
velocityondifference
sides
a cell causes
sheer stress
It is generally recognized that the last event

generates the greatest shear stress. As the bubbles
burst off the liquid surface, the high liquid velocity
at the interface with the air decreases to the bulk
liquid velocity over a very small distance (of about a
couple of hundred micron). The momentum of the
liquid movement caused by the bursting bubbles
causes the liquid to rise above the liquid level
surface. As a result, the liquid moving downward
to fill the empty space left by the bubble reverses
its direction to rise upward. In the upward moving
liquid cone, the velocity is highest in the center and
lowest on exterior. Also in the process of bubble
bursting, there is a rapid change of the gradient
direction when the liquid overshoots the surface
and then drops down after reaching a maximum
height. The damaging effect on cells entrapped
in the bubble-bursting zone can be tremendous.
Fig. 10.8: Possible ways of damaging cells by sparging in a

bioreactor
As
visualized
through
high-speed
video
photography, a bubble is often surrounded by
a layer of cells, possibly due to hydrophobic
interactions between the cell and bubble surfaces.
Once bubbles begin to rise, the liquid stream forces
cells to the tail, or the wake, of the moving bubble.
As the bubble bursts from the liquid surface, cells
can be severely damaged, if not outright killed.
Block copolymer pluronic F68 reduces the number
of cells adhering to the rising bubble through its
surface tension modulation effect. Pluronic F68 thus
protects cells from the damaging effect of sparging.
Fig. 10.9: Conceptual depiction of fluid movement surrounding

the point at which a bubble emerges from the liquid surface.
Fig. 10.10: Accumulation of cells to a bubble surface due to

hydrophobic interactions and accumulation to the tail end while the
bubble rises.
Mass Transfer Resistance in Cell Immobilization Reactor

Cells are sometimes cultivated inside particles
or in a stagnant liquid phase. Notable examples
are macroporous microcarrier and hollow fiber
systems. In these systems, oxygen not only needs
to be supplied continuously to the culture medium,
but it also needs to diffuse to the position where
oxygen is supplied to the interior, where cells reside.
As oxygen diffuses, it is also consumed by cells.
An oxygen concentration gradient is, thus, created
between the position of oxygen delivery and the
deepest point of cells where oxygen must reach.
Figure 10.11: Concentration gradient of oxygen in a cell

aggregate or an immobilized cell matrix particle
The rate of concentration decrease is dependent

on how fast the cells consume oxygen and the
geometry of the particle. Farther into the interior,
oxygen may become depleted or become too low
to support a high rate of growth or productivity.
To avoid such oxygen limitations, one might resort

to using smaller particles or to maintain a higher
oxygen concentration in the culture medium. The
extent of oxygen transfer limitation can be estimated
by comparing the mass diffusion and consumption
rates. The theoretical calculation is presented
as plots of two variables, often referred to as
effectiveness factor and Thieles modulus, which are
discussed in standard chemical reactor textbooks.
Oxygen Transfer in Plug-Flow Bioreactors

Oxygen transfer limitation occurs in both axial and
radial directions
Periodically reversing the direction of flow should
help
Some bioreactors used in cell cultivation employ

a cell compartment that lacks direct contact with
the gas phase. Hollow fiber bioreactors and some
recirculatory flat bed bioreactors, especially in
tissue engineering applications, fall into this
category. In such a system, oxygen is supplied
by recirculating the medium. Often, an external
oxygenator, such as a hollow fiber membrane device
or a mixing vessel, is used to replenish oxygen in the
medium before it is recirculated to the cell chamber.
In such cases, the oxygen concentration gradient

exists in two dimensions. As the medium flows
downstream in the cell chamber, oxygen is diffused
through the membrane into the cell compartment.

Oxygen concentration, thus, decreases along
the axial direction.
In the radical direction,
oxygen diffuses toward the cell chamber; along
the path the oxygen concentration decreases.
Fig. 10.12: Oxygen concentration gradient in the axial direction

of fluid flow in a plug flow reactor
In a hollow fiber bioreactor, oxygen in the

recirculating medium diffuses from the bulk liquid
to the wall of the fiber, then through the fiber
and into the cell chamber. Typically, the diffusion
resistance is highest across the wall. Outside, the
wall diffusion occurs simultaneously with oxygen
consumption by cells. At a high cell concentration,
the drop in oxygen in the cell chamber is steep
so, often, a very high concentration of oxygen is
maintained in the medium to ensure the center
of the cell chambers receive enough oxygen. To
improve oxygen transfer, one may increase the flow
rate of medium recirculation, introduce a gas phase
compartment (and have hollow fibers for air flow),
or reverse the flow direction periodically. In any case
the longitudinal distance that can be used before
oxygen transfer becomes limiting is restricted.
Fig. 10.13: Photograph of a hollow fiber bioreactor
Fig. 10.14: Oxygen gradient in the radial direction of a fiber in a

hollow fiber bioreactor
Concluding Remarks
The solubility of oxygen is very low in cell culture
medium. It must be supplied continuously to sustain
cellular demand from a gas phase. The transfer rate
of oxygen through the liquid-gas interface is affected
by mass transfer coefficient, interfacial area and
driving force, or the deviation from equilibrium. All
three factors can be manipulated to enhance oxygen
transfer. The most direct and effective method of
supplying oxygen is by bubbling air through the

bioreactor. Although air bubbles cause damage to
cells, the problem can be overcome. However, the
air supply rate per reactor volume tends to decrease
as scale increases. Therefore oxygen transfer and
carbon dioxide removal are among the key issues that
should be addressed in scaling-up or scaling-down.
This will be further discussed in the scale-up chapter.
Fedbatch Culture
and Dynamic Nutrient Feeding
With contributions from Weichang Zhou
Fedbatch Culture. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Types of Fedbatch Culture. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Intermittent Harvest and Feed. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Extended Fedbatch Culture: Fortified Feed and Addition. . . . . . . . . . . . . . . . . .
Fedbatch With Metabolic State Manipulation. . . . . . . . . . . . . . . . . . . . . . . . . . . .
Designing Feed Medium for Fedbatch Cultures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Feed Medium Design for Consumed Nutrients: Stoichiometric Principle. . . .
Feed Medium Design for Unconsumed Components. . . . . . . . . . . . . . . . . . . . .
Control Strategies for Fedbatch Cultures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Control Objective and Criteria: Productivity and Product Quality. . . . . . . . . .
Feeding Strategies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Feeding by Direct Measurement of Nutrient Consumption. . . . . . . . . . . . . . . .
Proportional Feeding With Base Addition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Proportional Feeding With Turbidity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Proportional Feeding With Oxygen Uptake Rate (OUR) . . . . . . . . . . . . . . . . . .
Delivery of Feed Medium. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Online Estimation of Stoichiometric Feeding. . . . . . . . . . . . . . . . . . . . . . . . . . . .
233
234
234
235
236
237
238
240
241
241
243
243
244
244
245
245
246
247
Fedbatch Culture
Batch Culture
Limited nutrient concentration in medium
Solubility of amino acids
High osmolarity at high nutrient levels
Growth inhibition by high nutrient levels
Feed nutrients throughout the cultivation extending
the growth and production period
Higher cell and product concentration
In the first decade of the 21st century, we witnessed

a rapid expansion of therapeutic proteins in
clinical applications. We also saw productivity
increase rapidly through intense process
development. During this time, fedbatch culture
processes have emerged as the predominant
mode for producing recombinant proteins.
In a batch process, the constraint of osmolarity
limits the amount of nutrients that can be added
initially. This low-nutrient level prevents the culture
from attaining high cell and product concentrations.
In fedbatch cultures, medium is added during
cultivation to prevent nutrient depletion, thus
prolonging the growth phase and ultimately
increasing cell and product concentrations.
FED-BATCH |233
Types of Fedbatch Culture

Fedbatch Culture
Predominant mode of production rDNA proteins
Intermittent harvest used in inoculum
preparation, also in virus production
Intermittent Harvest and Feed
Fig. 11.1: Time profile of a fedbatch culture with intermittent

harvest
Fedbatch processes are widely used in multipurpose, multi-product facilities because of their
simplicity, scalability, and flexibility. A variety of
fedbatch operations, ranging from very simple
to highly complex and automated, are seen in
current production facilities.
Compared to a
typical batch operation, the operation time of all
types of fedbatch cultures are extended, resulting
in higher total cell and product concentrations.
The intermittent harvest and feed fedbatch
process does not deviate significantly from a batch
culture. The culture is initiated using a medium and
culture conditions, very similar to that used in batch
culture. Cells are then allowed to grow exponentially,
with no additional manipulation, until cells reach
the late growth phase. A portion of the cells and
product are then harvested and fresh medium is
added to replenish the culture volume. This process
is repeated several times. This simple strategy is
commonly used for the production of viruses which
cause persistent infection in cells, as it allows for an
extended production period. It is also used in roller
bottle processes with adherent cells. In some cases
this type of fedbatch culture is also used to supply
inoculum for imitating a reactor production chain.
FED BATCH |234
Extended Fedbatch Culture: Fortified

Feed and Addition
feeding
initial
concentrated
feed
nutrients
final
Fig. 11.2: Depiction of a fedbatch culture with incremental

feeding
Extended Fedbatch Culture
Fig. 11.3: Time profile of a extended fedbatch culture with

increasing culture volume
In fortified feed and addition type of fedbatch

culture, the period of rapid cell growth is prolonged
by adding concentrated feed media (usually a 1015 times more concentrated than basal medium,
often without salts or with 1x concentrations of
salts). The concentrated medium is added either
continuously (as shown) or intermittently to supply
additional nutrients, thereby further increasing
cell concentrations and lengthening the production
phase. To accommodate the extra addition of media,
a fedbatch culture is initiated at a volume less than
the full capacity of the bioreactor. This initial volume,
typically enough to allow the impeller to be submerged,
is kept at a low level to allow for the addition of
extra media throughout the expansion phase.
The addition of extra nutrients results in a higher

cell concentration, prolongs the production
period, and, ultimately, results in a substantially
higher product titer. In recent years, it has become
common practice to decrease the operating
temperature after a period of rapid growth. The
temperature shift reduces the growth rate and
nutrient consumption rates, further prolonging the
production period and increasing the product titer.
Typical Fedbatch Culture for rDNA Protein Production

7 - 15 days
8x106-2x107 cells / mL
FED BATCH |235
In batch cultures and most fedbatch processes,

lactate, ammonium, and other metabolites eventually
accumulate in the culture supernatant over time.
These metabolites, along with high osmolarity
and reactive oxygen species, are growth inhibitory
and contribute to the eventual loss of viability and
productivity of the culture. The effects of lactate and
ammonia on cultured cells are complex. Detectable
changes in growth, productivity, and metabolism
have all been documented. Additionally, metabolite
accumulation has been found to affect product quality.
Fedbatch With Metabolic State

Manipulation
Figure 11.4: Growth profile of a fedbatch culture with

controlled glucose level to induce metabolic shift
Fig. 11.5: Stoichiometric ratio change during a fedbatch culture

with metabolic shift
Table 1 Characteristic Stoichiometric Ratios of Key

Nutrients for Cell Growing in Different Metabolic States
stoichiometric
ratio (mmole/
mmole)
Without
Metabolic
Shift
Metabolic
Shift
Lactate
consuming
cells
lactate/glucose
1.4 - 2.2
0.05 - 0.5
0.4 - 1.0
ammonia/
glutamine
0.5 - 1.3
0.1 - 0.3
alanine/
glutamine
0.2 - 1.3
0.01 - 1.3
1.0
1.0 - 2.0
oxygen/glucose
By limiting the availability of glucose and glutamine

using controlled feeding strategies, cell metabolism
can be directed to a more efficient state, characterized
by the reduced production of lactate. Such a change
in cell metabolism is often referred to as metabolic
shift. By extending the methodology to limit the
consumption of both glucose and glutamine, both
lactate and ammonium accumulation can be reduced.
In the example of fedbatch culture shown, glucose

concentration was kept at a low level by continuous
feeding of a concentrated glucose solution. In this
scenario, lactate was still produced and, after a period,
lactate production ceased, thus indicating a metabolic
shift. The shift can also be seen from the decrease
in the stoichiometric ratios of lactate production
to glucose consumption during the cultivation.
A metabolic shift alters the stoichiometric

ratio of lactate to glucose throughout the
entire culture period. Even after averaging
the initial culture period (which includes no
metabolic shift), less lactate and ammonium
are produced for a given nutrient consumed.
Cells can also consume lactate under slow growth and
high lactate and low pH conditions. Recently, it has been
shown that after the rapid-growth stage in fedbatch
cultures, cells can switch to lactate consumption
from lactate production. The consumption of
lactate alleviates growth inhibition and gives rise
FED BATCH |236
to a higher product concentration, compared to the

cultures that remain at a lactate production state.
Designing Feed Medium for Fedbatch Cultures

Medium Components
Consumed by cell (almost all organics, growth factors)
Concentration decreases unless being replenished
Not consumed
Or consumption rate too small to cause
concentration change in medium
Fedbatch Culture
Volume can be expaniding due to feeding
Volume expansion may cause dilution
The vast majority of fedbatch culture employs

fortified nutrient mixtures to prolong growth and
production phases. Critical to a successful fedbatch
process is the design of feed media. A well-designed
feed medium should enable cell growth and
product formation, without depleting any medium
component or allowing the build-up of excessive
nutrients or metabolites. Furthermore, adding feed
media to the culture should not introduce factors that
adversely affect cell growth or product formation.
The chemical species in media fall into two

general categories: 1) those whose consumption
rates are significant and measurable and 2) those
whose concentrations greatly exceed the amount
consumed by cells during the cultivation period.
Chemical species that are consumed should be
replenished by the feeding medium to maintain
their concentration in an optimal range. Conversely,
nutrients that are hardly consumed may not be
added; however, because the culture volume
increases substantially in a fedbatch culture,
even a nutrient that is not consumed by cells
may be diluted due to culture volume expansion.
Consequently, feeding of nutrients not consumed by
cells is necessary to sustain growth and production.
FED BATCH |237
Feed Medium Design for Consumed

Nutrients: Stoichiometric Principle
Glucose, amino acids can be added proportional to
their stoichiometric ratio during fedbatch culture
Determine the stoichiometric ratio of medium
components
Choose one reference component (e.g. glucose,
lactate, glutamine, oxygen), feed all other
nutrients by stoichiometric ratio
Stoichiometric ratio may change over time of
cultivation
During the growth phase, the goals of feeding

are largely to replenish the nutrients which have
been consumed and to ensure that nutrient
levels stay within an optimal range for optimal
growth or for optimal product formation.
Under balanced growth conditions, the specific

consumption rates of various nutrients are relatively
constant and, correspondingly, the consumption
of nutrients or production of metabolites relative
to one another are proportional. These rate
proportionalities are called stoichiometric ratios.
A well-formulated feed media is designed to add
nutrients at appropriate stoichiometric ratios
to match their consumption rates. This strategy
should allow all nutrients to be replenished at
each particular cell lines rate of consumption.
Stoichiometric ratios can be calculated using
historical culture data obtained from the cell line
of interest, as discussed in the stoichiometry and
kinetics chapter. Ideally, the data should be obtained
from cultures under relevant cultivation conditions.
Typically, one medium component is chosen as a
reference nutrient. The consumption rates of other
nutrients are related to the reference nutrient,
using stoichiometric ratios. Common choices for
reference nutrients are glucose, glutamine, oxygen,
and lactate, as they are consumed or produced
in larger quantities among all nutrients and
metabolites and are relatively easy to measure.
The idea of stoichiometric feeding is that if all
nutrients are consumed in relative proportions,
once the amount of the reference nutrient
consumed is known, one can calculate the amount
consumed of all the other nutrients using the
known stoichiometric ratios. The stoichiometric
ratio can thus be used as a basis for designing feed
media. Using this strategy, one would determine
the consumption of the reference nutrient and
then feed the culture with the feed medium to
appropriately replenish lost nutrients, according
to their calculated stoichiometric consumption.
Stoichiometric ratios are not necessarily constant.
FED BATCH |238
Cells may not necessarily consume all nutrients

in the same proportion in different culture
conditions or stages. The feed media composition
may need to be adjusted, according to known
changes of stoichiometric ratios under different
culture conditions or in different growth stages.
Fig. 10.6: Use of cumulative nutrient consumption data to

calculate stoichiometric ratio
Table 2. Amino Acid Specific Consumption Rate and

Composition in Cell Biomass and IgG
Normalized Specific Rate (relative
to lysine at low L/G state)
Normalized
Composition
High L/ Intermediate Low L/

G state
State
G state
Cell
IgG
ALA
-11.50
-0.93
0.08
1.32
0.76
ARG
1.25
1.25
1.25
0.69
0.35
ASN
0.23
0.25
0.03
0.00
0.50
ASP
0.05
0.23
0.10
1.47
0.57
CYS
0.00
0.00
0.00
0.04
0.35
GLN
16.25
9.00
4.50
0.00
0.72
GLU
0.50
0.25
0.13
1.84
0.74
GLY
0.35
1.28
0.10
1.33
1.00
HIS
-0.75
0.75
0.50
0.32
0.24
ILE
-0.75
1.25
1.00
0.84
0.35
LEU
0.75
1.75
1.25
1.31
0.98
LYS
1.50
2.25
1.00
1.00
1.00
MET
0.78
0.58
0.10
0.33
0.20
PHE
-0.40
0.38
0.45
0.54
0.50
PRO
0.00
0.00
0.00
0.80
1.02
SER
1.08
0.90
0.45
0.90
1.85
THR
2.00
1.50
0.75
0.79
1.11
TYR
-0.50
0.25
0.50
0.40
0.59
VAL
-1.25
0.75
1.00
0.95
1.30
L/G: stoichiometric ratio of lactate to glucose
Feed medium design, using this method, is best

approached as an iterative process, where a design
is tested, analyzed, and refined until an optimal
medium emerges. Most basal media are designed
with an excess supply of amino acids, especially
glutamine. In addition to being used for protein
synthesis, glutamine supplies precursors for
nucleotide synthesis and replenishes intermediates
for the tricarboxylic acid cycle. The consumption of
glutamine and other amino acids can often be reduced
without affecting growth and product formation.
The reduced consumption of these nutrients also
decreases ammonium production. In the feed
medium optimization process, the concentration
of some amino acids may also be reduced if excess
ammonium accumulation becomes growth inhibitory.
The first round of iteration usually employs data
from batch culture growth, in which amino acids
are supplied in excess. The stoichiometric ratios
determined from batch culture are then used to
prepare the initial version of feed medium. Often,
the concentration of nutrients in the feed medium
is about 10- to 15-times that in basal medium,
in order to reduce the required feed volume.
During the optimization runs, the concentration of

the reference nutrient (usually glucose, glutamine,
or oxygen), which can be measured on-line or offline rapidly, is measured periodically to determine
the consumption rate. The amount of feed medium
needed to replenish the reference nutrient to the
original level or a set point is then calculated and
added. If the stoichiometric ratios are estimated
correctly, the level of nutrients should remain within
the set range. If the stoichiometric ratio for a nutrient
is not accurate, its concentration will then increase
or decrease. After the optimization run, a new set of
stoichiometric ratios is then determined and is used
FED BATCH |239
Feed Medium Design for Unconsumed

Components
For Mg+2 and PO4-3, intracellular concentration >>

medium concentration
At high cell concentration, cell consumption can
cause depletion in medium
Need to supplement about 1x in feed medium
for the design of the next feed medium. Using these

methods, a very effective medium is typically achieved
with fewer than three rounds of optimization.
Some medium components are consumed in very

small quantities and their concentrations remain
virtually unchanged in culture. This includes many
inorganic ions, such as sodium, calcium, and sulfate.
From a stoichiometric point of view, the feed
medium does not need to include these nutrients.
As feed is added, however, the volume of the culture
increases and causes the medium components to be
diluted. Thus, these components are included at low
levels in feed medium (typically 1X concentration
or less). In some cases, inorganic salts, such as NaCl,
are completely eliminated from the feed medium
to reduce any changes in culture osmolarity.
Some ions, such as magnesium (complexed with ATP),

phosphate (as free phosphate or in nucleotides),
and potassium, are present at a much higher
concentrations inside the cell than in the medium.
At high cell concentrations, the amount of these
ions taken up by the cells may become significant.
Therefore, it is often necessary to compensate their
consumption by supplying them in the feed medium.
In addition to basal medium components, protein
hydrolysates, serum, insulin, transferrin, vitamins,
and lipid additives are also used in culture. These
additives supply minute nutrients, which are
consumed or become inactive with time. However,
the concentrations of these medium components
are not usually measured and their consumption
rate is mostly unknown. In the absence of a reliable
measurement consumption rate for those additives,
one has to rely on an order-of-magnitude estimate
of the upper and lower limits of their consumption
rate. The feeding rate of those additives is then
chosen to maintain their concentrations above
a minimum threshold, while below a maximum
tolerable limit which is experimentally determined.
FED BATCH |240
Control Strategies for Fedbatch Cultures
Developing Fedbatch Strategy

Define objective of feeding. Possibilities include:
Sustain nutrient level
Manipulate growth rate in a range
Define what criteria to manipulate
e.g., what level of nutrient to control?
Mode of feeding
Continuous vs. intermittent
How to deliver the feed medium
By direct measurement of nutrient or infer from
indirect measurement
Control Objective and Criteria:

Productivity and Product Quality
Control objective extends beyond acheiving high
productivity to high product quality and consistency
Require an understanding of relationship between
manipulated variables and control objective
How the feed medium is delivered to a culture may

affect the performance of a fedbatch process. Online measurements may be employed to determine
the consumption of the reference nutrient in real
time, then used to drive a fully-automated system
to feed nutrients continuously. At the other extreme, one can use off-line monitoring of nutrient
level and manually feed the nutrient periodically.
Three elements should be considered in developing a feeding scheme: 1) the control objective
and the control criterion (level of different nutrients to be maintained), 2) the mode of feeding (continuous or intermittent feeding), and
3) a control strategy for determining the timing and amount of feed medium to be delivered.
The simplest strategy is to allow nutrient levels

to vary, within a wide range, by adding large
quantities of feed media at widely-separated time
intervals (e.g., once or twice per day) based on
off-line measurements or historical data. Such
a periodic feeding scheme is very simple and is
usually sufficient to avoid depletion or overfeeding
of nutrients. For more specialized cases, especially
those aiming to manipulate cell metabolism, a more
frequent measurement of parameters, along with
well-controlled feeding schemes, are necessary.
A common objective of nutrient feeding is to increase
product titer by increasing cell concentration.
In recent years, the objective has been gradually
extended to assuring product quality and consistency.
Culture conditions may affect glycoform or protein
folding. The depletion of a particular amino acid
may cause an amino acid to be mis-incorporated
into the protein product. Culture conditions
may also affect the frequency of deamination or
glycation of the protein product after it is secreted
into the medium. Furthermore, extensive cell death
may cause product degradation or desialylation.
The control objective of feeding may be to reduce
glycation by setting a control criterion to keep
FED BATCH |241
Possible Control Objective and Control Criteria

Control Objective
Reducing desialylation by reducing cell death and
sialydase release
Reducing glycation
Control Criteria
Control osmolarity below set point
Control glucose level below set point
glucose concentrations below a certain level, or

to minimize amino acid mis-incorporation by
maintaining amino acids above a certain level.
Another method of control is to minimize protein
quality deterioration by maintaining a long,
slow growth phase and avoiding a death phase.
After defining the control objective, the next step is to

identify the process variables that cause the process
to deviate from the control objective. Furthermore,
a relationship between the key process variable
and the control objective, such as inducing lactate
consumption, should be available so that the process
variable can be controlled within an optimal range.
For mammalian cell culture, although the control
objectives can be defined, the factors that affect
the path to the objective are not easily identified.
Lacking knowledge of the relationships between
manipulated variables and the control objective
makes the optimization of control criteria difficult.
In the past few years, it has become empirically evident
that maintaining a moderately high osmolarity in a
late stage of culture can increase the productivity of
recombinant proteins. A high osmolality increases
productivity, perhaps, by enhancing the expression
of stress responsive proteins, which facilitate
protein folding, although the exact mechanisms
are not known. By keeping the osmolality only at a
moderately high level to avoid causing a rapid decline
of cell viability the culture can be sustained over a
longer period of high specific productivity, leading
to high final product concentration. This practice
is commonly used in industrial manufacturing.
An increasingly common practice is to employ
atypical culture conditions, which may not be
optimal for long-term maintenance or expansion
of cells in the final production reactor. These
conditions may include very high glucose
concentrations (15 g / L or 83 mM), low pH (6.9 vs.
7.2 for optimal growth), and/or low temperature
(33/34 oC vs. 37 oC). These conditions have been
found to affect the duration of culture and the
overall cell metabolism, resulting in an increased
product titer. A high concentration of glucose
FED BATCH |242
also increases osmolality. This is compensated by

reducing the salt concentrations to keep osmolality
around 300 mOsm during the cell growth stage.
Feeding Strategies
Feeding by Direct Measurement of
Nutrient Consumption
In-Time Measurement for Control

On-line or off-line
Example

On-line sampling coupled

HPLC for amino acids
Enzyme assays for glucose, glutamine,
lactate
Different Way of Coupling Feeding to On-Line

Measurement
Single stream with fixed stoichiometric
ratio
Multiple streams ratios among different
nutrients can be adjusted
Direct measurement of nutrient concentrations

is the most straightforward way to determine the
amount, rate, and timing of feed to be added. Based
on current concentrations of nutrients, one can
determine how much medium should be added to
sustain the nutrient level for a given period of time.
The concentrations of some nutrients can

be determined on-line, although off-line
nutrient measurement is more common.
Glucose and glutamine are the two nutrients
most commonly measured and used as
controls because their concentrations can be
determined rather rapidly in laboratories.
Direct measurement of these compounds on-line
can be implemented using an auto-sampling device,
in series with commercially-available immobilized
enzyme/membrane-based measurement devices.
HPLC can also be implemented as an on-line

approach for the measurement of glucose and amino
acids. This technique requires a series of processing
steps for sample preparation before injection into
the HPLC. A significant lag time between sampling
and delivering control action is unavoidable.
However, since the doubling time of mammalian
cells in culture in generally exceeding fifteen hours,
even an hour lag time in HPLC measurement of
amino acids is acceptable. Industrially HPLC on-line
analysis has been implemented at pilot plant scales.
FED BATCH |243
Proportional Feeding With Base

Addition
Rationale
In lactate production state L / G is relatively constant
Base added to neutralize pH at 1 mole base / mole
lactate
Use stoichiometric ratios to lactate to feed
Easy to implement, but its sensitivity is limited by
buffer in medium
A widely-used control strategy in bioprocessing is to

proportionally add medium according to the amount
of base added to the culture for controlling pH. This
provides the option of continuous on-line feeding
without an on-line nutrient measurement device.
In most cultures, a large fraction of the glucose
consumed is converted to lactic acid. To
maintain culture pH, one mole of base is added
to neutralize each mole of lactic acid produced.
If the stoichiometric ratios between lactic acid
production, glucose consumption, and other
nutrients are relatively constant, the rate of
lactic acid production can be used to estimate the
consumption rates of glucose and other nutrients.
Barring the effects of pH buffers (such as sodium

bicarbonate or HEPES), the base addition rate is
indicative of lactic acid production. This method
is simple and easy to implement. However, it is
highly sensitive to CO2 level in the gas phase and
sodium bicarbonate concentration in the medium.
Proportional Feeding With Turbidity

Turbidity is a good indicator of total cell density
(but not viability)
Feed nutrients proportional to cell density
Reliable during growth stage
Capacitance probe can measure viability
Proportional feeding according to base addition

is not well-suited for processes requiring a highly
accurate control of feed rate. Medium buffer
capacity can cause delays in base addition and
decrease the overall sensitivity of the method.
The use of an online laser turbidity probe can provide

accurate estimations of cell concentration in culture.
Simple proportional feeding with turbidity works
well during the exponential growth phase, when
viability is high and the growth rate is relatively
constant. However, near the end of the exponential
growth phase, when viability drops, an assessment
of viability or metabolic activity must be used to
adjust feed rates to avoid over-feeding. An alternative
measurement to turbidity is measurement of
capacitance, which reflects viable cell volume, i.e.,
viable cell concentrations and cell sizes. This may
be used for better control of nutrient feeding rate.
FED BATCH |244
Proportional Feeding With Oxygen

Uptake Rate (OUR)
Most sensitive among all on-line measurement of

metabolic activity
Require developing a computer algorithm
Need to develop stoichiometric ratios to oxygen
Among different measurements of nutrient

consumption, the oxygen uptake rate (OUR) is most
accurate for assessing cellular metabolic activities.
OUR measurements, unlike pH, are not masked by
buffers in the medium. A small amount of oxygen
consumed in culture, even in the range of 10 M,
can be accurately determined using a dissolved
oxygen sensor. For other nutrients which are often
present at millimolal or hundreds of micromollal
levels, such sensitivity of measurement cannot be
easily achieved especially for on-line measurement.
With its high sensitivity, small changes of OUR

can be confidently detected on-line and in real
time, thereby providing an immediate indication
of changes in metabolic rate. On-line oxygen
consumption data can then be used to determine
nutrient demand, using established stoichiometric
ratios
and
control
continuous
feeding.
Delivery of Feed Medium

Parameters in Feeding
Initial culture volume vs. feeding volume
Feed concentration
High concentration desired, but stability/
precipatation concern
High salt content from solubilization
Single feed vs. multiple feed solution
Need adjustable stoichiometric ratio
Continuous vs. step-wise feeding
Step-wise feeding
Frequency
Equal volume each time or proportional to
cell density
Operational simplicity vs. controllability
After determining the amount of feed media to be

added, one needs to decide on the mode of medium
delivery (e.g., the frequency and amount of feed to be
delivered at each feeding). The method of feed media
delivery is constrained by equipment and also by the
composition of the feed medium. The feed medium
usually consists of a solution of concentrated
amino acids and other organic nutrients, which,
when kept over a long time, tend to precipitate.
The main consideration in determining the proper

feeding frequency is the acceptable range of nutrient
concentration. The concentration of nutrients will
fluctuate between a high level at the time of feeding
and the low point immediately before the next
feeding. More frequent feeding reduces the deviation
from the set point. On-line feeding, by coupling to
base addition, turbidity, or to OUR measurement,
is easily implemented by computer control and
is almost continuous. When feeding is coupled
to less frequent off-line measurements, medium
is typically added manually, a few times a day.
FED BATCH |245
Processes employing continuous feeding to

control nutrient levels in a small range will
require substantially more effort than intermittent
feeding strategies. While allowing more control
over environmental conditions, the superiority of
continuous feeding in terms of extending culture
lifespan and increasing productivity has not been
clearly documented, except in the case where a
metabolic shift is the desired outcome. In fact, with
simple off-line monitoring, robust, intermittent
feeding strategies is standard industrial practice.
Online Estimation of Stoichiometric

Feeding
Detecting Stoichiometric Ratio Change During
the Culture
Monitor glucose, lactate, glutamine, OUR
Monitor L, G, G in, OUR and check their ratios
over time
A challenging issue of stoichiometric feeding is the

adjustment of the feeding rate, or feed composition,
in response to metabolic changes throughout a
culture [12]. This is particularly critical when
fedbatch cultures are used to elicit a metabolic shift.
Changes in metabolism over the course of a
culture are commonly seen, as evidenced by the
nonlinear relationships between specific nutrient
consumption rates. Many such changes bear
little consequence on cell growth or productivity,
but in some cases the effect is profound.
For simplicity, major changes in feed composition

or feed rate are only made when profound changes
in metabolism are observed. For example, in a
late stage of growth, cells may cease to produce
lactate and consume glucose at a slower rate.
Excessive glucose feeding may, then, reverse the
metabolism in the favor of lactate production.
Major changes in the metabolic rates of glucose,

lactate, and glutamine can be detected by
monitoring their stoichiometric ratios throughout
the course of a culture. On-line measurement of
nutrient levels would allow for timely detection
of changes in stoichiometric ratios; however,
its widespread implementation in industrial
processes will require further development
of reliable, automated sampling methods.
Without on-line, direct measurements of glucose
and lactate to detect changes in stoichiometric
FED BATCH |246
ratios, one may resort to indirect estimation. For

example, changes in OUR and the base addition
rate may be an indication of a changing metabolic
state, as the ratio of OUR to lactate consumption
changes significantly when cells switch from a
lactate production state to a consumption state.
Concluding Remarks
Fedbatch culture is the prevailing mode of cell
culture in the final production of manufacturing
recombinant proteins. It is also commonly used
for extending the period of cell expansion for
pre-production culture. In the latter, the culture
conditions and feeding strategy are designed
for optimal cell growth, whereas in the former,
the conditions are often designed to elicit a high
productivity, high cell concentration, and high

product accumulation. The design of feed medium
and the selection of the feeding control strategy are
critical to the successful implementation of fedbatch
culture. By applying stoichiometric principles to
feed medium design and by using a well-designed
feeding strategy, an optimal fedbatch culture
process can be implemented with relative ease.
FED BATCH |247
FED BATCH |248
Cell Retention and Perfusion

With contributions from
Sadettin Ozturk, Chun Zhang, and Weichang Zhou
Practice of Perfusion Culture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Analysis of Perfusion Culture. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Material Balance on Perfusion Culture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Effect of Recycling Factor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Methods of Cell Retention. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Sedimentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Incline Settling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Centrifugation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Acoustic Resonance Enhanced Settling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Spin Filter Separation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Microfiltration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Alternating Tangential Filtration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
249
251
251
253
254
254
255
256
257
258
259
260
261
Practice of Perfusion Culture

Marketed cell culture products
using perfusion bioreactors
Product
Company
RecombinantTM, Antihemophilic
Factor (recombinant), (Factor VIII)
Kogenate-FS (Factor VIII)
Aldurazyme
Naglazyme
ReoPro (IgG Fab Fragment)
Remicade (IgG1
Simponi (IgG1)
Xigris (Protein C)
Cerezyme
Fabrazyme
Myozyme/Lumizyme
Gonal-F (follicle stimulating hormone)
Vpriv (velaglucerase alfa)
Replagal (agalsidase alfa)
ReFacto (Factor VIII)
Baxter
Bayer
BioMarin
Centocor
(J&J)
Eli Lilly
Genzyme
(Sanofi)
Serono
(EMD)
Shire
Wyeth
(Pfizer)
Many sectors of the chemical process industry

underwent major transformation during the
second half of the 20th century. One of the most
significant changes is the switch from batch
processes to continuous processes. By minimizing
equipment turnover time and process start-up
time, a continuous process can be sustained at
a high rate of productivity over a long time and
achieve a higher throughput than a batch process.
Increasing adoption of the continuous process

method also sparked much research into
bioprocessing. Many enzymatic biocatalyst
processes have long been operated continuously,
similar to most waste treatment or biodegradation
processes. Contrary to the practices of the
chemical process industry, however, continuous
process only became more common in the later
years of microbial and cell culture bioprocessing.
A vast majority of biochemical processes involving
microbial or animal cell cultivation are batch
CELL RETENTION AND PERFUSION |249
Advantages
Better Product Quality
Better controlled culture environment (nutrients
& byproducts)
Pseudo steady state operation (ease of control)
Shorter residence time
Higher cell viabilities & lower concentration of
impurities
Critical for unstable molecules
More Economical
Higher cell concentrations & higher
productivities
Smaller bioreactor size
More flexible
Faster start up in process development
Disadvantages
Longer Cycle Time
Longer process development & validation time
Higher contamination risk
Higher equipment failure risk
Potential regulatory/licensing issues
processes for many reasons. First, unlike catalysts

and reactants do not change their behavior over
time in chemical reactors (although catalysts may
gradually lose their activity), microbes and cells
may mutate, evolve, and change the make-up of
their population and production capacity. Second,
the risks of microbial contamination and equipment
failure make long operations undesirable, especially
in manufacturing, for which process robustness is
of paramount concern. This is especially true when
producing high value pharmaceuticals. Finally,
the current product capture and purification
operations are all designed for batch mode. Even
if the production is operated in a continuous
mode, the process is not designed to realize
all of the advantages of a continuous process.
A continuous culture is constrained by the maximal

flow rate at which it can operate. One cannot operate
at a flow rate that is faster than cells growth rate;
otherwise cells are washed out and incapable of
replenishing the culture content. This is particularly
daunting for processes that must be performed at a
flow rate higher than the cell growth rate. In some
cases, cells produce growth inhibitors that must be
continuously removed by media replenishment. In
other cases, cells must be grown at a low growth rate
to achieve a high productivity, resulting in the media
being removed at a faster rate than the growth rate.
To overcome this shortcoming, a cell recycling

system can be added to continuous culture. By
recovering cells from the effluent flow and returning
them to the reactor, one can operate a continuous
culture beyond its natural limitation of the dilution
rate (flow rate divided by the bioreactor volume).
With a higher cell concentration in the reactor, the
overall throughput of the reactor is then also higher.
When cell culture was first adopted as the production

vehicle for biopharmaceuticals, continuous
operation was explored as an industrial process.
Researchers realized the growth of mammalian cells
in batch culture is impeded by lactate and ammonium
accumulation. A continuous process, thus, alleviates
growth inhibition by removing metabolites through
a continuous flow of medium. However, the high cost

of medium, especially serum, and the low product
concentration made continuous culture impractical.
To enable continuous operation, cells are retained

in the reactor while the media flow flushes
out metabolites. This can be accomplished by
immobilizing cells on some solid particles to
prevent them from being flushed out by media
flow, or by separating cells out from the effluent
stream and recycling them back to the reactor.
Thus, the general idea is continuous culture with
cell retention. Theese processes are often called
perfusion, which is reminiscent of the procedure
of flowing fluid through an organ or tissue.
Analysis of Perfusion Culture

Material Balance on Perfusion Culture
Fig. 12.1: Schematic of a continuous culture with cell recycle
A number of biotherapeutic proteins are produced

by perfusion processes. Some recombinant
antibodies could easily be produced with a fedbatch
process instead. It should be noted that the selection
of process mode is often based on the availability
of in-house expertise and many other factors.
However, in the cases of labile products that may
be degraded or otherwise inactivated over time, a
perfusion culture alleviates the loss of productivity
that cannot be easily overcome in a batch culture.
For products that accumulate only at very low
concentrations, a perfusion process may also present
a competitive advantage over a batch or fedbatch
One can analyze a perfusion culture system by

performing material balances on the reactor system.
The flow rates (F) of fresh medium entering and
exiting the system are the same, to keep the volume
in the reactor constant. The fresh medium stream is
free of cells. We also assume that the reactor is well
mixed, so that the nutrient (substrate) concentration
in the reactor is the same that it is in the effluent
stream. A cell separator is used to process the reactor
effluent stream to return a concentrated cell stream
with a given cell concentration, cx, back to the reactor.
Note that the effluent stream from the reactor has a
higher flow rate than the fresh feed, with a flow rate
of (1+)F instead of F, to balance the recycle stream
Forthebalanceequationonbiomassforthebioreactoris
dx
V = Fcx (1 + ) Fx + xV
dt
Balanceonthecellrecyclesystemgives:
(1 + ) Fx = Fcx + Fx2
(1 + c) Fx =
Fx2
x2
=1 + c
x
Thebalanceonthesubstrateonthebioreactor
V
x
ds
= Fs0
V F (1 + ) s + Fs
dt
Y
DefiningF/V=D.Applyingsteadystateconditions:
=
0 Dcx (1 + ) Dx + x
= D(1 + c)
0=
Fs0
x
Yx s
0= D( s0 s)
D = Dilution rate
practically c>1, so D>
V Fs
x
Yx s
Fig. 12.2: Comparison of cell concentration profile at

different dilution rates with and without cell recycle
Cell recycle (or retention) can accomplish:

A higher cell density in reactors
A higher dilution rate than maximum specific
growth rate
at the F flow rate. The symbols used are the same

as those in the stoichiometry and kinetics chapter.
Material balance can be performed on both the

reactor and the cell recovery/recycling device for
both cells and substrate. An important parameter
affecting the performance of the system is the
ratio of cell concentration in the purge stream, x2,
to that in the reactor, x. This ratio is affected by
the recycling factor, , and the cell concentration
factor, c by x2/x = 1+-c. Note the ratio should
be always smaller than 1, so that there is a
concentration effect by the cell recovery device.
An important conclusion from the steady state

analysis is that with cell recycling, the dilution rate,
which is defined as the flow rate divided by the
reactor volume, is larger than the specific growth
rate; whereas without cell recycling (i.e., =0), the
dilution rate is always the same as the specific growth
rate. As seen in the figure about cell recycling, the cell
concentration in the reactor is higher than without
cell recycling and the reactor can be operated at a
dilution rate higher than the growth rate. Therefore,
for cells with a doubling time of one day, the maximum
flow rate that can be used without cell recycling
would be one volume a day. With a perfusion
culture, even a few volumes a day may be operated
depending on the cell retention factor (c). A higher
retention factor permits a higher cell concentration.
Depending on the retention device used, the
efficiency of retention may vary; for instance, it
may decrease rapidly at the high dilution rate,
causing the cell concentration to decrease rapidly.
The analysis described above is for a bioreactor

with an external cell recovery device. The same
principle applies to a system with an internal
device. The consequence of employing an internal
device is the same: the dilution rate can be higher
than otherwise. It allows a fast nutrient flow rate
to be used to reduce metabolite concentrations,
while keeping cells in the reactor and allowing
cell
concentrations
to
become
maximal.
Effect of Recycling Factor
In a perfusion system operated at steady state, the

amount of cells purged (Fx2) from the cell recovery
device equals the amount of cells produced through
reproduction (xV). To keep a high viability and
steady state, a small amount of cells are purged,.
The system is operated at a dilution rate, D.
For a given dilution rate, one may employ a different

recovery device with a different cell retention
efficiency. Using the equation as an illustration,
given the same dilution rate and specific growth rate,
one may choose different combinations of and c. If
one chooses a highly efficient cell concentrator with
a large c, then that is used would be smaller, and
vice versa.
Fig. 12.3: A high cell concentration factor allows for a low

recycle ratio while achieving the same enhancing effect of cell
cycle
This concept illustrates that at a given dilution rate,

when using a highly efficient cell separator (such
as a centrifuge that creates a concentrated recycle
stream), a low recycling rate (F) can be employed.
Conversely, when using an inefficient cell separator
that gives a low degree of cell concentration, then a
large recycling rate needs to be used; in other words,
cells will need to be pumped out of the reactor and
pass through the cell separator to recycle more
frequently.
In large-scale operations, the cell stream could
potentially stay outside the reactor for a long time,
so oxygen starvation is a concern. Consequently, the
fluid stream out of the reactor is often chilled, first,
to reduce oxygen consumption. With a device that
gives a low c value, cells will need to be subjected
to the extra environmental perturbation of being
chilled and pumped more often. This factor should
be considered in selecting a cell recovery method.
Methods of Cell Retention

Selecting the method of cell separation very much
depends upon the way cells grow. The larger the
particles are, the easier they are separated. Many
processes employ microcarriers with particle
diameters ranging from 0.2 mm to 2 mm. These cellladen microcarriers can be easily separated from
media stream by sedimentation. In some cases,
cells are grown as large clumps, or aggregates, of 1
2 mm in size, also making sedimentation readily
applicable. Other considerations in selecting a cell
separator are the required throughput purge rate, the
concentration factor (c), and the scale of operation.
Because of the limit of the concentration factor

(c) of a cell separator, even though the purge
rate (F) may not be large, a large recycling factor
() may be necessary. Note that the capacity of
the separator required is not only based on the
dilution rate (D, or F/V), but also on the flow rate
out of the reactor ((1+)F). If a device with a small
concentration factor, c, is used, the capacity of the
separator will have to be substantially larger than
what is needed to process the purge stream alone.
Sedimentation
Conical Settler
Selective removal of dead cells
Low separation efficiency
Cell settling velocity ~ 2-10 cm / h
Good for larger cells
Long residence time outside of the bioreactor
Fig. 12.4: A settling cone for cell retention
The simplest cell separator is perhaps a settling

cyclone. The fed stream from the bioreactor enters
into the settler in its midsection, where the stream
splits into flows in two directions. The upward
stream, drawn by a pump for a purge stream,
encounters a large cross sectional area and, thus, the
vertical velocity is much smaller than the cells setting
velocity, so the cells move downward. The downward
stream, on the other hand, faces a decreasing cross
sectional area and, thus, increases its vertical
velocity as it carries cells downward. A transient
zone separates the two well-developed upward and
downward flow regions. In this transition zone, cells
are separated and carried downward into the reactor.
Such a settler is a convenient device for laboratory
operations. As the reactor scale increases, the flow
rate also needs to increase proportionally, but the
cross sectional area of the settler increases only
with 2/3 power of the settler volume. The efficiency

of cell separation decreases rapidly as the scale
increases, making it ill suited for larger operations.
Incline Settling
In a simple settling tank, the direction of fluid flow

and cell settling are along the same axis (both
vertical). A sufficiently long transient zone is
necessary to separate the cell and cell-free streams.
Operating parameters
Cell size and concentration
Perfusion rate
Settling area
Length of the plates
Inclined Settling
While a particle is moving upward with flow, it also
settles toward the bottom plate
It is collected upon hitting the bottom
Eventually, the particle rich zone has a higher fluid
density and begins to move downward
The particle-rich stream is recycled to the bioreactor
purge stream
Particle
Recovery
co
n
ce
t
rat
ed
ce
ll
str
ea
Particle
Setting
feed from reactor
recycle to reactor
Fig. 12.5: Cell separation in an inclined settler for cell

retention
To enhance the separation efficiency, the

settler is often inclined so that an angle exists
between the fluid flow direction and the particle
settling direction.
A particle is considered
collected once it settles on a settlers surface,
since the fluid velocity on the surface is zero.
In an inclined settler, the feed cell stream enters

at the bottom and moves upward. Inside the
settler, cells begin to settle down vertically
due to gravity. If a cell particle hits the surface
at the lower plate, it is collected and does
not get carried out by the effluent stream.
Eventually, the cells settled on the bottom plate form
a layer of fluid with a higher density than the stream
above. This heavy stream then moves downward,
carrying the cells along. At the steady state, there
are three streams in the system: the feed stream,
the effluent stream (carrying unsettled cells),
and the concentrated cell stream at the bottom.
In industrial design, multiple inclined plates
are used in a single settler. In such designs, the
feed stream and the returning cell stream do not
cross each other by partitioning their flow path
in different zones. In some cases, mechanical
vibration is applied to the plates to prevent settled
cells from sticking to the surface and being lysed.
The residence time in the settler has to be at least

as long as the particle settling time. With a high cell
concentration in the stream, oxygen starvation is a
major concern, as it may induce apoptosis and cell
lysis. Therefore, the stream passing through the settler
is often chilled to reduce the cells metabolic rate.
Centrifugation
Centrifugation Technology
Excellent separation efficiency
High perfusion capacity
Little clogging
Easy scale-up
Vulnerable to mechanical failure in long term
continuous operation
Three different designs:
Westfalia Disc type
Continuous cell recycle back to fermenter
Centritech Lab
Disposable separation unit
Fig. 12.6: The cell loading and cell ejecting regions of a

disc centrifuge for cell recycle
Centrifugation is a standard unit operation in many

downstream recovery processes. In the early days of
perfusion process development, it was among the first
to be exploited for cell retention. As early as the mid
1980s, the Japanese company Teijin had employed
centrifuges for perfusion culture of hybridoma cells
used to produce antibodies for cancer imaging.
However, most centrifuges are not designed for

long term and aseptic operations. Early use of the
centrifuge for perfusion was only intermittent and was
for batch harvest and cell biomass recycling, as well
as for the replenishment of fresh medium. A number
of autoclavable or stem-sterilizable centrifuges
are now available. These and the disposable bagbased centrifuge are all capable of processing
up to hundreds of liters of medium a day, and are
suitable for continuous use in perfusion culture.
The disc-type centrifuge is analogous to a multiple

parallel plate settler, except that the parallel plates
are rotating and generate a centrifugal field for
cell settling. The disposable bag system employs
three tubes: a feed tube, an outflow tube for the
heavy (cell-rich) stream, and an outflow tube for
the light stream. The unique design of an inverted
question mark allows the three tubes to rotate,
along with the centrifuge, without being twisted.
concentrated
cell stream
cell
dilute stream
feed
cell movement
Fig. 12.7: A disposable bag based centrifuge (Centritech) for

cell recycle
Acoustic Resonance Enhanced Settling This device uses acoustic energy to enhance
Mechanical/Acoustic Trapping
Enhanced sedimentation (Cell aggregation)

Slightly favorable for viable cell retention
Heat generation may create temperature gradient
Low separation efficiency
Moderate capacity (200L/day per unit)
Operating parameters
Cell density, perfusion rate, power input
cell agglomeration. As the cell stream from the

reactor passes through the acoustic chamber,
cells are induced to agglomerate. This gives
rise to a faster settling velocity. With increased
settling
velocity,
sedimentation
is
easily
accomplished without resorting to multiple plates.
As cell concentration or operating conditions (i.e.,

flow rate and temperature) change, the energy
and residence needed for agglomeration may also
light single cell stream

moves upward
heavy aggregate
stream moves
downward
change. Sensors for detecting cell agglomeration

will help stabilize the operation of the device.
transducer
reflector
agglomeration
zone
feed
recycle stream cell aggregates
Fig. 12.8: An acoustic cell agglomeration

device for cell retention
Spin Filter Separation
Fig. 12.10: A spinning filter (or rotating cage) for

cell retention
Rotational Cage
Internal and External
Remove dead cells/debris
Operating Parameters Affecting Performance
Screen pore size (1-120 m)
Perfusion rate
Rotation speed
Screen surface area
Draft tube
Screen materials:
Stainless steel, DNA & RNA deposit,
Teflon, Polyamide 66, polyethylene, better
Fig. 12.9. Picture of an acoustic cell retention

apparatus
The term spin filter is used to refer to two different

designs that may have rather different mechanisms
of operation. Both employ high porosity filter with
relatively large openings (~20 - 100 m) installed
on the wall of a rotating cage. Both are submerged in
culture fluid. A pump draws medium from inside the
cage as the purge stream. The culture fluid passes
through the filter to enter the cage, then is withdrawn
by the pump to become the effluent stream exiting
the reactor. The flow into the cage has a lower
cell concentration than the bulk culture fluid, thus
achieving the overall retention of cells in the reactor.
A rotating cage rotates along the center shaft of the
impeller agitator at a low speed. The centrifugal
field is typically insufficient to push away cells along
the outside wall of the cage. Yet, a boundary layer
of liquid around the cage probably exists, in which
the cell concentration is lower than in the bulk. As
a result, the fluid drawn across the filter is lower
than in the bulk. Furthermore, there is little filter
cake formation. The screen on the cage is, thus,
not exactly a filter. Nevertheless, the system has
been employed in scales of up to hundreds of liters.
Centrifugal Cage
dilute purge stream
recycle stream
Using such a device, the operation is less prone

to variation, due to changes in fluid dynamics.
It gives the freedom of operating in different
regions in the reactor. In some variations, it is
installed outside of the reactor and used as an
external cell retention device. However, the device
spinning
stationary
differs from the traditional centrifugal filter used

in the recovery process in chemical industry
in an important way in that no filter cake is
formed. In fact, if cell cake is formed on the
screen wall of the cage, cell death is likely to
occur in the cake and leads to process failure
Fig. 12.11: A Centrifugal filter for cell retention
Microfiltration
External Loop for Cell/Harvest Separation

High shear, tangential flow
At high transmembrane pressure, cell deformation
occurs
Fouling caused by high molecular weight DNA, protein,
lipids, anti-foam occurs after days of operation
Difficult to scale-down
The degree of concentration in a single pass is
relatively small
Cross Filtration Model
feed
The rotating cage is notoriously difficult to scale up,

as its operating mechanism is not well understood.
Later modifications of spin filters increased its
rotation rate up to hundreds of rpm, allowing it
to operate like a centrifugal filter. The centrifugal
force is sufficient to push the cells away from the
surface of the filter, thus drawing liquid through
with a lower cell concentration than in the bulk.
Microfiltration uses membranes of different

configurations, including parallel plates and hollow
fiber devices, and has a pore size of around 4 m.
Microfiltration was among the first techniques
used for cell retention. Its widespread use has been
impeded by membrane fouling, which is especially
severe when a high concentration of proteins or
complex medium is used in the culture. With the
use of low protein medium in the past decade,
the problem of protein fouling has lessened,
but clogging by dead cells remains problematic.
return to
reactor
cell-free permeate
cell
membrane
very high flux may
cause cell damage
Fig. 12.12: Microfiltration membrane for cell retention
Alternating Tangential Filtration
Different
configurations
of
microfiltration
devices all use tangential flow, meaning the
feed stream flows in a direction parallel to the
membrane, while the filtrate flows across the
membrane. The more recent use of a pulsatile flow
system, in which a diaphragm is used to rapidly
reverse the flow direction, has reduced fouling.
from reactor
return to reactor
product harvest
stream
product harvest
stream
diaphragm in
closed position
diaphragm in
open position
air
air
Rapid pulsatile flow in reverse
directions minimizes fouling.
Fig. 12.13: An alternating tangential flow hollow fiber device for cell retention
Concluding Remarks
Perfusion culture operation was explored very early
on in cell culture development for obvious reasons:
(1) the low throughput from batch operation can
be enhanced by switching to continuous operation,
(2) a simple continuous culture would have too
low of a cell concentration to make the process
economical, and (3) a dilution rate faster than the
cell growth rate is needed to purge the metabolites
accumulated in culture. Its widespread adoption
was inhibited, however, by concerns about
regulatory requirements and the lack of a clear
definition of BATCH for product manufacturing.
Researchers were also concerned about the lack of
a scalable cell recovery device for long operations.
These hindering factors have largely been overcome.
Although cell line stability for sustained production
of a product is still a concern, evidences have
shown that, with proper control of cell age and cell
stock, stability may not be an overriding concern.
volumes of inoculum that are used to scale up. The

carrying over of a large amount of spent medium
from seed culture is undesirable; evidence seems
to suggest that a high metabolite concentration at
inoculation can negatively affect a cells metabolic
characteristics in the main culture. Thus, using
the cell recovery devices for preparing inoculum,
especially to increase the initial cell density,
could potentially enhance process performance.
Another area that may change in the near future is

the increased use of fortified medium in perfusion
culture, as we have seen in fedbatch cultures.
Using fortified medium (instead of medium with
a standard composition) can reduce the flow rate
and increase cell and product concentrations.
Given the intrinsic advantages of continuous

operations and the advances in cell retention
technology, we may begin to see more widespread
application of perfusion culture in the coming
As cell retention technologies become mature, one
years. It has considerable potential to increase
may also see this application go beyond perfusion
the capacity of high throughput processes, reduce
culture. The same devices can be used to concentrate
reactor sizes, and possibly minimize product quality
cells and remove metabolites, as well as to quickly
fluctuations through steady state operations.
replace the fluid phase for cell freezing operations.
Current inoculation operations are limited by the
Scaling Up and Scaling Down

for Cell Culture Bioreactors
Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Mechanical Agitation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Mechanism of Agitation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Power Consumption and Mixing Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Power Consumption of Agitated Bioreactors . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Other Dimensionless Numbers for Stirred Tank Reactors . . . . . . . . . . . . . . . . .
Effect of Scale on Physical Behavior of Bioreactors . . . . . . . . . . . . . . . . . . . . . . .
Mixing Time. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Nutrient Starvation Time. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Mixing Time. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Nutrient Enrichment Zone, Mixing Time vs. Starvation Time . . . . . . . . . . . . .
Mixing Time Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Scaling Up and Mechanical Forces on Cells. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Scaling Up and Oxygen Transfer and Carbon Dioxide Removal. . . . . . . . . . . . . . . . . .
Material Balance on Oxygen in Bioreactor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Aeration Rate and Superficial Gas Velocity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Gas Flow Rate in Scaling Up: A Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Effect of Scaling Up on CO2 Removal. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
263
265
265
266
266
268
269
271
271
272
272
273
274
275
276
278
280
281
284
Introduction
Translation of the process scale is one of the most
difficult issues in bioprocessing, and it is probably one
of the least systematically studied subjects in the field.
Few engineers are involved in designing large-scale
equipment using small-scale experimental data, but
many will be developing processes in laboratories
and at pilot plant scales for eventual implementation
in a production scale. Others may be involved
in troubleshooting investigations for production
plants using laboratory equipment. Therefore,
an understanding of the factors affected by scaletranslation is important in carrying out those studies.
SCALING UP AND SCALING DOWN ON CELL CULTURE BIOREACTORS |263
Translation of Scale
Objective
Prediction of process performance
Specify operating conditions from one scale to
another
Major Effect of Scale
Oxygen (Gas) transfer
Heat transfer
Shear force
Compression force
CO2 removal
Fig. 13.1: Schematic of a large

scale cell culture fermenter
A reactor may be scaled up geometrically similarly

or non-similarly. Geometrical similarity refers
to maintaining the ratio of the main geometrical
lengths, such as height over diameter, as well
as the relative size of the internal parts (e.g.,
impeller, flow diverter, etc). In this chapter, our
discussion will focus on geometrically similar cases,
which are conceptually easier to grasp, although
this is not always the best approach in scaling.
In scaling up geometrically similarly, all length
dimensions of the rector are scaled proportionally.
Consequently, the surface area will increase with
the length dimension to the second power, while
the volume will increase to the third power of the
length. As a result of scaling up, the scale-related
surface area per unit volume of equipment will
decrease. In microbial fermentation, the decrease in
surface area to volume ratio causes the impediment
for removal of heat generated from metabolism
and mechanical agitation. In mammalian cell
processing, the metabolic heat generated is less
a concern. However, the process may still be
sensitive to other variables related to scale change.
As the scale of an equipment changes, the physical
and mechanical parameters may not be maintained
constant. Frequently, it will not be possible to keep
all key operating parameters constant between
different scales. This may lead to changes in
the chemical environment and, ultimately, cell
physiology and productivity.
The objective of
scaling-up and scaling-down is, therefore, not to
strive to keep scale-related parameters at constant,
but to define the operating range of scale-sensitive
physical and mechanical parameters so that
the cellular physiological state and productivity
can be maintained within an acceptable range.
At times, this may require an adjustment of
the chemical environment at different scales.
SCALING
SCALING UP
UP AND
AND SCALING
SCALING DOWN
DOWN ON
ON CELL
CELL CULTURE
CULTURE BIOREACTORS
BIOREACTORS |264
Mechanical Agitation
Purpose of Agitation
Gas-liquid mass transfer
The higher shear field near the impeller tip
produces small bubbles, thereby increasing
gas-liquid interfacial area (provided that bubble
coalescence is not correspondingly increased)
Suspension of solid (e.g. microcarriers, soymeal) or
dispersion of liquid
Liquid-liquid, liquid-solid mass transfer (e.g.
hydrocarbon culture, quick mixing of pH neutralizing
base)
Minimization of pellets or aggregates
Pellets are cell aggregates or mycelial
microorganisms (streptomyces, molds)
Mixing, especially for viscous fluid (e.g. xanthan gum)
Fermentation, broth of mold culture
Mechanical agitation in a stirred bioreactor keeps

cells in suspension, provides mixing to create
a more homogeneous chemical environment,
and creates a flow pattern that increases the
retention time of gas bubbles in the culture fluid
to enhances oxygen transfer. In the cases that
cells are grown as aggregates, agitation also helps
reduce the formation of oversized particles.
In microbial fermentation, oxygen demand is rather

high (often exceeding 150 mmole / L-hr). To increase
the efficiency of the oxygen supply, extensive
agitation is used to break up air bubbles. In many
fermentations of mycelial mold or actinomycete,
extensive agitation is used to overcome the
high viscosity of culture fluid and to reduce the
mycelial pellet size to enhance oxygen transfer.
In cell culture processes, the oxygen demand is

nearly two orders of magnitude lower than that
in microbial fermentation.
Adequate oxygen
supply can usually be accomplished by much less
intensive agitation, which is also sufficient for
providing sufficient mixing and suspension of cells.
Mechanism of Agitation
Table 1. Characteristics of Impellers

Characteristics
Propeller
Disk Turbine
Flow direction
Axial
Radial
Gassing
Less suitable
Highly suitable
Dispersing
Less suitable
Highly suitable
Suspending
Highly suitable
Less suitable
Blending
Highly suitable
Suitable
The flow patterns generated by different impellers

in a stirred tank are generally classified as one of two
types: axial flow or radial flow. An axial flow pattern
refers to primarily upward or downward flow due to
the pumping action of the impeller. In a radial flow
pattern, the liquid moves primarily outward toward
the wall of the vessel. In cell culture processing,
impellers generating axial flow are used because
the shear fields generated by axial flow patterns are
lower than those generated with radial flow patterns.
The Rushton disk turbines, as are often used with
multiple installations in large reactors, are the
predominant type used in microbial fermentation.
In this design, the sparger is placed directly
underneath the disk turbine. Gas bubbles from the
sparger rise to hit the disk and are directed outward.
The blades, rotating at a fast speed, then break the

bubbles up. In the immediate surrounding region of
the blade, a very high-energy dissipation is predicted
by computer simulations, which would contribute
to bubble break up, but also create a high shear
zone which can potentially cause damage to cells.
Fig. 13.2: Fluid flow patterns in a stirred tank reactor: axial flow
type vs. radial flow type
Flat Blade (Rushton Turbin)
The propeller or pitch blade type impellers, on the

other hand, focus on moving the liquid to create
the lift for mixing and suspending solids. The
impeller diameter to tank diameter ratio should
be higher for microcarrier culture. While the
propeller three blades is used extensively in
microbial fermentation to enhance oxygen transfer,
the axial flow three blades provides less shear
stress and a more uniform velocity in the entire
discharged area than the propeller three blades.
Axial Flow Blade
Propeller Three Blade
Fig. 13.3: Three types of impellers commonly

seen in stirred tank reactors
Power Consumption and Mixing Characteristics

Power Consumption of Agitated
Bioreactors
N
PO
H
V
Di
DT
Fig. 13.4: Notation of an impeller

based mixing reactor. H: liquid
height, V: liquid volume, N: impeller rotation rate, Po: agitation
power, Di: impeller diameter, DT:
tank diameter.
In designing equipment and analyzing physical

systems, of which the scale spans over a wide
range, one often needs to develop a correlation
between different design variables. When data
collected from different scales and under different
conditions are plotted together, they inevitably
give rise to different correlations. Each of those
correlations is good for that particular scale or a
range of scales. They are, thus, of limited value.
In order to find a correlation that is applicable to

different scales over a wide range of operating
variables, the experimental data are often plotted in
dimensionless variables to develop correlations
among the experimental data. These dimensionless
variables are a combination of experimental variables.
In such combinations, the units of dimensions from
individual variables cancel each other out with the
idea being that a correlation between dimensionless
variables is not sensitive to scale. The dimensionless
correlations obtained from experiments performed
on different scales should hold on any scale,
including those that have not been investigated.
Fig. 13.5: Relationship between impeller Power number and

impeller Reynolds number for different types of impellers
Various relationships expressed in dimensionless

numbers are fundamental to fluid mechanics, mass
transfer, heat transfer, etc. The correlation between
friction factor (f) and Reynolds number (Re) is used
universally in the design of fluid flow in pipes. The
plot of friction factor and Re show two regions: at
low Re, f decreases linearly until Re = ~2000, where
there is a short break, then it continues at a relatively
constant value at a high Re region. The first region
is recognized as the laminar (or viscous) flow region
and the constant tail is the turbulent flow region.
A similar plot has been generated for power

n
consumption in a stirred tank reactor. The Reynolds
number is now denoted as ReI (Impeller Reynolds
Power Number (Np): 3Po 5
number) to indicate that it is based on the length
N Di t
(diameter) of the impeller. The dimensionless
number for power consumption by impeller is the
Po = Impeller power (ungassed power, Po indicated
indicated ungassed)
power number, Np. Correlations between Np and ReI
N = Impeller speed
have been generated for various types of impellers.
They all exhibit a similar behavior to the f vs. Re
Di = Impeller diameter
plot for fluid flow in a pipe. In all these Np vs. ReI
= Fluid specific gravity
plots, a rapid decrease with increasing impeller
= Fluid viscosity
Reynolds number is followed by a constant value
For all practical purposes a bioreactor is always
region; the decreasing region and the constant
operated in the turbulent region
region represent the two correlations for viscous
In turbulent regions: (Np) is constant, independent of flow and turbulent flow regimes respectively. In
(ReI)
the turbulent regime, the power number is constant

(Eq.
1)
over a wide range of impeller Renolds number, but
Po = K
3
5
tN Di
the value changes with different types of impellers.
2
Impeller Reynolds Number (ReI): NDi t
Other Dimensionless Numbers for

Stirred Tank Reactors
Three other dimensionless numbers, in addition to
power number, are used:
Dimensionless velocity, v / ND
Dimensionless pumping number, Qp / ND3
Dimensionless blending (mixing time), N
v: tip velocity
Qp: liquid pumping
: mixing time
At turbulent regions, all those numbers are

relatively constant.
In bioreactor operations the flow is always in the

turbulent regime. Viscous flow is encountered in
a stirred tank only when a very viscous fluid, like
glycerol, is used. Therefore the power number for
impellers is constant for the same type of impeller. In
other words, the impeller power divided by N3Di5 (N
is agitation rate, Di is impeller diameter) is constant.
Three other dimensionless numbers are frequently
used to predict the performance of a stirred tank
when the scale changes. They deal with three
important aspects of bioreactor operations:
velocity, volumetric flow rate, and mixing time.
The maximum velocity in a mixing tank occurs at the

tip of the impeller. This velocity can be represented
by the multiplicative product of the rotation speed
of the impeller times its diameter, NDi. (Note: in
this chapter, we will ignore in the discussion of
perimeter, area of circle, etc. The constant value
is cancelled out when comparing different scales.)
The amount of fluid that the impeller can move

(called pumping) is directly dependent on its
rotating velocity and the area of the impeller blades.
Because we are considering scale translation under
geometrically similar conditions, we can use the
length of impeller, instead of the impeller blade, to
represent the length scale. The pumping, then, is
the projected area (D2) of the impeller multiplied
by the velocity of its rotation (ND), which gives ND3.
For mixing time, the representative time

scale (called characteristic time) in a mixing
tank is the inverse of rotation speed (1/N).
The dimensionless numbers for the three properties

can by obtained by taking the representative velocity
(v), liquid volumetric flow rate (Qp), and mixing time
(), and divide by their respective characteristic
counterpart (e.g., ND, ND3, and 1/N). The plots of
dimensionless velocity, pumping, and mixing time
against ReI all show profiles similar to the power
number plot, with two distinct flow regimes: laminar
(viscous) flow and turbulent flow. The values at
high ReI turbulent regimes are relatively constant.
Effect of Scale on Physical Behavior of

Bioreactors
Scale Translation Approaches
Constant K a
L
Constant impeller tip speed (ND )
i
Constant power per unit volume (Po/V)
Constant mixing time
Table 2. Examples of Scaling Up by Keeping

Different Parameters Constant.
The reactor is scaled up 15.6 times by volume while
keeping geometric similarity.
Property
Pilot Scale
160 l
Plant scale, 2500 l
1.0
15.6
98.0
6.2
P/volume
1.0
1.0
6.2
0.4
1.0
0.54
1.0
0.4
1.0
2.5
2.5
2.5
Qp
(pumping)
1.0
8.5
15.6
6.2
Qp/volume
1.0
0.54
1.0
0.4
NDi
(tip speed)
1.0
1.35
2.5
1.0
Using the correlations based on the dimensionless

numbers, one can explore the effect of a changing
scale on different variables. We assume that
the equipment in different scales will remain
geometrically similar. In that case, the effect of
different reactor sizes can be compared using
characteristic length D (the tank diameter). If the
tank diameter increases by 10 fold, all the other
reactor parts (tank height, impeller diameter, etc.)
will increase by the same proportion of 10 fold.
In scaling up different processes, one needs to

keep the most important variable(s) constant or
within an acceptable range. The commonly used
criteria for scaling up are (1) a constant KLa, so that
mass transfer can be maintained, (2) a constant
impeller tip speed, to sustain a critical value of
high shear velocity to break up agglomerating
particles or pellets of mycelial cells, (3) a constant
power input per volume (usually for less power
intensive processes such as crystallization,
blending), and (4) a constant mixing time.
Consider the case of scaling up by maintaining power

input per reactor volume constant. Recall that the
power number is constant in a turbulent region and
the power input (PO) is proportional to N3D5. The
reactor volume is described by HD2. Because of
geometrical similarity, we can represent H by D and
ignore the constant that does not contribute to
scale comparison. The reactor volume (V) is thus
represented by D3. In keeping PO/V constant, N3D2
is also constant in different scales. In scaling up
as D increases, the rotation speed must decrease
by 1/D2/3. It is inevitable that larger reactors will
need to be operated at lower rotation speeds.
By similar algebraic manipulation one can also see

that scaling up by a constant power per volume (PO/
N3D2) constant will lead to increasing the amount f
total pumping (ND3) with the scale. However, pumping
per volume will decrease as the scale increases. For
mixing time, the trend is an increase with scale.
The table compares the effects of scaling up
by setting different parameters at a constant

value. Cases considered include constant power
per volume, constant agitation rate, constant
pumping rate, and constant tip speed. Often,
one chooses not to keep a single value constant,
and not to scale entirely geometrically similarly.
Scaling Up Geometrically Similarly By Keeping Power Per Unit Volume Constant
PO/N3DI5 is constant in turbulent region. Density of water, , is constant. Thus,

(Eq. 1)

PO = KN3DI5
The volume of reactor can be expressed as characteristic length raised to the third power,

V = HDt2 = cDi3
(Eq. 2)
The power per unit volume is described as

Po/V = K N3DI5/ cDI3 = KN3DI2 = constant
This leads to the conclusion that when power per unit volume is kept constant, N3DI2 is also constant.

N3DI2 = constant
(Eq. 3)
Effect on Agitation Rate
Comparing scale 1 and scale 2
N13DI12 = N23DI22
N1/N2 = (DI2/ DI1 )2/3
(Eq. 4)
The agitation rate N decreases with increasing scale.
When the diameter increases eight times, the agitation
rate is in the larger scale.
Effect on Impeller Tip Speed
Tip speed is described by N multiplied by Di. from Eq. 3

N13DI1 3/ DI1 = N23DI23/ DI2
N13DI13/ N23DI2 3 = DI1/ DI2

N1DI1 / N2DI2 = (DI1/ DI2)1/3 (Eq. 5)
Tip speed increases with increasing scale, but only at
1/3 power of the length of scale.
Effect on Mixing Time

The decreased pumping per volume also causes
an increasing in mixing time when scale increases.
The dimensionless mixing time is N. Its value
is relatively constant in the high Re number
turbulent region.

The capacity of liquid pumping can be described

by the impeller tip speed, NDi, by the area that it
moves against the liquid, Di2.
Under the condition of constant power per volume,
N13Di12 = N23Di22. Multiply both sides by diameter to
the seventh power.
(Eq. 7)
N13Di19 / Di17 = N23Di29 / Di27

N1Di13 / N2Di23 = (Di1 / Di2)7/3

Liquid pumping capacity increases with scale. By
dividing both sides by characteristic length raised
to the third power, we can obtain the pumping
capacity on a per volume basis
NDi3 / Di3 = pumping per volume = Qp/V (Eq. 8)
(Qp1/V1) / (Qp2/V2) = (Di1 / Di2)-2/3 = (Di2 / Di1)2/3
The pumping capacity per volume actually
decreases with increasing scale.
1N1 = 2N2
From Eq. 4

Effect on Liquid Pumping
(Eq. 6)
1 / 2 = N2 / N1 = (Di1 / Di2)2/3
Mixing Time
The nutrients that are added at the beginning of
the culture will eventually become well mixed
in the culture fluid. (Note: This may not be true
for some microbial fermentation in which some
nutrients are supplied in a solid form and dissolve
gradually). Mixing problems may arise for those
components that are added continuously or
intermittently during the cultivation. If the nutrient
is added in a fixed position(s), it may not get carried
to other locations in the reactor fast enough to
meet the cells metabolic needs. In some cases,
the additive needs to be supplemented at a high
enough concentration that has an adverse effect
to the culture, and must be dispersed quickly. In
these cases, adequate mixing must be provided.
Nutrient Starvation Time

Because of its solubility, oxygen is the first nutrient species
to be completely consumed
Table 3. Comparison of Oxygen and Glucose Saturation

Time in a Typical Culture (For 1010)
Oxygen
0.1 mM
(50% saturation with
air space)
C in Culture
Glucose
1 g / L (55 mM)
Specific
consumption
rate
1 x 10-10 mmole /
cell-hr
Volumetric
consumption
1 mmole / L-hr
0.15 1 mmole / L-hr
Time to
depletion
0.1 hr (6 min)
12 hr
0.15 1.0 X 10-10 mmole

/ cell-hour
In cell culture processes, oxygen is almost always the

first nutrient to be depleted. Due to its low solubility
in medium, it must be continuously supplied.
In comparison, the concentration of glucose
maintained in the medium is usually a couple orders
of magnitude higher than oxygen. Their ratio of
molar specific consumption rates ranges from about
1.0 (when most glucose is converted to lactate) to
close to 6.0 (when most glucose is converted to
CO2). For oxygen, at a high cell concentration, the
depletion time can be as short as a few minutes,
whereas the depletion time for glucose is orders
of magnitudes longer.
Therefore, in reactor
scaling, special attention is always paid to oxygen.
Mixing Time
Mixing Time Measurement
Measurement
At t = 0 add tracer
Measure terminal mixing time, defined as the
point when an arbitrary chosen uniformity, 90%,
is reached
To measure the mixing time, one can inject a

dye into the reactor under operating conditions
and then use a sensor placed in a fixed location
to record the dye concentration over time. The
concentration will fluctuate over a large range
initially, but will eventually reach a steady value.
The time needed for the concentration to reach
a range of steady value (such as 90% of its final
homogenous concentration) is considered the
mixing time.
If one plots the concentration
deviation from its final steady value, C, the time
profile can be approximated by first order kinetics.
Fig. 13.6: Measurement of averaged mixing time in a stirred

tank
Nutrient Enrichment Zone, Mixing Time

vs. Starvation Time
In a stirred tank reactor, a nutrient is added at a fixed

position. Consider a fluid element carrying cells. When
it passes by this position, it acquires the nutrient and
then moves away to circulate around the reactor. On
average, the same fluid would return to this feeding
zone after a duration of one characteristic mixing
time. It is important that the amount of nutrient
that the fluid acquires at the nutrient enrichment
zone is sufficient to sustain the metabolic needs of
the cells before it returns to the zone. Therefore,
the mixing time needs to be shorter than the
starvation time. The starvation time is dependent
on cell concentration and the consumption rate.
If the mixing time is longer than the starvation
time, a discernible concentration difference of the
nutrient would appear in different locations in
the reactor. Pockets of low concentrations would
emerge and their locations may change over time,
as the fluid flow pattern and cell concentrations are
not constant. A sensor that is at a fixed position
may, thus, not reveal the presence of such pockets.
Mixing Time Distribution
Fig. 13.7: Measurement of circulation time distribution

in a stirred tank
Mixing Time Distribution Measurement

Add radio emitter to the reactor i, a sensor picks up
the signal when the emitter circulates around and
passes by
Measure circulation time for each encounter and
plot the frequency distribution of the circulation
time
Determine the mean and median circulation time
and the standard deviation
One can plot distribution of circulation time as
a population density function. The portion of
circulation whose time of circulation lies between
t and t + t is the area under the curbe between t
and t + t.
If we follow a particular fluid element in a

reactor, it will come to the nutrient enrichment
zone, go away, and come back repetitively. The
time interval between its return to the nutrient
enrichment zone, however, will not be uniform.
The mixing time described above is an average mixing
time. But the mixing time that is physiologically
important (e.g., the return time to the nutrient
enrichment zone) is not a single value of the
average mixing time, but is distributed over a range.
Imagine that we use a ball that emits a radio signal.

The ball has the same density as the fluid and is
carried around by the fluid motion. A sensor at a
fixed position in the reactor would pick up the
signal when the ball is close and record the time
interval between consecutive returns. This time
interval of return will distribute over a range as
the ball sometimes returns shortly after it moves
away, while at other times it roams around the
reactor for a while before returning to the sensor.
The histogram of the time-interval distribution can
be converted to a mixing time distribution function.
In general, the distribution follows a logarithmic
normal distribution. The mean or median of mixing
time is a descriptor of mixing characteristics, but it
does not present the entire picture of mixing. Given the
same median or mean mixing time, two reactors may
still have a very different mixing time distribution.
Imagine that the location of the sensor is also

the position of nutrient feeding. Those cells
circulating with the particular fluid element
receive nutrient only when the fluid returns to
that position. If the circulation time is longer that a
critical time, then the nutrient level seen by those
cells may fall below the critical value. A wide
distribution
of mixing time can be a concern.
Even though the frequency of exceedingly long
circulation time is low, nutrient starvation may
occur in those rare occasions and trigger apoptosis
or cause other irreparable damage to cells.
In reactor operation, the fraction of mixing
Population Density Distribution f(t)
time that is longer than the critical time should

be minimized by either improving mixing or
by setting a limit on the cell concentration.
Probability of Starvation
1
Circulation Time
Critical Circulation Time
Fig. 13.8: Mixing time distribution and critical mixing

time for nutrient depletion
Mixing Time Distribution

The mixing time in a tank is not uniform. If the average mixing time is 6 minutes, many fluid elements will
have a mixing time shorter than 6 minutes, and others
will have longer times.
If the critical mixing time is 6 minutes, the average
mixing time should be shorter than that.
Scaling Up and Mechanical Forces on Cells

In a turbulent flow, the fluids kinetic energy is
transferred by swirling pockets of fluid, called eddies.
Turbulent regimes are comprised of eddies of
different sizes, characterized by velocity fluctuations.
The hydrodynamic forces experienced by the cells

may arise from fluid-cell and cell-mechanical
parts interactions. Cells, being neutrally buoyant
particles, follow the motion of the relatively larger
eddies. In general, the direct impact of cells
on the impeller is minimal and cells generally
flow by impeller blades without suffering much
direct mechanical impact. However, these large
eddies cause the formation of cascades of smaller
eddies, which may impart damage to cells by
dissipating all of their energy on the cell surface.
The size of the eddy relative to the size of the cell
is thought to be an important factor that damages
cells. Smaller eddies in the size range of cell surface
motif, i.e., much smaller than cell diameter, are
considered to be more damaging than eddies that
are larger than cells. Studies examining cell death
Microeddies cause cell damage
Fig. 13.9: Eddies surrounding a cell suspending in a stirred

tank
Eddy size increases with scale if the power per unit

volume decreases with scale
caused by turbulent flow often assume that the

cell death rate is proportional to the Kolmogorov
eddy concentration, and cell damage occurs when
the eddy size is smaller than a critical eddy size.
As the scale increases, the eddy length increases.
Thus, strictly from the viewpoint of the average eddy
size, the effect of turbulence on cell damage will
not become more severe in scaling up. One should
be cautioned that the eddy size is not uniform, but
distributed over a range. As the scale increases, the
distribution function of the eddy size also varies.
Further, the discussion here does not take into
account the effect of gas mixing. The phenomenon
of mechanical stress caused by combined agitation
and aeration is rather complex in scaling up.
Scaling Up and Oxygen Transfer and Carbon Dioxide Removal

When scaling up a process, we aim to produce cells
and product in quantities proportional to the scale.
To meet that goal, we normally provide all nutrients
in a proportional amount, to meet the increased
metabolic needs of cells. For liquid nutrients,
increasing the nutrient provision in proportional to
cellular needs on a large scale is easily met. However,
for oxygen and CO2, which are supplied and removed
through gas aeration, scaling up presents a challenge
because of the constraints of physical factors.
Material Balance on Oxygen in

Bioreactor
Supplying oxygen to the growing cells in the

bioreactor involves blowing air through the culture
fluid, and transferring oxygen from the gas bubbles
into the culture fluid. The amount of oxygen carried
out from the reactor is less than that supplied
into the reactor. This difference is the amount
transferred into the liquid phase. Material balances
on oxygen can be performed on the gas entering
and leaving the reactor (the gas phase balance),
as well as on oxygen being transferred from gas
bubbles to liquid (the liquid phase balance). The
oxygen transfer rates calculated from the gas phase
balance and from the liquid phase balance are equal.
The dynamics of the dissolved oxygen

concentration are described by the balance
between OTR and OUR at a quasi-steady state.
At steady state, OTR=OUR
(Eq. 9)
(Eq. 10)
Oxygen Balance on Gas Phase
On the gas side, the oxygen transferred from the

gas side to liquid side is reflected in the difference
of oxygen concentration between gas inlet and gas
outlet.
(Eq. 11)
Gas phase balance is performed by taking the

difference between the oxygen input rate at the
inlet and the oxygen output rate at the outlet.
That difference is the amount of oxygen that has
been transferred into the liquid. While oxygen is
transferred into the liquid phase, CO2 (produced by
cells in the medium) and water vapor are stripped
out of the culture broth, thus the volume flow rate
in the outlet may differ from the inlet. Assuming
ideal gas behavior (PV = nRT), the molar flow rate of
the oxygen at the inlet and the outlet is the total air
flow rate (PQ/RT) multiplied by the molar fractions
of oxygen at the inlet and the outlet YO and YO
respectively. The oxygen transferred into the
liquid is simply the difference between the molar
flow rates of oxygen in the inlet and the outlet.
2,in
2,out
Oxygen Balance on Liquid Phase
On the liquid side OTR and OUR is described by Eq. 11

since the rate of change of dissolved oxygen is small. In cell
culture process, the air flow rate can be considered to be
the same at the inlet and outlet. Overall, the relationship is
described as:
(Eq. 12)
(Note: P/RT converts volume flow rate Q to molar flow rate
using ideal gas law. P is the pressure of gas phase)
On the liquid side, the oxygen transfer rate (OTR)

is described by the overall mass transfer coefficient
(KLa) and the driving force (C*-C). For small
reactors, we assume that liquid is well mixed.
This assumes that the concentration measured at
the outlet is the same as the concentration in the
reactor. C* at the air outlet should be used for the
driving force calculation. For a large reactor, one
assumes that the gas phase behaves like a tubular
reactor (plug flow), and a logarithmic mean of
the driving force is used. We assume that a quasisteady state, i.e., the change in dissolved oxygen
(dC/dt), is very slow compared to the oxygen
consumption and rate of transfer. Thus, the oxygen
uptake rate (OUR) can be approximated by the OTR.
C* is the dissolved oxygen concentratoin in equilibrium

with the gas, which may differ in different parts of the
bioreactor.
For small scale bioreactors, one can assume both liquid
phase and gas phase are well mixed. The gas phase in the
reactor is thus the same as that in the exit gas stream.
Thus, C* is related to the oxygen concentration at the exhaust gas by Henrys law constant:
(Eq. 13)
For large scale bioreactors, the inlet and outlet oxygen concentrations may be very different. One uses the logrithmic
mean driving force described below:
(Eq. 14)
Aeration Rate and Superficial Gas

Velocity
D = tank diameter
Q = aeration rate
vl = culture volume in
reactor
Vs = gas superficial velocity
A = cross-sectional area of
the tank
H = heights of culture
volume
Superficial gas velocity = gas flow rate/reactor cross

sectional area

vs = Q/A = Q/(Dt2) (Eq. 15)
The reactor volume increases with length scale raised to
the third power, while the cross sectional area increases
only with the second power.

V1 / V2 = D13/ D23 A1 / A2 = D12/ D22
Now we examine the gas phase balance. In scaling

up, if we keep Q/V constant, we will be able to
maintain the same oxygen level at the inlet and
outlet (Yin, Yout) and sustain the same oxygen transfer
rate (OTR). However, when scaling up, Q/V is likely
to decrease. If OUR is to be sustained, then Yout
has to be smaller to maintain the material balance.
Then, consider the liquid phase balance. OTR is KLa

multiplied by (C*-C). Because oxygen level at outlet,
Yout, is lower, C* is also lower. How can OTR be kept
at the same level as in the small scale? One possibility
is to increase KLa while keeping C at the same level
as in the small scale (thus allowing (C*-C) to be
smaller). The will require an increase in agitation
power. The other possibility is to increase (C*-C) to
the same level as in the small scale by allowing C to
be maintained at a lower level, if the reduced C has no
adverse effect on culture performance. Alternatively
one can use oxygen enriched air to increase C* to
maintain (C*-C) at the same level as in the small
scale. However, this will lead to increased level of CO2
accumulation as will be discussed in the next section.
Therefore, overall oxygen transfer becomes a
challenge in scaling up because the aeration rate
cannot be increased proportionally with the scale.

When scaling up one may choose to increase the air

flow rate proportional to the increasing reactor volume,

Q1 /Q2 = V1 / V2
One can see that

vs1 / vs2 = D1/ D2
(Eq. 16)
Superficial gas velocity will increase linearly with increasing scale.
Aeration Rate and Oxygen Transfer

Driving Force
Now we examine the gas phase balance. In scaling

up, if we keep Q/V constant and keep KLa constant
in the liquid phase, we will be able to maintain
the same oxygen level at the inlet and outlet (Yin,
Yout) and sustain the same oxygen transfer rate
(OTR). However, when scaling up, Q/V will actually
decrease so if OUR is to be sustained, then Yout
has to be smaller to maintain the material balance.
Then, consider the liquid phase balance. We have

kept KLa constant in this analysis. To keep OTR
also constant, the driving force must be sustained.
However, the driving force is the logarithmic
mean of the oxygen concentration at the inlet and
the outlet. With a lower level of Yout, the driving
force for oxygen transfer will actually be lower.
Therefore, overall oxygen transfer becomes a
challenge in scaling up because the aeration rate
cannot be increased proportionally with the scale.
When scaling up, we aim to maintain OUR and OTR at the same level.
Considering mass balance in the gas phase:
From Eq. 12:
Given that Q2/V2 is smaller, and Yin (oxygen concentration in the inlet air) is the same, Yout,2 in the
large scale will be smaller
From Eq. 13 and Eq. 14, since Yout,2 is smaller, so is C*.
Considering the liquid phase, OUR and OTR will be maintain at the same level in the two scales:

(KLa)1(C*1 C1) = (KLa)2(C*2 C2)
This can be accomplished by
Allowing C*2 to be lower while maintain C2 = C1. KLa2 in the large scale must increase.
Keep KLa2 at the same level as in small scale, then the concentration driving force of oxygen must
be increased to the level of the small scale by
Allowing C2 to decrease
Increasing C*2 by using enriched oxygen
Gas Flow Rate in Scaling Up: A

Summary
The physical constraint of a reduced cross-sectional

area to a reactor volume ratio with increasing
scale poses a challenge in oxygen transfer. In
scaling up, the airflow rate may not be increased
proportionally to the culture volume because an
overly high superficial air velocity is problematic.
Table 4. Effect of Scale on Oxygen Transfer

Reference
Scale
Constant
Air Flow
Scale (volume)
1,000
Constant
Superficial
velocity
1,000
Cross Sectional
Area
100
100
Air Flow Rate
1,000
100
Superficial Air
Velocity
10
O2 Consumption/
CO2 Production
1,000
1,000
Q(yin yout)
Comments
May reach
flooding
(yin yout) has

to be very
large, i.e. yout is
small
Need to
increase KLa or
power input
If air supply increases proportionally with scale, foaming

can become serious and flooding may occur.
In microbial fermentation, a high airflow rate

eventually causes flooding (i.e., the impeller is
swamped by gas bubbles) and loses its capacity for
pumping liquid. For cell culture, the aeration rate
used is substantially lower than the flooding aeration
rate. However, potentially a different problem may
arise. Antifoam agents are not used as extensively
in cell culture as in microbial fermentation because
of potential damage to cells. At a high superficial
velocity control of foaming may become problematic.
When scaling up, aeration rate is not increased

proportionally to the culture volume. Because less air
is given to the same volume of culture, more oxygen
has to be taken out from the gas phase to meet the
oxygen demand. This causes the oxygen level in the
gas phase to be lower. As a result the driving force
for oxygen transfer is also lower in the large scale.
In scaling up it is a common practice to take

a middle ground. The airflow rate per reactor
volume is decreased somewhat, but the superficial
velocity is allowed to increase (albeit less than
proportionally to the culture volume) to minimize
the loss of the oxygen transfer driving force. In
some cases, the air is enriched with oxygen, while
in other cases, the agitation rate is increased when
oxygen falls below a set point during the cultivation.
Effect of Scaling Up on CO2 Removal

The mass transfer coefficients for oxygen and carbon
dioxide are about the same.
R.Q. (moles of CO2 produced/ moles O2 consumed)
for mammalian cells is very close to 1.0. So, oxygen
uptake rate (OUR) and carbon dioxide evolution rate
(CER) are about equal.
The toxic level of CO2 is around 1520% (110 mm Hg
150 mm Hg).
The carbon dioxide removed from the reactor
(at steady state) is the difference between its
concentrations in the inlet and outlet gas:
(Eq. 17)
CER and CCO2, in are the same in reactors of different
scales
If cell concentration and metabolic activity remain
constant.
Q / V is smaller in large scale.
Inevitably CCO2, out will have to be higher on a large
scale.
The driving force for CO2 removal decreases with
increasing scale.
The respiratory quotient for most cells, using

glucose and glutamine as the main source of
energy, is very close to 1.0. Thus, every mole of
oxygen consumed by the cells generates about one
mole of CO2. A very active culture with a high cell
concentration can produce more than 100 mmole/L
of CO2 per day. In comparison, a cell culture medium
has about 20 - 40 mM of sodium bicarbonate.
The amount of CO2 produced by cells, thus, far
exceeds that which is added as buffer to the media.
Many cells, such as hepatocytes, are rather tolerant
to CO2 but others are more sensitive. The growth
of most cells may begin to be affected at a CO2
concentration of 15% (114 mm Hg) so continuous
removal of CO2 from the culture is important.
When scaling up, the airflow rate per reactor

volume may not increase proportionally. As a
consequence, less air is used to strip CO2 from an
equal volume of culture media. If the metabolic
activity of the culture remains the same, the same
amount of CO2 produced by the cells is now being
removed using less air. The CO2 level will then be
higher in the air exiting from the large-scale reactor.
The rate of CO2 stripping is dependent on the

concentration difference of CO2 in the liquid and the
gas phase. A higher concentration in the gas phase
diminishes the stripping efficiency. The increased
CO2 concentration in the exit air, thus, causes
further accumulation of CO2 in the liquid phase.
To compensate for the reduced O2 driving force
from scaling up, one can use O2-enriched air.
However, such a measure cannot compensate
for the diminished stripping efficiency of CO2.
Therefore CO2 accumulation patterns in large scale
and small scale bioreactors can be rather different.
To achieve the same molar transfer rates for oxygen

and carbon dioxide (although in opposite directions),
the driving force for carbon dioxide has to be around
100 mm Hg, a level similar to that for oxygen transfer.
The upper bound of CL for CO2 should be below the
inhibitory level of 110-150 mm Hg. If the driving force
is 100 mm Hg, then C* will have to be 10-50 mm Hg.
That is 1.5-6.5% CO2 in air.
To strip off carbon dioxide a sufficiently fast air flow
rate must be used to ensure the CO2 concentration in
the gas phase is low enough to provide a large driving
force.
(Eq. 18)
The overall mass transfer coefficient for CO2 is slightly

lower than that of oxygen because of its larger molecular
weight. However, the difference, approximately the
square root of their molar weight ratio, is very small.
One can consider the KLa to be about the same.
Unlike O2, the solubility of CO2 in aqueous solution is

very high. At the gas bubble interface, O2 in the liquid
phase can be assumed to be in equilibrium with the
gas phase, so (C*-C) is a good estimate of the driving
force. The same assumption is not always valid for CO2.
CO2 in the medium exists as CO2, HCO3, and CO3-2. At

the interface, CO2 crosses the film and is transferred
out of solution, but HCO3- does not. So, HCO3-must
dissociate to CO2 before being transferred to the
gas phase. The kinetics of HCO3-, the dominant
form of CO2 in aqueous solution, to dissociate
to CO2 is slow. Because of the slow kinetics, the
actual driving force is smaller than what can be
estimated from the total CO2 (g) concentration.
Chemical Environment in Scale

Translation
In scaling up or scaling down, the chemical

environment may also be affected. For cell culture
processes, many of those changes are caused by
different CO2 accumulations at different scales. CO2is
a major contributor of the pH buffer in a cellular
environment. CO2 produced by the cells is excreted
to maintain a physiological range of intracellular pH.
Carbon dioxide is transported through the plasma

membranes as CO2 and HCO3-. CO2 can diffuse
through cellular membranes, while HCO3 transport
is mediated by transporters. A symporter cotransports HCO3- and H+, while another antiporter
co-transports Cl- and HCO3- in the opposite direction.
Fig. 13.12: Schematic of the removal of carbon dioxide produced

by cells
As CO2 builds up in the culture fluid, a higher Clgradient is needed to drive HCO3- out, otherwise
the intracellular HCO3- level will be higher. Because
the airflow rate per reaction volume is not kept
constant in scaling up, the CO2 level in the culture
will be higher on a larger scale. This will affect
the intracellular CO2/HCO3- levels. However,
experimental evaluation of the effect of scale on
cellular level of CO2 and intracellular pH is still lacking.
CO2 produced by cells can diffuse through the cell

membrane
Most CO2 becomes HCO3- and is excreted through
transporters
Accumulation of CO2 in medium may affect
intracellular pH
Concluding Remarks
In scale translation, the relationship of geometryrelated physical parameters are not kept constant,
because the volume and the surface area of the
reactor change in different proportions relative
to length. Therefore, one has to define the critical
range of various scale-sensitive variables and
choose scaling up criteria to ensure the operation
is within the optimal region. In most cases, one
chooses not to scale up completely geometrically
similarly. Most large-scale reactors have a larger
height to diameter ratio than the smaller scale ones.
Nevertheless, the physical constraints on scaling up
are the same regardless of whether one scales up
geometrical similarly or not. In scaling up, the gas
flow rate is also likely to change in its proportion to
the reactor volume. This causes the mass transfer
characteristics to be different for different scales.
While the dissolved oxygen can be controlled at the
same level, the CO2 concentrations profiles are likely
different for reactors of different scales. Differences
in CO2,concentration in the reactor will cause pH
control actions, including base and CO2,addition and
CO2 stripping, to vary at different scales. Difference

in pH control actions may further change the
chemical environment of the culture. Given that the
physical and chemical parameters related to scaling
up cannot easily be manipulated or controlled, one
may resort to selecting cells which are less sensitive
to those parameters. Understanding scale-sensitive
parameters and a sound knowledge of estimating
the range of those critical parameters will greatly
facilitate the scale translation of cell culture
processes.
In process development involving scale translation,
one should aim to reproduce or to predict the
conditions of physical constraints, as well as the
resulting chemical environment. It may not be
possible to replicate all physical and chemical
parameters on drastically different scales.
Ultimately, one should identify and aim to control
critical physical parameters, to minimize variations
in the chemical environment.
Cell Culture Genomics

Gene and Genome. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
What is a Gene?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Organization of Genome . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Packaging DNA into Chromatin. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Epigenome . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Molecular Mechanisms that Mediate Epigenetic Regulation . . . . . . . . . . . . . . .
Genome Scale Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Transcriptome . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Proteome Profiling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Sequencing Technologies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Sanger Method and Sequencing Technology Evolution . . . . . . . . . . . . . . . . . . .
High Throughput NextGen Sequencing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Gene Expression Exploration in Cell Culture Processing . . . . . . . . . . . . . . . . . . . . . . . .
285
285
287
290
291
291
293
293
299
302
302
304
306
308
Gene and Genome

What is a Gene?
Mouse Genome Encodes

~30,000 genes coding for proteins
~1,600 genes coding for RNA
800 tRNA genes
350 tRNA genes
~450 other ncRNA (noncoding RNA genes)
A gene is a sequence of DNA that encodes for an RNA

and protein product. Up until a quarter century ago
the prevailing notion was that one linear sequence
of DNA directly encodes for one gene. After the
discovery of alternative splicing, however, our
understanding quickly changed. Recent findings
have highlighted a relatively large number of
alternatively splicing genes in the mouse genome.
A large number of these genes are translated into
different protein sequences. Our knowledge of the
relationship between a gene and its expression
product is evolving. For instance, we now know
that two genes may reside on opposite strands of
the same segment of DNA (and will be transcribed
in opposite directions), or they may reside in
the same strand of DNA and are overlapping..
A gene may encode for two types of end products: 1)
CELL CULTURE GENOMICS |285
Table 1. Distribution of Mouse Genes

Protein coding
genes
~30,000
Pseudogenes
~4,000
RNA coding genes
~1,000
rRNA
~800
tRNA
~350
snRNA
~150
miRNA
~300
rcRNA
~450
Gene Structure in Eukaryotes

Exons make up the mRNA; intervening sequences
called introns are also present
Splicing of pre-mRNA entails removal of intronic
regions, addition of polyA tail and 5 cap
Alternative splicing is common
a protein, or 2) a non-protein-coding RNA (ncRNA).

In its entirety, the mouse genome encodes for a
total of about 37,000 genes. Among them, just over
30,000 genes code for proteins, 4,000 code for socalled pseudogenes (genes whose identity can be
traced to an ancestor but are no longer translated
into a functional protein), and the remaining
1,600 code for RNAs (including 800 rRNAs, 350
tRNAs, and other non-coding RNA, or ncRNA).
The average mammalian gene spans a region
encompassing approximately 23 to 28 thousand
base pairs (kbp) in length. Typically, the promoter
region is located upstream of the transcribed region
(about 1 kbp or longer, sometimes as long as 10
kbp) and the transcribed region is used to generate
a primary transcript, or pre-mRNA. The primary
transcript starts with a 5 end untranslated region
(5 UTR) and ends with a 3 UTR. It also contains
a number of protein coding exons and introns.
After the introns are spliced out, the mRNA, or
mature transcript, remains and includes: 1) the
5UTR, 2) the exons, 3) the 3UTR, and 4) a newlyadded polyA tail. The mRNA is ultimately exported
out of the nucleus for translation in the cytoplasm.
Many primary transcripts undergo alternative
splicing. In these cases, portions of exons are
stitched together under different conditions (e.g., in
different tissues, at different times, or occurring at
different frequencies) to give different mature mRNA
species. The stitching point of two consecutive
exons may not be the same under alternative
splicing. In this case, splicing may result in a shift
of the reading frame, which would affect the exons
downstream of the splice junction and perhaps
give rise to completely different protein sequences.
The Functional Annotation of the Mammalian

genome (FANTOM) consortium has generated the
most complete mouse gene sequence database to
date. It has uncovered a large number of protein
coding genes with alternative splicing. These
events have been found to produce very different
protein sequences, a large number of polymerase
II transcribed ncRNAs (with polyA tail), and
antisense RNAs that may have regulatory roles.

The prevalence of so many classes of variants poses
ambiguity in the definition of a gene. For instance,
if one were to count each independent gene
product as a different gene, the total number of
genes could be estimated to be upwards of 50,000.
Organization of Eukaryotic Genes

23-28kbp
~1 kbp
AUG
intergenic region
Promotor
UGA
Exon1
Intron1
Exon2
Intron2
Exons
Transcription
5 UTR
intergenic region
3 UTR
UGA
AUG
primary transcript
Alternative splicing
Splicing
polyadenylation
5 UTR
AUG
AUG
UGA
UGA
3 UTR
AAAAAAAAAAAA
AUG
UGA
AAAAAAAAAAAA
start
stop
AUG
UGA
start
stop
AAAAAAAAA
Export to cytosol
Translation
protein 1
protein 2
Fig. 14.1: Expression of an eukaryotic gene. Alternative splicing into two different proteins are shown.
Organization of Genome
Composition of a Typical Mammalian Genome

Coding Sequences
Intronic Sequences
Repetitive Sequences
Other intergenic sequences
~ 1-2%
~ 20-25%
~45-55%
The number of genes in each organism varies greatly,

from 700 genes in a simple parasitic mycoplasma
to over 30,000 in mammals. Those studying the
minimum set of genes that are required for life have
derived a gene set ranging from 270 to 350 genes.
As the number of genes increases with increasing

complexity of organism, the additional genes
and gene products acquired tend to affect a cells
interaction with its environment (e.g., transporters,
signaling molecules, and their receptors). In
other words, data suggest that the increased
complexity of higher organisms requires the
formation of a more sophisticated communication,
both at the cellular and organism level.
With an increasing number of genes, one also sees

an increasing size of the genome. By convention, the
size of a genome is quantified using the number of
bases in a haploid genome. A bacteriums genome
Eukaryotic Genome is Organized into

Chromosomes
Human: 22 pairs; x,y
Mouse: 19 pairs; x,y
Rat: 20; x,y
Chinese hamster: 10 pairs; x, y
Mouse chromosome size: 61 Mbp (Chr 19)to 197
Mbp (Chr. 1)
Repetitive Sequence
Short RNA-derived interspersed elements (SINES,
90-300 bp)
Long interspersed nucleotide element (LINES,>500
bp)
Retrovirus like elements with long terminal repeats
(LTR)
DNA transposons
Widely distributed simple sequence repeats:
Direct repetitions of short k-mers such as (A)n,
(CA)n or (CGG)n
Segmental duplications, consisting of blocks of
10300bp of the genome
Blocks of tandem repeated sequences, such as at
centromeres, telomeres, the short arms of acrocentric chromosomes and ribosomal gene clusters,
these also include satellites and microsatellites.
is about 3 Mbp to 8 Mbp, while a funguss genome

is a bit less than one order of magnitude larger
than that of bacteria, from 12Mbp to 30Mbp.
In contrast, mammals have about a genome size

of 2-3 Gbp, nearly 1,000 fold higher than bacteria.
While the mammalian genome is nearly 1,000-fold
larger, it contains only 10 times more genes. A typical
protein-coding gene in bacteria is about 1 kbp, or
330 amino acids, while an equivalent sequence in
mammals is about 1.3 kbp. Including introns and
UTRs, a mammalian gene in total is 23-28 kbp,
which is substantially larger than a bacterial gene.
The gene-encoding regions, including introns,

account for only about 25% of the mammalian
genome. The rest of the genome consists of other
intergenic sequences, including promoters,
regulatory elements, and regions not yet explored
by scientific inquiries. Additionally, nearly 50%
of a mammalian genome consists of repetitive
sequences. That number is even higher in some
plants, giving them a genome size even larger than
mammals. The repetitive sequences reside not only
in intergenic regions, but also in introns, UTRs and
upstream or regions adjacent to genes. Repetitive
sequences fall into different categories; some are
short, while others are long, up to 500 bp. Some
are the result of transposons or the remnants
of retrovirus infection throughout evolution.
The presence of repetitive sequences presents a

barrier to the quick and accurate assembly of DNA
sequencing reads. In DNA sequencing, the DNA
molecule is first fragmented and then the independent
fragments are sequenced. The assembly algorithm
searches for the overlapping regions and attempts
to stitch them together into longer contiguous
sequences (or contigs). Therefore, a fragment
that has a repetitive end can often be assigned to
multiple loci as it is not a unique sequence. One
solution to this is to sequence very long fragments
of DNA, over the stretches of repetitive sequences.
Table 2. Genome Size of Representative Organisms

Common Name
Taxonomy
Species
Genome Size
Estimated Number of Genes
Mycoplasma
TENERICUTES
Mycoplasma pneumoniae FH
8.11E+05
670
Bacteria
PROTEOBACTERIA
Escherichia coli DH10B
4.69E+06
4,271
Bacteria
FIRMICUTES
Bacillus subtillis subtillis 168
4.21E+06
4,354
Yeast
FUNGI-ASCOMYCOTA
Saccharomyces cerevisiae
S288C
1.21E+07
6,273
Slime Mold
PROTISTS-MYCETOZOA
Dictyostelium discoideum AX4
3.40E+07
13,362
Roundworm
NEMATODES
Caenorhabditis elegans
1.00E+08
20,935
Fruit Fly
ARTHROPODA
Drosophila melanogaster
1.37E+08
21,116
Chicken
CHORDATA-BIRDS
Gallus gallus
1.00E+09
17,935
Frog
CHORDATA-AMPHIBIA
Xenopus tropicalis
1.70E+09
20,500
Human
CHORDATA-PRIMATES
Homo sapiens
3.17E+09
53,894
Mouse
CHORDATA-MAMMALS
Mus musculus C57BL/6J
2.72E+09
37,261
Rat
CHORDATA-MAMMALS
Rattus norvegicus BN/SsNHsdMCW
2.70E+09
35,427
Dog
CHORDATA-MAMMALS
Canis lupus familiaris
2.40E+09
24,661
Chinese Hamster
CHORDATA-MAMMALS
Cricetulus griseus
2.70E+09
32,476
Packaging DNA into Chromatin
DNA molecule
packed with
histone proteins
to form
chromatin
chromatin
forms fiber-like
structure
may condense tightly

or packed loosely in
different regions of
chromosomes
Fig. 14.2: Packing of a segment of DNA into highly condensed

region and relatively open region
Chromatin: complex of DNA and protein in which

genetic material is packaged within the cell
In bacteria, the genome is typically arranged into a

single, large chromosome. Most chromosomes are
circular, although linear chromosomes are also seen.
Conversely, in eukaryotes, the genome is segmented
into different chromosomes. Chromosomes are not
merely DNA molecules wrapped loosely together.
A typical E. coli cell (2 m x 1 m in size) has a
genome of about 1.5 mm in length. If all 46 diploid
chromosomes of the human genome were strung
together, they would form a ~2 m x 2 nM string. The
chromosome is, thus, not merely a string of DNA
packed haphazardly into the nucleus. It requires
extensive manipulation and the work of specialized
machinery to allow it to be packed into a dense volume,
while still remaining accessible for transcription.
Each chromosome is a molecule of double-stranded

DNA. Packaging of DNA occurs at multiple levels.
At the local level, small regions of a DNA molecule
form a complex with DNA binding proteins,
chiefly histones, to form a beads-on-a-stringlike structure. That form is further condensed
into packed beads, called nucleosomes. In further
condensation of the structure, some regions are
more open and accessible to transcription (this is
called euchromatin), while other regions are more
densely packed, and are called heterochromatin.
Histones: principal protein components of chromatin

Nucleosome: fundamental sub-unit of chromosome
which consists of 165 bp of DNA wrapped around an
octamer of core histones
Epigenome
The term epigenetics was introduced in the 1940s to
describe the interactions of genes with their environment,
which bring the phenotype into being.
Heritable changes in gene expression not encoded in
the DNA
Essential for genome packaging and fundamental to
development
Epigenetic alterations influenced by the environment;
for example, identical twins can be susceptible to
different diseases
Epigenomics: Representing the totality of epigenetic
marks in given cell type
Molecular Mechanisms that Mediate

Epigenetic Regulation
Although the genome of each cell of a higher mammal

may encode more than 30,000 genes, at a given time
it may actively transcribe only a fraction of these. A
typical CHO cell in culture, for instance, expresses only
about 16,000 genes. Some genes are ubiquitously
expressed in all tissues; others are tissue- or timedependent. At the transcriptional level, a large
array of transcription factors are responsible for
regulating the expression of genes according to the
tissue type, timing by developmental stage, or by
event (such as stress or exposure to some signaling
molecule). At a higher level, transcription is also
regulated by the accessibility of the gene loci, which
is further controlled through epigenetic events.
Epigenetic regulation does not result in a mutation, as
no change occurs to the underlying DNA sequence. It
does, however, cause a heritable change in the cellular
phenotype. Unlike a mutation, which originates in
a single genomic locus of a single cell, epigenetic
changes can occur and affect gene expression
on a global level. For instance, when stem cells
differentiate, or when fibroblasts are transformed
into iPS (induced pluripotent stem) cells, global
epigenetic changes occur on the chromosomes
to affect the reprogramming of genetic circuits.
In cell culture, cells are often adapted to new

culture conditions, such as differing growth factors or
adjusting to growth in suspension. In such processes,
the entire population of cells shift their phenotype.
Such processes are less likely to be mutation events
and are more likely to involve epigenetic regulation.
In the generation of producing cells, the host cell
transforms from a non-secretor to a professional
secretor in a short time, accompanied by a vast change
of cellular properties. Although it is possible that
mutations may be responsible for some of the changes,
it is very likely that epigenetic events are the major
Chromatin modification
Covalent modifications of histones
Histone variants
Nucleosome remodeling
DNA Methylation
Non-coding RNAs
Table 3. Types of Histone Modifications

Chromatin
Modifications
Residues
Modified
Functions Regulated
Acetylation
Lysine
Transcription,
Repair, Replication,
Condensation
Methylation
Lysine
Transcription, Repair
Methylation
Arginine
Transcription
Phosphorylation
Serine,
Threonine
Transcription, Repair,
Condensation
Ubiquitination
Lysine
Transcription
drivers to in the dramatic shift to high productivity.
The major mechanisms of epigenetic modification

involve
DNA
modifications
and
histone
modifications. Of the DNA modifications, the
methylation of Carbon 5 of cytosine is one of the
most common. Once methylation occurs, the mark
is highly stable and can be passed on to daughter
cells. Note that this change does not constitute a
change or mutation in the DNA sequence. Much of
the DNA methylation occurs on cytosines that reside
in CpG dinucleotides (its complementary strand, 3GpC -5, is also methylated). Regions of the genome
with a high GC content, where the CpG sequence
is very frequent, are often called CpG islands.
Such regions, when upstream of a promoter, can
play key roles in the regulation of gene expression.
Methylation of a CpG islands primarily leads to
gene silencing, as has been shown in the cholesterol
dependence of NS0 cells. For instance, Hsd17b7, a
key gene in cholesterol synthesis, is silenced by CpG
methylation in NS0 cells. Accordingly, demethylation
treatment of NS0 cells led to the rapid emergence of
cholesterol-independent cells. Methylation is also
likely to be involved in the glutamine dependence
of CHO cells, due to methylation of a CpG island
upstream of the glutamine synthetase gene.
Of note, methylation does not only occur in CpG
islands; nearly a quarter of methylation seen in
embryonic stem cells is not in a CG context. This
non-CG methylation, however, is more transient
and mostly disappears after differentiation.
At the histone level, acetylation, methylation,

phosphorylation, and ubiquitination may occur
on different amino acid residues of histone
proteins. Each histone protein has multiple
sites that may be modified, resulting in a large
combination of possible histone modifications,
each affecting the packing of DNA and the
accessibility of genes in the region. Both histone
and DNA chemical modifications require specific
enzyme-mediated reactions. Their maintenance
and removal also requires specific enzymes.
Knowing the mechanisms of epigenetic regulation,

it is possible to intervene using chemical inhibitors
of the necessary enzymatic reactions. For example,
5-Azacytidine is used to facilitate demethylation.
TrichostabinA and sodium butyrate are also used to
interfere with histone modifications. The alternation
of the epigenetic status can also be induced by the
introduction of exogenous genes, as in the process of
reprogramming induced in the derivation of iPS cells.
reprogramming induced in the derivation of iPS cells.
Genome Scale Analysis

Transcriptome
Mouse Genome Encodes
~15,000 genes are expressed in a given cell
Highly abundant genes generally dont change
transcript level over a wide range
Rare genes can be very dynamic
1,000 fold change in transcript level in 30 min
is common in bacteria. A similar change in
differentiating stem cells usually takes days.
Table 4. mRNA in a Typical Somatic Human Cell
Superprevalent
(Abundant)
Intermediate
Complex (rare)
Number of Species
% of mRNA by
mass
10 - 15
10 - 20
1,000 - 2,000
40 - 50
15,000 - 20,000
40 - 45
At a given time a typical mammalian cell transcribes

about 15,000 genes into RNA. The vast majority
of those transcripts are present at only very low
levels, with some even as low as a few copies in
each cell, with another small fraction expressed at
intermediate levels. Only a very small number of
genes are expressed at extremely high levels. Genes in
the last class, the so-called abundant genes, encode
proteins such as ribosomes, GAPDH (3-phosphoglyceraldehyde
dehydrogenase),
and
actin.
In some recombinant cells, the product gene, which

is highly amplified, also falls into this category.
Because the sheer number of transcripts for each
abundant gene is very high, the total mass of those
RNAs can constitute up to ~10% of all mRNA.
Such genes usually do not undergo a very large
degree of change in their expression level. For
instance, one rarely sees even two-fold changes in
the transcript level of most abundant genes under
different culture conditions. However, keep in mind
that, because they are so abundant, even a 10%
change in the level of a gene in this category is much
greater than even a 10-fold change in a rare gene.
The genes that most frequently undergo very
large changes in expression are the rare genes.
These rare genes often encode for products
that are gene regulators or other products that
are powerful even at minute levels. For this
reason, they are kept at very low expression
levels and are not expressed when not needed.
Expressed Sequence Tags (ESTs)

Transcripts from cells of tissues are isolated, reverse
transcribed to cDNA and cloned into E. Coli to
construct an EST library
Clones of E. Coli are sequenced to give rise to fragment
or the entire length of transcript
Sequences are assembled and annotated
The data gives transcriptome profiles of (abundance
level) of transcripts of different genes under different
conditions
The sequence data are used to design microarray and
assemble the genome
The transcript levels of many genes are relatively stable

at different times and under different conditions,
while some are relatively dynamic. Overall, the rate
of change of gene expression in mammalian cells is
rather slow compared to bacteria. We see over three
orders of magnitude decrease in transcript levels
within half an hour in bacteria; however, even under
stem cell differentiation conditions, a similar level of
change in mammalian cells usually occurs over days.
To explore the dynamics of gene expression in
different tissues and in different diseases or
differentiation stages, transcripts were isolated
from those tissues and directly sequences. Those
transcripts are typically called expressed sequence
tags (ESTs). The collection of those ESTs form the core
of the database of various genes in different species.
Capturing the dynamics of transcripts at a global level,

i.e., on a genome scale, has become readily available in
the past decade through the use of DNA microarrays
and, more recently, through deep sequencing.
Transcriptome profiling through arrays and
sequencing remains the cellular analytical tools that
are truly global and capable of genome-wide survey.
DNA Microarrays
Fig. 14.3: Classical DNA microarray prepared from cDNA clones

of an EST library and the use of it as a two-dye microarray
Table 5. Available Microarray Technologies

Affymetrix
Agilent
Nimblegen
Manufacturing
technology
Photolithographic
manufacturing
Noncontact
inkjet
printing
Maskless
synthesis
using digital
micromirro
device
Probe Length
Feature size
Multiplexing
25 - mer
5 - 18 m
No
60 - mer
65 m
2-, 4-, and
8-plex
60 - mer
16 m
2- and 4 - plex
Two prevailing forms of microarrays are currently

utilized: cDNA microarrays (commonly used for
complex mammalian genomes) and oligonucleotidebased microarrays. The difference between these
methods lies in their type of probes. In the case of
DNA microarrays for microbial species the primers
are designed to specifically amplify gene fragments
from genomic DNA that will then serve as probes.
These probes are designed specifically for unique
segments of gene sequences. The probes spotted on a
cDNA microarray for mammals, on the other hand, are
amplified from cloned cDNAs using universal vector
primers. Usually designing probes based on specific
gene sequences is too costly with the large number
of genes involved. They usually lack the specificity to
differentiate many isoforms or alternatively-spliced
variants.
Oligonucleotide-based
microarrays,
in contrast, utilize much shorter (20 - 80 bp),
specifically designed, then chemically-synthesized
probes. Multiple probes covering different regions
of each gene are often used to increase the specificity.
With the decreasing cost of oligoDNA microarrays

and direct sequencing, cDNA microarrays
are being phased out. cDNA arrays rely on
using two fluorescence channels for relative
measurement renders them inconvenient for
comparison of a large number of samples.
Long oligo microarrays are typically comprised

of 50 - 70 bp probes synthesized onto a glass
slide. Affymetrix arrays are made by the direct
synthesis of eleven sets of short 25-mer probes
onto the chip through photolithography-based
technology. Typically, multiple probes are employed
for a given gene or contig over a region of a few
hundred base pairs of each target transcript.
The photolithographic in situ synthesis technique
requires the construction of masks for each layer
of nucleotides added to the probes. The process
is extremely costly. In contrast, using a digital
micromirror, NimbleGen technology directs light to
tiny spots to allow chemical reactions to occur only in
the lighted spots without using masks, thus drastically
reducing the manufacturing cost of making the array.

light source
mask (each layer

four masks for A,
T, G, C)
array layer by layer synthesis (total ~25

layer for 25-mers of DNA)
Fig. 14.4: Mask-based layer-by-layer in-situ photosynthesis
of oligonucleotide on array surface (Affymetrix system).
digital mirror
light
source
Both Agilent and NimbleGen allow multiplexing (i.e.,

multiple independent samples can be hybridized
to separate arrays on a single slide). Both of these
array formats also support a dual mode system
that provides the option of using the routine twocolor experimental design (Cy3 / Cy5 based)
or one-color (Cy3 only) on a single platform.
When the DNA array is used for two-channel

comparison, usually a common reference is used.
Such common references are usually acquired
by mixing mRNA samples from different tissues
under different culture conditions to ensure
that the vast majority of transcripts are present.
When the DNA array is used for two-channel
comparison, usually a common reference is used.
Such common references are usually acquired
by mixing mRNA samples from different tissues
under different culture conditions to ensure
that the vast majority of transcripts are present.
layer by
layer
synthesis
~60-mers of
DNA
no mask needed
Fig. 14.5: Digital mirror-based photosynthesis of oligonucleotides on array surface (NimbleGen system).
RNA-seq
Direct counting of abundance level of reads for
each gene, normalized to gene length
Very wide dynamic range of detection
Quantification not affected by low sensitivity or
saturation as in fluorescence detection
Does not require sequence information for
With the ability to generate tens of gigabases

in a single run, high throughput sequencing
technologies are becoming an affordable and
powerful tool for transcriptome profiling.
A first step in transcript profiling is the removal of

rRNA by oligo(dT) capture of mRNAs, which targets
their polyA. Subsequent reverse transcription
for cDNA synthesis is performed either by polyAbased priming or by using random primers.
RNA-seq methods take mRNA samples, shears them
into fragments, and reverse transcribes them into

cDNAs, which are then sequenced using one of the
high throughput sequencing technologies. The
resulting output is a large number of sequences on
the order of 50 - 150 bp. Each string of sequence
is called a read. Depending on the depth of
sequencing and the abundance level of the nucleic
acid fragment, a segment of nucleic acid may be
sequenced only a couple times or up to million times.
RNA-seq is not 3 end biased; the abundance level of

a transcript is represented by the number of times
that segments of the transcript have been sequenced.
The more abundant and the longer the transcript is,
the more frequently its sequence reads will appear.
Using this method, one can sequence as deeply

as is necessary to detect almost all transcripts in
the cells, even those that are rare. In microarray
analysis rare transcripts usually do not yield enough
signal to give confidence to results. Furthermore,
very high abundant genes are usually detected in
the non-linear, near-saturation range of signal,
thus lacking good sensitivity for quantification.
Such problems are not present when RNA-seq
is used, as it gives a much wider dynamic range.
Another major advantage of RNA-seq is that

it requires no prior EST database or genome
sequences of the species to be probed, whereas for
DNA array probe design, at least the sequences of
the genes to be probed must be available. For the
species whose genome sequence or EST database
is available, the reads from sequencing are mapped
to the exon sequences for enumeration of hit reads
and for normalization to sequence length. Even if
no genome sequence or EST database is available,
the reads can be still assembled into contigs. In
most such cases, the contig annotation can be
obtained from a related species, and the reads can
be mapped back to each contig and then counted
to give an estimate of transcript abundance.
Table 6. Commonly Used DNA Microarray Formats

Microarray
Type
cDNA
oligoDNA
Probe sequence
generation
Common Uses
Sample Numbers
Hybridization
method
Quantification
Method
Universal primer
amplified EST probes
Commonly used
for mammalian EST
library based array
Sequence specific primer

amplified probes
Generally used for

microbial species
whose genome has
been sequenced
Normally
two channel
comparisons (can
go up to four)
Typically mRNAs
are isolated and
reverse transcribed
to cDNA which is
the labeled with a
fluorescent dye
Synthetic specific
50-60mer, can have
multiple probes per gene
Cross hybridization
among different
closely related
sequences can be
minimized
Measure the ratio

of intensity of the
same transcript from
different samples.
Comparison of
multiple specimens
from different
samples must
use ratio of ratios
unless a common
reference for all
samples is used.
Can use a pooled
RNA as a common
reference to facilitate
the comparison of
multiple samples.
Direct comparison
of levels of different
transcript in the
same specimen is
difficult.
Two Channel
Single Channel
RNA-Seq
Photolithographically
synthesized 25mers,
multiple probe set per
sequence
Interrogate
multiple segments
of about 500 bp
region using both
perfect match
and mismatch
probes to compute
the signal
Single Channel
mRNA is reverse
transcribed, then
transcribed into
cRNA, which is
then fragmented
and biotin labeled
before hybridizing
to the probes on
the array.
The intensity gives

an estimate of
the abundance of
transcript for each
gene. Data can be
used for both intraarray comparison
(different genes in
the same specimen)
as well as inter-array
comparison (ratio of
expression level as in
cDNA array).
Direct sequencing.
Coverage depends on
sequencing depth. For
CHO, 20GB gives good
coverage of most genes
Use for reaching

depths sufficient
to detect rare
genes. Discerning
heterogeneity
in transcript or
genome. Also for
transcriptome
profiling of
unsequenced
genome
Single sample
per channel,
or multiplexing
barcoded sample
Short reads
are sufficient
for sequenced
genome. If
assembly is
required, long
reads are
preferred.
Direct counting of
sequence reads per
gene, normalized to
gene length.
Proteome Profiling
Proteome profiling is another means of surveying

global changes in gene expression. It is a powerful
complement to microarray transcriptome profiling,
serving to examine gene function directly at the protein
level. Proteomic analysis allows for the investigation
of the primary amino acid sequence, protein-protein
interactions, and post-translational modifications
on a large scale. In proteomics, mass spectrometry
has established itself as an indispensable tool.
2D Gel Electrophoresis
2D-gel electrophoresis allows for the simultaneous

analysis of a many protein molecules. In this
method, a complex protein sample is resolved in
two dimensions according to charge (isoelectric
point) and mass on a polyacrylamide gel, followed
by staining. The stained protein spots are then
characterized using sophisticated image analysis
tools (such as PDQuest).
The difference in
staining intensity of observed spots allows for
relative quantification between protein samples.
Figure 14.6: Typical flow of 2-D electrophoresis-based protemics.
2D gel electrophoresis
First Dimension: Isoelectric focusing (IEF),
separation by charge. IPG (immobilized pH gradient)
strips; usually pH 4 to 7
Second Dimension: SDS-PAGE, separation by
molecular weight
Staining types: Coomassie Blue (sensitivity in the ug
range); silver staining (sensitivity in the ng range);
SYPRO RUBY(sensitivity in the pg range; fluorescent
dye; most suitable for quantification)
Image Analysis: identification of differentially
expressed spots by eye or with the aid of specific
software packages (PDQuest)
This method, however, is not suitable for some

proteins. For instance, proteins present at
lower levels are not easily detected. Also, some
proteins co-migrate and cannot be resolved.
Finally, proteins with charges outside of the
isoelectric range of the gel or highly hydrophobic
proteins are also unable to be resolved in the gel.
Once a protein spot of interest is identified, the

spot is excised and purified for further analysis,
such as direct sequencing or mass spectrometry
for protein identification. Electrospray ionization
(ESI) and matrix-assisted laser desorption/
ionization (MALDI) are two techniques commonly
used to volatize and ionize the proteins or
peptides for mass spectrometric analysis.
MALDI-TOF and ESI

Differentially expressed spots are extracted from 2D
gel, and subjected to proteolytic digest, and peptide
finger print analysis using MALDI-TOF
2D LC
Shotgun LC Based Methods

SILAC
Use non-radioactive isotope for labeling cultured
samples
Mix samples for isolation or enrichment of some
cellular fractions
Sample can be analyzed by PAGE or 2D LC
The fractions can be analyzed in mass-spec for
identification
iTRAQ
Use isobaric tag for different samples
2D LC
Protein mixture is subjected to proteolytic digestion
peptides are amenable to LC separation to reduce
complexity
Column types: First dimension often exchange, second
dimension reverse phase
Electrospray injection into mass spectrometer
For identification of molecules, not quantification
Fig. 14.7: iTRAQ labeling of proteolytic peptides for 2-D liquid

chromatography.
In applying two-dimensional liquid chromatography

(LC) for analysis of protein mixtures, a proteolysis of
the protein (typically using trypsin) is first performed.
This process reduces proteins to oligopeptides of
mostly 15 - 30 amino acids long, which facilitates
separation by liquid chromatography. In the first
dimension, the peptides are separated into fractions.
Each of the selected fractions is then injected into
the second LC for further separation before injection
into the mass spectrometer (MS) for detection.
LC-LC/MS methodologies have the advantage
of being capable of analyzing complex protein
mixtures; however there was previously no way to
quantify different expression levels among samples
until recently, due to the application of stable-isotope
labeling to LC-LC/MS. This method makes use of the
fact that pairs of chemically-identical analytes with
different isotope compositions can be differentiated
in a mass spectrometer by their difference in mass-tocharge ratio. The ratio of signal intensities for the pair
accurately indicates an abundance ratio for the two
analytes. Two commonly used labeling techniques
are SILAC (Stable Isotope Labeling by Amino Acid
in Cell Culture), and iTRAQ (Isobaric Tagging
for Relative and Absolute Protein Quantitation).
Although all proteins in a sample are, technically,

included in the analysis using non-gel-based
techniques, these methods are still not a true global
surveying tool. In actuality, only a few hundred to a
few thousand proteins are identified in the analysis
of a sample due to limitations in the resolving
power of liquid chromatography. The capture
of a peak from a sample is stochastic, and highly
abundant proteins are detected over those present
at lower levels. The isolation and identification
of low abundance proteins can be enhanced by
repetitively analyzing the same fractions in the mass
spectrometer and by excluding peptides that have
already been identified in the previous analysis. This
is very expensive and tedious so exhaustive surveys
of the proteome space are not commonly practiced.
Chemical isotope labeling of proteins in vitro

(after protein isolation) using isobaric Tags for
Relative and Absolute Quantification (iTRAQ) is a
highly versatile and widely-used method. iTRAQ
tags have a reporter group, a balancer group, and
a peptide reactive group that tags the N-terminus
of every peptide. One of the unique features of
iTRAQ is that up to eight samples can be labeled
with different tags and analyzed simultaneously,
while other methods can only label two samples.
ITRAQ
The combined balancer and reporter groups have

an identical mass of 145 for all four labeling pairs of
balancers (mass 28 to 31) and reporters (mass 114117). The MS spectra for each of the four labeled
samples look identical. The reporter-balancer
fragment stays intact, giving rise to the same
m/z ratio. This allows for protein identification
using the combined signal of the four samples.
Upon MS/MS fragmentation, the bond between
the reporter and balancer group is broken. The
reporter groups then appear as peaks in the low
mass region, and quantification of the peak area for
each reporter group gives the relative abundance
for a peptide between the labeled samples.
Fig. 14.8: Isotopic labeling of peptides.
SILAC
Fig. 14.9: Stable isotope labeling of amino acids in cell culture

(SILAC) for proteomics.
In vivo labeling methods, such as Stable Isotope Labeling with Amino acids in Cell culture (SILAC), use
deuterated leucine, or other isotope labeled amino
acids, to differentially label one of the protein samples by replacing an amino acid in the cell culture medium, thereby allowing the isotope to be incorporated into the cellular proteins. The combined labeled
and unlabeled samples are then analyzed by LC-LC/
MS, generating two spectra for each peptide fragment, one shifted precisely by the mass of the deuterated amino acid. Differences in peak height provide the means of quantification between samples.
This method allows two samples to be mixed prior
to protein isolation, thereby eliminating systematic
errors due to protein isolation efficiencies. It is
well suited for use when subcellular enrichment
protocols are used (as in the case of organelle
fractionation) prior to LC-LC/MS analysis. One
notable limitation of this method is that the labeling

time required is relatively long, to allow for sufficient
incorporation of the isotope into cellular proteins.
Sequencing Technologies
Sanger Method and Sequencing
Technology Evolution
Table 7. Summary of Sequencing Methodologies

Technology
Read Characteristics
Applications
Sanger Sequencing
(ABI)
500-1000 bp 384
reads/run
Gold standard all

purpose sequencing
Roche Sequencer
FLX/454
400 bp/read
1-7 Gbp/run
Re-sequencing,
expression profiling,
SNP/methylation
Illumina/Solexa
36-150 bp/read
20 Gbp/run
Re-sequencing,
SNP/methylation
SOLiD (ABI)
30-50 bp/read
20 Gbp/run
Re-sequencing,
SNP/methylation
Helicos
35 bp/read
20 Gbp/run
Re-sequencing,
SNP/methylation
Target DNA template is used for DNA synthesis

Fluorescently labeled nucleotide analogues
(dideoxynucleotides) is added to the synthesis
reaction; wherever an analogue is incoporated into the
elongating DNA, the reaction terminates
Each of the four dideoxynucleotide chain terminators
is labelled with a different florescent dye.
The newly synthesized and labeled DNA fragments
are separated by size (with a resolution of just one
nucleotide)
DNA sequencing technology has undergone

revolutionary changes in the past few years.
Traditional Sanger sequencing has dominated
the field for nearly three decades, and it is still
the prevailing method used for sequencing small
regions of DNA. For large-scale, or genome-wide
sequencing, a number of high throughput methods
have rapidly changed the scope of sequencing.
They are now used for genome sequencing,
transcriptome profiling, transcription initiation
site surveys, transcription factor binding site
profiling, and epigenetic alteration studies.
Sanger sequencing gives relatively long reads,

while newer methods give shorter reads. Some
of the reads are too short for efficient assembly
and are used primarily for resequencing, i.e., for
sequencing the genome of individuals of a species
whose genome sequence is already available. Other
newer methods, such as 454 and Illumina, produce
reads that are at least long enough for assembly.
All of the new methods and the Sanger method share

the same reading scheme. Each time a nucleotide
is incorporated into an elongating DNA molecule a
signal is emitted. Two approaches are adopted to
detect the emitted signal, (a) amplifying the signal
by having many DNA molecules emitting the same
signal, (b) using very sensitive detection methods to
detect even the signal emitted from a sigle molecule.
In Sanger sequencing, each target DNA molecule is

first cloned into E. coli so that a sufficient amount
of pure DNA molecules can be obtained by growing
the E. coli clone. Those DNA molecules are then used
as templates for DNA synthesis. By using altered
nucleotide analogues that cannot be used in DNA
synthesis, the elongation stops randomly as soon as
an analogue (instead of the genuine nucleotide) is
incorporated. Given a proper titration of the ratio
of analogues to normal nucleotides, there will be a

DNA molecule terminated at every base of the DNA
template. After synthesis, the mixture is elongated
and the terminated molecules are separated by
chromatographic separation at a single base
resolution. Since each analogue is labeled with a
fluorescent color, the colorimetric chromatogram
is finally used to read off the sequence.
This clone handling is very expensive and

time consuming.
Furthermore, single-base
resolution can be accomplished only up to at
most 1.2 kbp, even with capillary electrophoresis.
Fig. 14.10: Conventional Sanger sequencing of DNA.
Sanger Sequencing
Enrich target DNA fragments by cloning into a E. coli
plasmid

- Specific primers at two ends of target DNA

Collect plasmids, use specific primer to start
DNA synthesis from one end
Add nucleotide analog, ddNTPs in addition to dNTPs at

low frequency in the reaction mixture
Incorporation of a ddNTP terminates DNA elongation.
Probabilistically a small fraction of elongating DNA
is terminated at every base, resulting a collection of
synthesized DNA fragments of different length
Separate those DNA molecules using capillary
electrophoresis
Each of the four ddNTPs fluoresce at a different
wavelength. As each DNA fragments comes out of
electrophoresis, the fluorescence is read to give the
identity of the base
The most fundamental change from the Sanger
High Throughput NextGen Sequencing method to newer methods is the omission of E. coli
Next Generation Sequencing Technologies

Massively Parallel Local Amplification
Sequencing-by-Synthesis
clones. Two approaches were taken to bypass cloning.

In the first approach, localized PCR amplification is
performed: a single DNA fragment is amplified in
a droplet or in a locus on a surface. This allows a
large number of identical DNA fragments to be used
to generate signals for detection as they incorporate
nucleotides when used as templates for synthesis.
In the second approach, single molecule detection is
achieved. As long as each molecule is separated from
the others, the signal from nucleotide incorporation
into an elongating DNA can be detected. The
increased sensitivity eliminates the need for cloning.
The current prevailing methods, sometimes

referred to as next generation sequencing
technologies, fragment the DNA molecules
randomly without resorting to cloning. A key
feature of these methods is the massive and parallel
amplification of each DNA fragment individually,
thus creating up to millions of clusters, or
colonies, of DNA molecules of the same sequence.
With the 454 technology, each fragment is
immobilized on a bead contained in an oil /
water emulsion droplet, thus forcing all of
the PCR products of that particular fragment
to also be immobilized on the same bead.
Fig. 14.11: Two strategies of localized PCR amplification of DNA

molecules for enhanced signal detection.
On the other hand, the Illumina Solexa sequencing

technology immobilizes the fragment on a
solid surface and confines the PCR products of
the fragment in the locale, thereby forming a
cluster of identical sequences on the surface.
These ingenious approaches allow a large number

of identical sequences to be isolated so that the
light emitted from reactions in these clusters
can be detected using a single sensor, such as a
CCD camera. This contrasts with the traditional
Sanger method, which requires a sensor at
the end of each electrophoretic capillary, thus
drastically reducing the cost of sequencing.
Both methods then employ base-by-base synthesis.
The 454 technology works by sequential addition of
454 Sequencing Technology

Immobilize one DNA fragment per bead
Suspend each bead in an emulsion droplet
PCR amplify DNA in each bead
Place each bead in one well
Perform synchronized synthesis by adding one
nucleotide at a time
Each nucleotide incorporation emits a photon
Homo-oligomer emits stronger light intensity (works
only for short homo-oligomer sequences)
Can sequence up to 1000 bp in length
Illumina Sequencing Technology
Immobilize DNA fragments on surface with adaptor
Localize PCR reaction to create clusters of DNA
fragments with complementary strands
Perform synchronized DNA synthesis from one strand,
then on the other strand
Upon the extension of one base, the reaction is
terminated because the functional group for further
extension of the added nucleotide is protected; this
prevents incorporation of more than one nucleotide
when homo-oligomer sequences are encountered
Can sequence up to over 100 bp; very high throughput
one of the four nucleotides and photons are emitted

upon the incorporation of a base. The Illumina
Solexa methodology employs fluorescently-labeled
nucleotides that are also reversible terminators.
One base is incorporated and interrogated at
a time, since further elongation of the chain is
prevented. When all clusters are scanned at the
end of a cycle and the base has been determined
for each colony, the fluorophores are cleaved
off and terminating bases are activated, thereby
allowing another round of nucleotide incorporation.
The drawback of the current generation of these

new sequencing technologies is their reliance on the
complete synchronization of serial reactions. The
reads obtained are relatively short, compared to those
from Sanger sequencing. However, this drawback
is more than compensated by their improved
capability of massively paralleled reading of
millions of sequences at high speeds and at relatively
low costs. Their tremendous parallel sequencing
abilities allow for up to a million fragments of a very
abundant mRNA species (such as the transcript for
the recombinant protein product) to be read in a
single run, while only single reads can be obtained
for thousands of rare genes. Such a dynamic range
in sequencing outputs has made them great tools
for assessing transcriptome-wide abundance levels.
Gene Expression Exploration in Cell Culture Processing

Transcriptome analysis is a powerful tool for
studying cell culture processes, as it is extremely
revealing for both subtle and not-so-subtle
changes that can occur in many situations (e.g.,
cell cultivation, the variation of cell behavior over
time, and cell line development). It has been used
to compare cell performance in different reactor
scales, cells of varying productivities, and cells
growing under different media composition, and
to probe cell changes after long-perfusion culture.
In a striking example, a cell line was subjected to

studies under differing culture conditions. The
variables included temperature, pH, and reactor
size. There was also an uncontrolled variable. Two
different lots of the same hydrolysate were used in
each run, as is customary (when one lot of material
is exhausted, another lot is used). The time-course
transcriptome data from all runs were collected and
subjected to clustering. All the runs with the same lot
of hydrolysate were found to be clustered together,
thus overriding the other variables of the experiment.
Fig. 14.12: Comparative transcriptome dynamics between

(a) mouse hybridoma cells treated with butyrate and (b)
differentiating mouse stem cells (MAPC). MAPC cells were
grown on liver lineage differentiation medium for 6 days.
Mouse hybridoma cells were treated with 1mM butyrate for 27
h. Both MAPC and hybridoma cell samples were referenced to
a corresponding, untreated time sample. The transcript levels
were probed with the Affymetrix MOE430A array.
For each dataset, average intensities are plotted along the y-axis,
and the log2 of the expression ratios are plotted on the x-axis.
Each marker represents a gene on the array. The vertical lines
mark the bounds of a two-fold expression change. Markers
lying outside of these lines are more than two-fold up or down
regulated between the two samples compared. In comparing
Figures 2a and b, the number of genes that are more than two
fold differentially expressed is substantially higher for the stem cell
differentiation study than for the butyrate treated hybridoma cells.
Such revelation is possible because of the large

number of genes probed. Many subtle changes which,
individually, may not be used to draw a confident
conclusion, can collectively point to a trend or
correlation that is not otherwise easily discernible.
There has been some effort in comparing cells of
varying levels of productivity. It has now been
realized that the hyperproductivity trait is a
complex phenomenon that is not simply a result of
turning on a small number of master genes. Rather,
the generation of high producers is likely to involve
colossal changes in gene expression that occur in a
wide range of genes, but each only to a modest extent.
A most important feature of either sequencing- or
microarray-based transcriptome analysis is its global
coverage of gene expression. This type of analysis
sometimes reveals unnoticed cellular processes that
may play a key role in some physiological events.
For example, it was found that the endocytosis and
secretory vesicle retrograde transport pathways
are upregulated in high recombinant proteinproducing cells. This led to the realization that
high producers have an increased capability
to recycle components of membrane vesicles
that carry proteins destined for secretion from
the Golgi apparatus to the cytosolic membrane.
One of the important features of transcript

profiling in cell culture processes is the small
number of genes differentially expressed and
the lower magnitude in the observed foldchanges, as compared to those observed in other
biological processes. For microbial cells, it is not
unusual to observe more than 20% of the genes
as differentially expressed by more than two-fold
within ten minutes of changing culture conditions.
For cultured mammalian cells, it is rare to see over
two-fold expression changes occurring in more
than a small percentage of the genes surveyed.
Fig. 14.13: Intracellular processes differentially expressed

between high and low producing NS0 cell lines. (a) Each node
depicts an intracellular process with a large number of differentially expressed genes. (b) Schematic of the steps involved in
vesicle-mediated transport (nodes 4, 5, and 6).
It is important to realize that the cells we deal with

are cultivated in artificial conditions, not in their
native niche. For an organism the development
of a fertilized egg to an adult body incurs many
major events and gene expression changes to
guide those events. Once the cell is guided to its
destined developmental status, the perturbations
of other environmental factors are relatively
minor in magnitude by comparison. The various
events that cells in culture may encounter are
only relatively small perturbations compared to
developmental or differentiation events that their
genome has been evolved to accommodate. It is
not surprising that colossal big changes in gene
expression are rarely seen in cell culture bioprocess
but are frequently seen in stem cell differentiation.
The small change in gene expression, as compared

to other organisms and to the in vivo processes,
requires one to be more versatile and more
skillful in data analysis. However, the power of
global transcriptome analysis will fundamentally
change our practice in cell culture processing.
nor perturbations in comparison to development.
The small degree of gene expression changes, as
compared to other organisms and to the in vivo processes, requires one to be more versatile and more
skillful in data analysis. However, the power of global transcriptome analysis will fundamentally change
our practice in cell culture processing.
Concluding Remarks
In the past few years we have seen dramatic advances
in genome science and technology. The availability
and affordability of sequencing technology has
changed sequencing effort from species sequencing
(i.e., focusing on obtaining a representative genome
sequence), to individual sequencing (i.e., aimed to
acquire genome sequence of an individual human,
organism or cell line). The depth of information
we are acquiring from genome, transcriptome,
proteome and epigenome, is transforming the
way we deal with bioprocess challenges.

The
application of -omics technology in cell culture
processing is still largely limited to process analysis.
One can foresee that, in a not too distant future,
-omics technology will be increasingly used for
the generation of producing organisms, as well
as in the design of biological processes. In some
ways, genomics science may also bring about
transformative changes in cell culture bioprocessing.
Index
Note: Page references followed by f or t indicate material in
figures or tables, respectively.
A
ABC transporter, 43, 44f, 84
Abraham, Edward Penley, 2
Abundant genes, 293, 293t
Acetylation of histones, 292293, 292t
Acetyl CoA
in cholesterol synthesis, 87
in lipid metabolism, 86, 86f
metabolism of, 81
in tricarboxylic acid cycle, 61, 73
Acetyl CoA shuttle, 86, 86f
Acoustic resonance enhanced settling, 257258, 258f
Actin fibers, 37, 3839, 38f, 46
Active transport, 4042, 41f
Activity parameters, 156, 160f
Adaptation, in stable cell lines, 131, 142, 142f, 143f
ADCC. See Antibody dependent cellular cytotoxicity
Adenosine, in medium, 109, 109t
Adenosine triphosphate (ATP)
in active transport, 40
consumption in glucose metabolism, 6061
mitochondrial production of, 29
production in glucose metabolism, 6064, 66, 70
Adenosine triphosphate synthase (ATP synthase), 61, 63
Adherent cultures, vs. suspension cultures, 202
Adhesion, cellular, 4546, 53, 142
microcarriers for, 203204, 204t, 205t
vs. suspension, 202
Aeration. See also Oxygen transfer, in bioreactors
stripping of carbon dioxide by, 219
surface, for oxygen supply, 220
Aerobic glycolysis, 6566
Affym etrix microarrays, 295296, 295t
Agarose cell immobilization, 209
Aggregation, 206, 206f, 209, 209f, 212
Agilent microarrays, 295t, 296
Agitation, in bioreactors, 265266
impellers for, 206, 265268, 265t, 266f
mechanism of, 265266
purpose of, 265
Airlift bioreactor, 207, 207f, 225
Akt, in regulation of glucose metabolism, 77, 77f
Alanine
in medium, 108
metabolism of, 8182
Albumin
production of, 13
recombinant, 122
secretion of, 31
serum, in medium, 118, 122
Alginate microcarriers, 205t

Allosteric regulation, differential, of glucose metabolism, 7576,
75f
Alternating tangential filtration, 260, 260f
Alternative splicing, 285286
Amino acid(s)
alterations, in bioprocessing, 1617, 16t
analysis of, 299302
in cellular growth, 150, 151, 151t
degradation of, 8182
essential, 81, 108, 108t, 151
in fedbatch culture, 238240, 239f, 239t, 246247
in intracellular fluid, 153, 153t
in medium, 81, 106108
metabolism of, 58, 8182, 81f, 82f, 186
non-essential, 81, 108, 108t, 151
stock solutions of, 108
transport of, 4041, 43, 81
Aminophospholipid translocases, 84
Ammonia, in cultured cell metabolism, 67, 236
Amphipathic lipids, 2223
Amplification
localized, in DNA sequencing, 304, 304f
for stable cell lines, 131132, 138141, 140f
Amplification maker, 135
Anaerobic glucose metabolism, 58, 65
Aneuploid cells, 54, 133, 143, 150
Animal(s)
as production vehicles, 199
transgenic, 12, 14, 14t
vaccines for, 8t, 9
Animal component-free serum, 102103
Animal serum, in medium, 102, 117119
Anterograde transport, 34, 35, 35f
Antibiotic(s)
declining price of, 23
development of and advances in, 24
in medium, 117, 117t
as selection markers, 137t
Antibody dependent cellular cytotoxicity (ADCC), 90
Antibody products, 58, 7t
manufacturing of, 1112
quantity for administration, 11, 11t
stoichiometric ratio in, 11
Antibody synthesis, analysis of, 186
Antioxidants, in medium, 116
Antiporters, 41, 42f
Apf-1, 52
Apoptosis, 5153
cell cycle and, 50f
death receptor pathway in, 5152
mitochondrial pathway of, 5253, 53f
morphological changes in, 51
vs. necrosis, 51
Apoptosome, 52
Ascorbic acid (vitamin C), in medium, 109
INDEX |309
Asparagine
in medium, 108
metabolism of, 81
Aspartate, metabolism of, 8182
ATP. See Adenosine triphosphate
ATP-binding cassette (ABC) transporter, 43, 44f, 84
Attenuated vaccines, 45
Autocatalytic growth, 156
Automation, for cell line production, 145146, 145f
Axial flow impellers, 206, 265266, 265t, 266f
Azacytidine, 293
B
Baby hamster kidney cells. See BHK cells
Bacterial genome, 287288, 289t, 290
Bag-based centrifuge, 256
Bag culture systems, 201202, 202f
Bak protein, 52
Balance equations, 162165, 175180. See also Material balance
Basal medium, 101
components of, 106117
optimal concentration of organic nutrients in, 110, 110f
Base addition, proportional feeding with, 244
Base-by-base synthesis, 304305, 304f
Batch culture (processes), 233. See also Fedbatch culture
in bioreactors, 196198, 197f
vs. continuous processes, 249251
Bax protein, 52
B cells, transition to plasma cells, 33, 33f, 130
Bcl-2 proteins, 52
Bcl-xL protein, 52
Beef extract, in medium, 124
Beta-carotene, in medium, 116
Beta cells, pancreatic, 31
BHK cells, 8t, 19t
adhesion of, 142
aggregates of, 209
as continuous cell line, 49
for transient expression, 129
veterinarian vaccines from, 8t
Bicarbonate buffer, 113114, 115f, 115t
Biogen, IDEC ISM facility, 12
Biological fluids, in medium, 118
Biologics, 127. See also specific types
Biomass, 147148, 148t
growth of. See also Growth
material balance on, 149154, 149f
as objective of medium, 9799, 125
overall synthesis equation for, 149150
intracellular fluid in, 153154
medium to generate, 103104
metabolic flux analysis of, 184185, 185f
as output of reaction, 176
on substrate, yield of, 158
Bioreactors, 191212. See also specific types
agitation in, 265266
impellers for, 265268, 265t, 266f

purpose of, 265
basic types of, 193196
batch processes in, 196198, 197f
cell culture, 206212
cell retention in, 249261. See also Perfusion culture
cell support systems in, 202206
continuous processes in, 196197
disposable or single-use systems in, 199202
fedbatch cultures in, 198, 198f, 212
mixing characteristics of, 193, 266270
mixing time in, 206, 268274
operating mode of, 196198
oxygen balance in, 220, 276280
oxygen supply for, 220228
by medium recirculation, 221f
by silicon tubing/membrane, 220
by sparging, 220228
by surface aeration, 220
oxygen transfer in, 206207, 213231
driving force for, 214216, 220, 279280
enhancing or improving, 216, 220, 221
experimental measurement of, 224225, 224f, 225f
hydrostatic pressure and, 225
in immobilization reactor, 229, 229f
mass transfer coefficient in, 216
objective of, 219
in plug-flow reactors, 229230, 230f
rate of, 215216, 220
scaling up and, 275280, 280f, 280t
surface/interfacial area and, 216
through gas-liquid interface, 212219, 213f, 214f
phases in, 196
power consumption in, 266270, 266f
reactions and reaction kinetics in, 194196
scale translation for, 263284
and carbon dioxide removal, 275, 281283
and chemical environment, 283, 283f
constant parameters in, 269270, 269t
dimensionless variables in, 267270
and driving force, 279280
effect of scale on physical behavior, 269270
geometric nonsimilarity in, 264
geometric similarity in, 264
major effects of scale, 264
and mechanical forces on cells, 274275, 275f
mixing time in, 271274, 271t
nutrient starvation time in, 271274, 271t
objective of, 264
and oxygen transfer, 275280, 280f, 280t
physical and mechanical parameters of, 264
Reynolds number in, 267268, 267f
and superficial gas velocity, 278, 278f
tracer concentration response in, 193195, 194f
velocity of, 268, 278, 278f
INDEX |310
volumetric flow rate in, 268270

Biosimilars, 3, 910
approval in Europe, 10, 10t
marketed in China, 11t
marketed in India, 10t
uncertainty about quality of, 10
Biotin, in medium, 109
BiP chaperone protein, 33, 35
Bispecies-transporters, 4142
Blasticidin S, as selection marker, 137t
Blood bags, 201202
Bok protein, 52
Bristol Myer Squibb, Devens (Massachusetts) facility, 12
Bubble size, in gas sparging, 222223
Buffer systems, in medium, 113115, 115f, 115t
Bulk ions, in medium, 111112, 111t
C
Calcium
intracellular and extracellular concentrations of, 100, 100t
Calcium phosphate precipitation, 133134, 134t
Calculated differentiated data, 169
Calculated integrated data, 169
Carbohydrates, as cellular component, 21, 148
Carbon
in amino acid metabolism, 8182
in cellular growth, 149150
flow or flux of, 67
in glucose metabolism, 59, 6167
in glutamine metabolism, 59
metabolic flux analysis of, 182f, 183
in pentose phosphate pathway, 6465
Carbon dioxide
in buffer system, 113114
cellular tolerance of, 281
concentration in medium, 217218
in glucose metabolism, 60, 61, 66
production of, 219
removal in bioreactors, 219
scaling up and, 275, 281283
transfer (diffusion) of, 212219
g-Carboxylation, 12, 14, 16
Carrier-mediated diffusion, 4042, 41f
Carrier proteins
in medium, 117, 118, 122
transport by, 122, 122t
Caspases, in apoptosis, 5153
Catabolism
of glucose, 60
of lipids, 36
in metabolic flux analysis, 188t
Catalase, in medium, 116
Catalytic macromolecular components, of medium, 105106
CDK4/6-cyclin D complex, 4849
CDK inhibitors (CDI), 4849
cDNA microarrays, 295, 295f, 298t

Cell(s)
chemical environment of, 2122, 21t, 97101
composition of, 2122, 21t, 148, 148t
in culture. See Cultured cells
death of, 5153
apoptosis vs. necrosis, 51
cell cycle and, 50f
consideration in growth rate, 157
from injury (necrosis), 51
diameter of, 21
mass of. See Biomass
movement of, 4647
nutritional requirements of, 9799. See also Medium
senescence of, 5355
size of, 21, 148, 148t
sources of, 1921, 19t
volume of, variation in, 150151, 150t, 151f
Cell adhesion, 4546, 53, 122123, 142
microcarriers for, 203204, 204t, 205t
vs. suspension, 202
Cell adhesion molecules, in medium, 122123, 123t
Cell aggregates, 206, 206f, 209, 209f, 212
CellCube Module, 201
Cell culture engineering. See also specific processes and
materials
advances and growth in, 14
process robustness in, 1417
Cell culture products, 412. See also specific products
alternative technologies for, 1214
biosimilar or follow-up biologics, 3, 910, 10t, 11t
in-process structural alterations to, 1517, 16t
from perfusion bioreactors, 249t
quality of, 1417
Cell cycle, 4749, 48f
and apoptosis, 50f
checkpoints in, 4749
cyclins and CDKs in, 4849
positive and negative cues in, 4748
Cell expansion. See also Growth; Growth control
medium for, 9799, 125
Cell lines
adaptation of cells in, 131, 142, 142f, 143f
amplification for, 131132, 138141, 140f, 141f
automation and high throughput technology for, 145146,
145f
basic steps for generating, 131133, 132f
continuous, 49, 143144
development of, 127146
gene expression in, 306308, 306f, 307f
genomics and, 146, 306308, 306f, 307f
host cells for, 127129
INDEX |311
hyper-producing, 129133, 130t, 131f

immortalization of, 54
industrial, 8t, 9
screening for, 131
selection of cells for, 131132
single-cell cloning for, 131133
sources of, 1920, 19t
stability of clones selected for, 143145
stability of product quality in, 144145
stable, 127, 129146
transfection for, 131137
veterinary, 8t, 9
vs. cell strains, 53f, 54
Cell membrane, 2227
composition of, 2225
dynamic nature of, 23, 26
homeostasis of, 26
lipid bilayer of, 2227, 83, 87
potentials across, 2526, 25t
protein content of, 25
transport across, 23, 2627, 3947
turnover rate of, 26
Cell pool, 132133
Cell recycling, 250251
Cell retention
methods of, 249261
in perfusion culture, 249261
Cell separation methods, 254261
Cell signaling, 4546
Cell strain, 53f, 54
Cell support systems, 202206
Cellulose, for microcarriers, 203204, 205t
Centrifugal cage, 259, 259f
Centrifugation, 256, 256f, 257f
Centritech Lab centrifuge, 256
Ceruloplasmin, in medium, 116
Channel-mediated diffusion, 4042
Chaperone proteins, 33, 35, 90
Checkpoints, in cell cycle, 4749
Chemical environment
of bioreactors, scale translation and, 283, 283f
of cells, 2122, 21t, 97101
Chemically defined medium, 102, 103
Chemical reaction systems
cellular system vs., 175176
material balance for, 175180
Chick egg, 4, 9, 19
China, biosimilars marketed in, 11t
Chinese hamster ovary (CHO) cells, 8t, 9, 19t, 20
adhesion of, 142
aggregates of, 209, 209f
doubling time of, 154t
genome of, 288, 289t
on microcarrier, 203f, 204
non-antibody products from, 6t
recombinant proteins from, 8t

for stable expression, 129130, 130t, 146
therapeutic antibody products from, 7t
volume of, 150t
Chloride ions
intracellular and extracellular concentrations of, 100, 100t
in medium, 104
transport of, 4345
Chlortetracycline, in medium, 117
CHO. See Chinese hamster ovary cells
Cholesterol
biosynthesis of, 8688, 88f
function of, 83, 8687
in lipid bilayer, 2425, 24f, 83, 87
structure of, 85, 85f
Choline, as backbone of phospholipid, 22
Chondroitin sulfate, in extracellular matrix, 46
Chromatin
DNA packaging into, 290, 290f
modifications of, 292293, 292t
Chromium, in medium, 112
Chromosome(s)
abnormalities, in cell line, 143144
in cultured cells, 5455
in mammalian genome, 288
Cis Golgi, 32, 33
Cisternae maturation model, 33
Citric acid cycle. See Tricarboxylic acid cycle
City water, contaminants in, 106
Clonal growth, 110
Clones
in Sanger sequencing, 302303
selected for cell lines, 131132, 143145
Cloning, single-cell, 131133
CMV promoter, 135
cMyc, in regulation of glucose metabolism, 77, 77f
Cobalamin, in medium, 109
Cobalt
in intracellular fluid, 153
in medium, 112
Codon optimiziation, 135
Collagen
in extracellular matrix, 4546
in medium, 123t
for microcarriers, 204, 204f, 205t
Combustion, 6163, 176178
Compartmentalization, 184
Complex glycans, 92, 92f
Complex medium, 102
Complex supplements, in medium, 117125
Concentration factor, in perfusion culture, 254
Concentration gradient
across cell membrane, 2122, 21t, 4042, 4345
across mitochondria, 2930, 30f
INDEX |312
Conditional promoters, 134135

Conical settler, 254255, 254f
Constitutive promoters, 134135
Contact inhibition, 46, 48, 53, 54f
Continuous cell lines, 49
Continuous processes, 249251
advantages of, 250
in bioreactors, 196197
with cell retention, 251. See also Perfusion culture
disadvantages of, 250
economics of, 250
flow rate vs. growth rate in, 250
Continuous stirred tank reactor (CSTR), 193196, 193f, 212
reaction and reaction kinetics in, 194196
Copper
in medium, 112
COS cells, for transient expression, 129
Co-transporters, 4142, 42f
Cow pox vaccine, 4, 199
CpG islands, 292
CRFK cells, 8t
Crisis, 54
Cross filtration model, 259, 259f
CSTR. See Continuous stirred tank reactor
Cultured cells
crisis of, 5455
epigenetic changes in, 291293
growth of. See Growth
Hayflicks phenomenon and, 5455
life span of, 5455
medium for. See Medium
metabolism of. See also Metabolism
glucose, 5870
overview of, 5859
passages of, 5355
stoichiometry and kinetics of, 147166
Cumulative data, 171
Cyclins, 4849
Cylinders on shakers, 201202
Cytidine, in medium, 109t
Cytochrome C
in apoptosis, 52
in electron transfer chain, 63
Cytokine(s)
in medium, 9798
quantity for administration, 11
Cytoplasm, 2736
Cytoskeleton, 27, 3739
Cytosol, 2729
acetyl CoA shuttle in, 86
lipid metabolism in, 83, 85, 86, 87
D
Data, types of, 169
Data analysis, 167174

Data processing, 167174
cell culture, 169171
fedbatch culture, 160, 160f
mapping data to pathways, 173, 173f. See also Metabolic flux
analysis
pipeline for, 168173
spreadsheets for, 168, 170171
standardized templates for, 168169
Data visualization, 172, 172f, 174f
Death, cellular
death of, 5153
apoptosis vs. necrosis, 51
cell cycle and, 50f
consideration in growth rate, 157
from injury (necrosis), 51
Death phase, of cell growth, 154155, 154f
Death rate, specific, 157
Death receptor pathway, 5152
Decline phase, of cell growth, 154155, 154f
Delivery of feed medium, 245247
Deoxyribonucleic acid. See DNA
Derived parameters, 160, 160f
Dextran
in medium, 116t
for microcarriers, 203204, 205t
DH82 cells, 8t
DHFR amplification system, 138141, 140f
Differential allosteric regulation, of glucose metabolism, 7576,
75f
Differentiated data, calculated, 169
Diffusion
in bioreactors, 212219, 213f, 214f
carrier-mediated, 4042, 41f
channel-mediated, 4042
facilitated, 4042
Dihydroxyacetone-phosphate (DHAP), 61
Dilution rate, in perfusion culture, 254, 254f
Dimensionless variables, in scale translation, 267270
Diploid cells, 5355, 143, 150
Diploid chromosomes, 290
Direct DNA sequencing, 295
Direct measurement, for fedbatch culture, 243
Disc centrifuge, 256, 256f
Disposable bag-based centrifuge, 256
Disposable cell culture systems, 199202
Dissociation, of cells from surface, 5354, 54f
Disulfide-bond formation, in protein therapeutics, 5, 14
DNA
as cellular component, 21
location in cell, 28
methylation of, 292293
mitochondrial, 2930
INDEX |313
modifications of, 291293

packaging into chromatin, 290, 290f
synthesis of, 28, 47
DNAcalcium phosphate Coprecipitation, 133134, 134t
DNA microarrays, 294297, 298t
cDNA, 295, 295f, 298t
layer-by-layer synthesis in, 295, 295t, 296f
oligonucleotide-based, 295296, 298t
for pathway-related data, 173
RNA-seq, 296297, 298t
for two-channel comparison, 296
DNA sequencing, 288289, 302305
in cell culture processing, 306308, 306f
direct, 295
evolution of technologies for, 302303
next generation technologies for, 304305
Sanger method of, 302303, 302t, 303f
Dog (MDCK) cells, 8t, 9, 1920, 19t
Dominant marker, 136
Doubling time, 154157, 154t
Driving force, for oxygen transfer, 214216, 220
scaling up and, 277, 279280
Dry mass, of cells, 148, 148t
E
E. coli. See Escherichia coli
E2F transcription factor, 49
ECM. See Extracellular matrix
Eddies, in bioreactors, 274275, 275f
EF-1. See Elongation factor 1
Electron transfer chain, 61, 6364, 66
Electroporation, 133134, 134t
Electrospray ionization (ESI), 299
Elongation factor 1 (EF-1), 135
Endocytosis, 2627, 36
Endoplasmic reticulum, 27, 31
expansion of, 33
glycan biosynthesis in, 93
glycosylation in, 90
lipid metabolism in, 83, 88f
protein secretion through, 3135, 35f, 83
rough, 31
smooth, 31
Endosomes, 27
Endothelial cells, volume of, 150t
Enzyme(s)
in glucose metabolism regulation, 7476, 75f
MichaelisMenton kinetics of, 41
Epigenetics
in cell culture, 291293
definition of, 291
inhibition of, 293
mechanisms of, 292293
molecular mechanisms mediating, 291293
Epigenome, 291293
Epigenomics, definition of, 291
Epithelial cells
movement of, 4647
use in bioprocessing, 1920, 19t
EPO. See Erythropoietin
ER. See Endoplasmic reticulum
Erythropoietin (EPO)
development of, 5
glycan and, 89
product quality and process robustness, 15
Escherichia coli
genome of, 28, 290
as host system, 12
in Sanger sequencing, 302303
ESI. See Electrospray ionization
Essential amino acids, 81, 108, 108t
ESTs. See Expressed sequence tags
Ethanol, transport of, 39
Euchromatin, 290
Eukaryotes
chromosomes of, 290, 290f
gene structure in, 286
genome of, 287288, 289t
European Union, biosimilar approval in, 10, 10t
Ex-Cyte, 124
Exocytosis, 3940
Exons, 286
Exponential phase, of cell growth, 151f, 154155, 154f
Expressed sequence tags (ESTs), 294
Extended fedbatch culture, 235, 235f
Extracellular fluid, 100101, 100t
Extracellular ion concentration, 2122, 21t, 100101, 100t
Extracellular matrix (ECM), 4546
functions of, 4546
Extracellular matrix extract, in medium, 123
F
Facilitated diffusion, 4042
F actin, 39
Factor VIII
in-process structural alterations to, 16
recombinant, 5
tissue-derived, 5
FADD. See Fas-associated death domain
FADH2, from glucose metabolism, 63, 68
FANTOM. See Functional Annotation of the Mammalian genome
Fas-associated death domain (FADD), 5152
Fatty acids
in lipid bilayer, 2324
in medium, 84
metabolism of, 8586
oxidation of, 36
saturated, 85
transport of, 39, 84, 122
FBS. See Fetal bovine serum
INDEX |314
Fedbatch culture, 233247

in bioreactors, 198, 198f, 212
control objective and criteria in, 241243
control strategies for, 241243
data processing and plotting of, 160, 160f
delivery of feed medium in, 245247
feeding parameters in, 245
feeding strategy for, 241245
direct measurement of nutrient consumption, 243
proportional with base addition, 244
proportional with oxygen uptake rate, 245
proportional with turbidity, 244
fortified feed and addition, 235, 235f
habitation-conducive components of, 104
for industrial production, 125
intermittent harvest and feed, 198, 198f, 234, 234f
lactate and production in, 59
medium for, 233, 237240
for consumed nutrients, 238240, 239f, 239t
for unconsumed components, 240
with metabolic state manipulation, 236237, 236f, 236t
metabolic stress in, 161, 161t
on-line estimation of stoichiometric feeding in, 246247
productivity and product quality of, 241243
for stable cell lines, 130
stoichiometric ratio in, 160f, 236240, 236f, 236t, 239f, 239t,
246247
types of, 234237
Fermentation technology, 1
Fermentors, 206207
large-scale, 264f
Ferrous ion, in medium, 111
Fetal bovine serum (FBS), 118
Fetuin, in medium, 123t
Fibroblast(s)
doubling time of, 154, 154t
human vs. chicken embryo, 19
life span in culture, 54
on microcarrier, 203f
use in bioprocessing, 1920, 19t
vs. epithelial cells, 20
Fibroblast growth factor, 48
Fibronectin
Filopodia, 3839, 46
Filtration, 259260
FL72 cells, 8t
Fleming, Sir Alexander, 2
Flippases, 84, 90
Flooding, in bioreactors, 280
Fluid
extracellular, 100101, 100t
intracellular, 153154, 153t
Fluidized bed bioreactor, 207
Flux vectors, 179180
Folding, of proteins, 3233, 8991

Follow-up biologics, 3, 910
approval in Europe, 10, 10t
uncertainty about quality of, 10
Formalin, viral inactivation with, 4, 9
Fortified feed and addition culture, 235, 235f
454 sequencing technology, 304305
Friction factor, in scale translation, 267
Fructose
in glycan biosynthesis, 9394
in medium, 108
transport of, 41, 71
Fructose 1,6-bisphosphate (F16BP), 75, 75f
Fructose 2,6-bisphosphate (F26BP), 77
Fructose-6-phosphate, 67
FS-4 cells, 19t
Functional Annotation of the Mammalian genome (FANTOM),
286287
Fungus, genome of, 288
Fusion proteins, 8
G
G1 phase of cell cycle, 4748, 48f
G2 phase of cell cycle, 4748, 48f
G actin, 39
Galactose
in glycan biosynthesis, 9394
in medium, 108
transport of, 71
Gangliosides, in lipid bilayer, 22, 24
GAPDH. See Glyceraldehyde dehydrogenase
Gap phase of cell cycle, 4748, 48f
Gas-liquid interface, oxygen transfer through, 212219, 213f,
214f
Gas phase, in bioreactor, 196, 276, 278279
Gas sparging, 220229
damage to cells by, 226228, 226f, 227f, 228f
direct, 221
orifice and bubble size in, 222223
Gelatin, for microcarriers, 203204, 205t
Gene(s), 285287
abundant, 293, 293t
alternative splicing of, 285286
coding for non-protein RNA, 285286
coding for proteins, 285286, 288
coding for RNA, 285286
definition of, 285
environment and, 291293
minimum set of, 287
number of, 287, 289t
rare, 293294, 293t
structure in eukaryotes, 286
Gene expression, 285
in cell culture processing, 306308, 306f, 307f
epigenetic regulation of, 291293
proteome profiling of, 299302, 299f
INDEX |315
transciptome analysis of, 293297

Genentech, Vacaville (California) facility, 12
Geneticin, as selection marker, 137t
Gene transfer
amplification in, 131132, 138141, 140f, 141f
cell adaptation to, 142, 142f, 143f
direction to transcriptionally active region, 141
methods of, 133134, 134t
promoters for, 134135
selectable marker in, 135137, 137t
for stable cell line, 131145
Genome(s)
and complexity of organism, 287
environment and, 291293
eukaryotic, 28
mammalian, 287289
mitochondrial, 2930
mouse, 285287, 288, 289t, 293
organization of, 287289, 287f
prokaryotic, 28, 287288
repetitive sequences in, 287288
species comparison of, 287288, 289t
Genome engineering, 3
Genome scale analysis, 293297
Genomics, 3, 285308
cell line, 146, 306308, 306f, 307f
protein (proteome profiling), 299302
transcriptome, 293297
Gentamicin, in medium, 117, 117t
Geometric-nonsimilar scale translation, 264
Geometric-similar scale translation, 264
Germanium, in medium, 112
Glass, as microcarrier, 203204, 205t
Gluconeogenesis, 67
Glucosaminoglycans, in extracellular matrix, 4546
Glucose
in fedbatch culture, 236237, 238240, 239f, 246247
lactate ratio to, 158159, 236237, 236f, 236t
in medium, 106107, 149
metabolism of. See Glucose metabolism
specific consumption rate of, 156
transport of, 3943, 42f, 7172, 71f, 72t
Glucose-6-phosphate, 64, 67
Glucose metabolism, 5877, 62f, 70f
aerobic, 6566
amino acid metabolism and, 8182, 82f
anaerobic, 58, 65
ATP consumption in, 6061
ATP production in, 6064, 66, 70
carbon flow or flux in, 67
carbon production in, 59, 6167
in culture, 149
in electron transfer chain, 61, 6364, 66
glutamine and, 80, 80f
insulin in, 120

lactate consumption in, 59, 7880, 79f
lactate conversion in, 5859, 60, 6566, 67
lipid metabolism and, 86
metabolic flux analysis of, 186
NADH balance in, 6869, 69f
reaction intermediates in, 67
regulation of, 7478
differential allosteric, 7576, 75f
growth control and, 77, 77f
isozymes in, 7476, 75f
signaling pathways and, 77, 77f
transport in, 7174
Warburg effect in, 6566
yield of, 6061, 70
Glucose oxidation, 5864, 70f
Glutamate, metabolism of, 8182
Glutamine
in medium, 106108, 149
metabolism of, 58, 59, 67, 8182
in regulation of glucose metabolism, 80, 80f
Glutamine oxidation, 58
Glutamine synthetase (GS)
for amplification, 138141, 141f
for glutamine synthesis, 108
Glutathione
reduced, in medium, 116
reduction of, 64
GLUT transporter(s), 71, 72t
GLUT1 transporter, 4041, 43, 71, 72t, 107
GLUT2 transporter, 72t
GLUT4 transporter, 43, 71, 72t, 77
GLUT5 transporter, 41, 71, 72t, 107
Glycan(s)
biosynthesis of, 67, 8996
diversity among species, 9596
effect/importance of, 8990
extension in Golgi apparatus, 9192, 91f, 9394
and immunogenicity, 9596
macroheterogeneity of, 89
microheterogeneity of, 89, 9293, 92f
N-linked, 8992, 90f, 91f, 174, 174f
nucleotide sugar precursors of, 9394, 93f, 94f
O-linked, 89
types of, 9293, 92f
visualization of data on, 174, 174f
Glyceraldehyde dehydrogenase (GAPDH), 135
INDEX |316
Glycerol, as backbone of phospholipid, 22, 22f, 24

Glycine
as buffer in medium, 115t
in medium, 108
Glycoforms, heterogeneity in, 89
Glycolipids, in lipid bilayer, 22, 24
Glycolysis, 5870
aerobic, 6566
carbon flow and reaction intermediates in, 67
in culture, 149
lactate consumption in, 7880
metabolic flux analysis of, 186
pentose phosphate pathway as shunt from, 60, 6465
regulation of, 7478
transport and, 7174
yield of, 6061, 70
Glycosylation, 5, 1214, 8996, 90f
diversity among species, 9596
multiple sites of, 89
N-linked, 8992, 90f, 91f
O-linked, 89, 90
in secretion process, 33
visualization of data on, 174, 174f
Glycosyltransferases, 34
Glycylglycine, as buffer in medium, 115t
Glyeraldehyde-3-phosphate (G3P), 61
Golgi apparatus, 27, 3234
classical view of, 32
compartments of, 32
dynamic nature of, 32
glycan extension in, 9192, 91f, 9394
glycosylation in, 90
protein secretion through, 3235, 35f, 83
transport across, 3334
Green monkey kidney cells, 8t, 9, 19t, 129, 201
Growth
autocatalytic, 156
balance equations for, 162165
cell death consideration in, 157
doubling time of, 154157, 154t
kinetic model of, 162165
mammalian cell, 154155, 154f, 154t
material balance on, 149154, 149f, 162165
medium for. See Medium
multiplicative model of, 165
overall synthesis equation for, 149150
phases of, 151f, 154155, 154f
quantitative description of, 156161
Growth control
apoptosis in, 5153
cell cycle and, 4749
contact inhibition in, 46, 48, 53, 54f
Hayflicks phenomenon and, 54
loss, in continuous cell lines, 49
in regulation of glucose metabolism, 77, 77f
telomeres and, 5455, 54f
Growth curve, 154155, 154f

Growth factors
in cell cycle, 48, 49
in cell migration, 4647
in medium, 9798, 117, 118
Growth hormone. See Human growth hormone
Growth rate, 5355, 53f, 156157
GS. See Glutamine synthetase
GTP-mannose, 67
Guanosine, in medium, 109t
H
Habitation-conducive components, of medium, 103104
Hamster cells, 8t, 9, 19t, 20. See also BHK cells; Chinese hamster
ovary cells
Hayflicks phenomenon, 54
Heavy metal ions, in medium, 112
HEK 293 cells, 8t, 19t
adhesion of, 142
aggregates of, 209, 209f
Helicos sequencing technology, 302t
Henrys law, 217218
Heparin
in medium, 123
Hepatocyte(s)
cell membrane of, 25, 25t
endoplasmic reticulum of, 31
size of, 21, 151
Hepatocyte growth factor, 4647
HEPES buffer, 115, 115t
HepG2 cells, on microcarrier, 204, 204f
Heterochromatin, 290
Heterogeneous bioreactor, 196, 196f
Hexokinase (HK), in regulation of glucose metabolism, 75
hGH. See Human growth hormone
High-mannose glycans, 92, 92f
High-molecular-weight supplements, in medium, 117125
High-performance liquid chromatography (HPLC), for fed-batch
culture, 243
High throughput technology
for cell line production, 145146
for DNA microarrays, 296297
for DNA sequencing, 302305
for proteome profiling, 299302, 299f
Histone, 28, 290, 290f
Histone modification, 292293
HMG-CoA, 8788, 88f
HMG-CoA reductase (HMGCR), 8788, 88f
HMG-CoA synthase (HMGCS), 8788, 88f
Holding time, in secretion, 32, 34
Hollow fiber bioreactor, 208, 208f, 229
Homeostasis, of cell membranes, 26
Hormone(s)
in interstitial fluid, 100
INDEX |317
in serum, 118
Host systems, 56, 1214, 127129. See also specific systems
Hot spot integration, 141
HPLC, for fedbatch culture, 243
Human growth hormone (hGH)
biosimilar, 910
Humira, expiration of patent, 9
Hyaluronic acid, in extracellular matrix, 46
Hybrid glycans, 92
Hybridoma, 5
adhesion of, 142
gene expression in, 306, 306f
growth of, 149
stoichiometric ratio and metabolic stress in, 161, 161t
volume of, 150t
Hydrogen
transport of, 4345
Hydrogen peroxide, 116
Hydrolysates, in medium, 124, 240
Hydrostatic pressure, and oxygen transfer, 225
Hydroxylethyl starch (HES), in medium, 116t
Hygromycin B, as selection marker, 137t
Hyper-producing cell line, 129133. See also Stable cell lines
characteristics of, 129131, 130t, 131f
gene expression in, 306308
I
IEF. See Isoelectric focusing (IEF)
IGF. See Insulin-like growth factor(s)
IgG. See Immunoglobuin G
Illumina sequencing technology, 302t, 304305
Immobilization reactors
agarose, 209
oxygen transfer in, 229, 229f
Immortalization, 54
Immunogenicity, glycans and, 9596
Immunoglobuin G (IgG)
in cellular growth, 151, 151t
Fc fragment, in fusion protein, 8
secretion time of, 34f, 34t
Impellers, in bioreactors, 206, 265268, 265t, 266f
amount of fluid moved by (pumping), 268270
constant tip speed, for scale translation, 269270, 269t
velocity of, 268
Inactivated vaccines, 4, 9
Incline settling, 255, 255f
India, biosimilars marketed in, 10t
Inducible promoters, 134
Industrial cell lines, 8t, 9
Industrial production
bioreactors for, 192193. See also Bioreactors
medium for, 125126
scale translation for, 263284. See also Scale translation;

Scaling up
In-process calculations, 169
Insect cell culture, 1213, 13t
Insulin
in cell cycle, 48, 49
in glucose metabolism, 120
in glucose transport, 43, 71
in medium, 120121, 240
mitogenic response to, 120
recombinant, 121
as tissue-derived protein therapeutic, 5
Insulin-like growth factor(s), 48
Insulin-like growth factor-1, in medium, 120121
Insulin-like growth factor-2, in medium, 120
Integral cell concentration, 159, 159f
Integrated data, calculated, 169
Interactive data exploration, 172
Interferon(s), as tissue-derived protein therapeutic, 5
Intermediate(s), of reaction modes, abbreviation of, 189t
Intermediate filaments, 37, 38, 38f
Intermittent harvest and feed, 198, 198f, 234, 234f
Interstitial fluid, 100101, 100t
In-time measurement, for fedbatch culture, 243
Intracellular fluid, 153154, 153t
Introns, 286, 287, 288
Ion(s)
bulk, in medium, 111112, 111t
intracellular vs. extracellular concentration of, 2122, 21t,
100101, 100t
transport of, 4345
Ion channels, as transport mechanism, 4042, 41f
Iron
free vs. bound form of, 45
in medium, 111, 112
reactivity of, 45
transport of, 45, 122t. See also Transferrin
Iron chelators, 121, 121t
Isobaric Tagging for Relative and Absolute Protein Quantitation
(iTRAQ), 300301, 300f, 301f
Isoelectric focusing (IEF), 299
Isoleucine, metabolism of, 8182
Isozymes, in glucose metabolism regulation, 7476, 75f
iTRAQ, 300301, 300f, 301f
J
Jenner, Edward, 199
K
Karyotype, variable, in cell lines, 143144
a-Ketoglutarate
in glucose metabolism, 61, 68, 80
in medium, 108
Kinetics
INDEX |318
in bioreactors, 194196
of cell cultivation, 147166
model of growth, 162165
KL. See Mass transfer coefficient
Krebs cycle. See Tricarboxylic acid cycle
L
Lactate
accumulation in culture, 78, 236
in fedbatch culture, 236237, 246247
metabolism of, 5859, 6566, 7880
consumption in glucose metabolism, 59, 7880, 79f
correlation with productivity, 58f, 59
production in glucose metabolism, 5859, 60, 6566, 67
specific production rate of, 156
transport of, 4142, 42f, 43, 7273, 72f
Lactate dehydrogenase, 66, 67, 79
Lactate-to-glucose ratio, 158159, 236237, 236f, 236t
Lag phase, of cell growth, 154, 154f
Lamellipodia, 3839, 46
Laminins
Large scale, translation for. See Scale translation; Scaling up
Large-scale fermenter, 264f
LDL. See Low-density lipoprotein
Lehmann, Jrgen, 210
Length, in scale translation, 264
Leucine, metabolism of, 8182
Lipid(s)
amphipathic, 2223
catabolism of, 36
in cell membrane, 2227
functions of, 83
in medium, 84, 125, 240
metabolism of, 8388
acetyl CoA shuttle in, 86, 86f
subcellular localization of, 83
transport of, 84
Lipid bilayer, 2227, 83, 87
characteristics of, 23
dynamic nature of, 23, 26
function of, 25
permeability of, 23, 39
phase transition of, 23, 23f, 24
transport across, 23, 2627, 3947
Lipofection/lipid-mediated gene transfer, 133134, 134t
Lipoprotein(s)
in medium, 105106
transport of, 84
Liquid chromatography, 2D, 300302
Liquid chromatography/mass spectroscopy, 300
Liquid phase, in bioreactor, 196, 277279
Live attenuated vaccines, 45

Liver, endoplasmic reticulum of, 31
Localized amplification, in DNA sequencing, 304, 304f
Logging data. See also Data processing
standardized templates for, 168169
Low-density lipoprotein (LDL)
in medium, 105106
transport of, 84
Lymphoid cells, use in bioprocessing, 1920, 19t
Lysis, 157
Lysosome, 36
M
MA104 cells, 8t
Macroheterogeneity, of glcyans, 89
Macromolecules
catalytic, in medium, 105106
transport of, 3940
Macroporous microcarriers, 203, 204, 204f, 205t, 229
Madin-Darby bovine kidney (MDBK) cells, 8t
Madin-Darby canine kidney (MDCK) cells, 8t, 9, 19t
Magnesium ions
in fedbatch culture, 240
intracellular and extracellular concentrations of, 21, 100,
100t, 153, 153t
Malate-aspartate shuttle, 6869, 186
MALDI. See Matrix-assisted laser desorption ionization
Mammalian cells. See also specific cells
bioreactors for, 192193. See also Bioreactors
critical feature of rDNA proteins from, 14
fragility of cells in, 22
genome of, 28
growth of, 154155, 154f, 154t
limitations of, 14
product quality and process robustness in, 1417
products from, 56, 6t
transgenic animals in, 12, 14, 14t
Mammalian genome, 287288, 289t
Manganese, in medium, 112
Mannose, in glycan biosynthesis, 9094
Manufacturing plants, 12
Manufacturing processes, 1112, 12f. See also specific processes
and products
Mass. See Biomass
Mass spectrometry, in proteome profiling, 299302
Mass transfer coefficient (KL)
for carbon dioxide, 282
for oxygen, 216
constant, in scale translation, 269270, 269t
experimental measurement of, 224225, 224f
scaling up and, 277
sparger orifice and bubble size and, 222223
Material balance
on cell growth, 149154, 149f, 162165
in fedbatch culture, 236237, 236f, 236t, 238240, 239f, 239t,
INDEX |319
246247
on oxygen, in bioreactors, 220, 276280
on perfusion culture, 251253, 252f, 253f
for reaction systems, 175180. See also Metabolic flux
analysis
setting up equations, 178179
systematic way to solve problems, 178
Mathematical model
of growth, 162165
information needed for developing, 163
Monod and Monod derivatives, 163165, 164f
purposes of, 162
Matrigel, in medium, 123
Matrix-assisted laser desorption ionization (MALDI), 299
Matrix operations, 179180
MDBK cells, 8t
MDCK cells, 8t, 9, 1920, 19t
MDR. See Multidrug resistance gene
Measurement data, 169
Mechanical/acoustic trapping, 257258, 258f
Mechanical damage protective agents, in medium, 116117,
116t
Medial Golgi, 32, 33, 35f
Medium
amino acids in, 81, 106108, 238240, 246247
animal component-free, 102103
antibiotics in, 117, 117t
basal, 101, 106117
buffer systems in, 113115, 115f, 115t
bulk ions in, 111112, 111t
carrier proteins in, 117, 118, 122
cell adhesion molecules in, 122123, 123t
for cell expansion, 9799, 125
chemically defined, 102, 103
classical, composition of, 58
complex vs. chemically defined, 102
components of
basic, 103117
classes of, 103106
non-nutritional, 113
stochiometric vs. catalytic macromolecular, 105106
stochiometric vs. habitation-conducive, 103104
design for, 97126
for fedbatch culture, 233, 237240
for consumed nutrients, 238240, 239f, 239t
for unconsumed components, 240
fundamental influence of, 97
high-molecular-weight and complex supplements in, 117125
for industrial production, 125126
insulin and insulin-like growth factors in, 120121, 240
lipids in, 84, 125, 240
nucleosides in, 109
objectives of, 98
optimal concentration of organic nutrients in, 110, 110f
optimization of cell growth environment in, 97101
osmolarity of, 111112
for production, 9799, 125126

protective agents in, 116117, 116t
protein-free, 103
protein hydrolysates in, 124, 240
serum albumin in, 118, 122
serum-free, 102103, 108, 109
serum in, 102, 117119, 240
for stem cells, 9799, 117
sugars and energy source in, 106107
tolerance of deviation from optimum, 100101, 101t
trace elements in, 111112, 112t
transferrin in, 105, 116, 118, 121, 240
types of, 101103
vitamins in, 109, 240
water in, 106
Membrane, cell. See Cell membrane
Membrane bioreactor, 208209
Membrane fusion, 3940
Membrane potentials, 2526, 25t, 2930, 30f, 4345
Membrane stirred tank, 210, 210f
Mercaptoethanol, in medium, 116
Messenger RNA (mRNA), 21, 286
abundant, intermediate, and rare, 293294, 293t
in RNA-seq, 296297
Metabolic flux analysis (MFA), 173, 173f, 175187
abbreviation of intermediates in, 189t
biomass equations in, 184185, 185f
catabolism reactions for, 188t
on cellular system, 181186
compartmentalization in, 184
example of, 186
general approach to, 182f, 183185
selecting reactions for, 183, 183f
solution and analysis in, 185
utility and limitations of, 181182
Metabolic state, of culture, 158, 236237, 236f, 236t
Metabolic stress, stoichiometric ratio as indicator of, 161, 161t
Metabolism. See also specific types
amino acid, 58, 8182, 81f, 82f
central, in cultured cells, 5859
glucose, 5870
lactate, 5859, 6566, 7880
lipid, 8388
transport and transporters in, 7174
Methane combustion, 176178
Methionine sulphoximine (MSX), in glutathione synthetase
amplification, 138141
Methotrexate (MTX), in DHFR amplification, 138141, 140f
Methylation
DNA, 292293
histone, 292
Methylcelluloses (MC), in medium, 116t
MFA. See Metabolic flux analysis
Micelles, 2223
MichaelisMenton enzyme kinetics, 41
Microarray analysis. See DNA microarrays
INDEX |320
Microcarriers, 203204, 205t

characteristics of, 204t
macroporous, 203, 204, 204f, 205t, 229
microporous, 203204, 203f, 205t
solid, 203204, 203f
Microencapsulation, 210, 210f
Microfiltration, 259260, 260f
Microheterogeneity, of glcyans, 89, 9293, 92f
Microporous microcarriers, 203204, 203f, 205t
Microsphere-induced cell aggregates, 209, 209f
Microtubules, 37, 37f, 46
Minerals, in medium, 111112
Minimum set of genes, 287
Mitochondria, 2930
acetyl CoA shuttle in, 86
as cells power plant, 29
genome of, 2930
lipid metabolism in, 83, 85, 86, 88f
membrane potential of, 2930, 30f
pH of, 2930
protein content of, 25
size of, 29
transport across, 7374
Mitochondrial apoptotic pathway, 5253, 53f
Mitochondrial DNA, 2930
Mitogenic factors, 48, 49
Mitosis, 4748, 48f
Mixing time, in bioreactors, 206, 268274
distribution of, 273274, 273f, 274f
measurement of, 272, 272f
in scale translation, 269274
vs. starvation time, 272
Molybdenum, in medium, 112
Monkey cells, 4, 8t, 9, 19t, 129, 201
Monocarboxylate transporter (MCT), 42, 43, 7273, 72f
Monod-derivative models, 164165
Monod models, 164165, 164f
Monolayer, 53
MOPS buffer, 115t
Mouse cells, 8t, 9, 19t, 49. See also NSO cells; SP2/0 cells
embryonic stem, doubling time of, 154t
gene expression in, 306, 306f
Mouse genome, 285287, 288, 289t, 293
Movement, cellular, 4647
M phase of cell cycle, 4748, 48f
MRC-5 cells, 19t
MRCS cells, 8t
mRNA. See Messenger RNA
MSX, in glutathione synthetase amplification, 138141
MTX, in DHFR amplification, 138141, 140f
Multidimensional data exploration, 172
Multidrug resistance (MDR) gene, 137
Multiple membrane plate bioreactor, 208209
Multiple plate system, 199, 201, 201f, 212
Multiplicative model, of growth, 165
Myc, in regulation of glucose metabolism, 77, 77f
Myelin membrane, protein content of, 25

Myeloma cells, 20. See also NSO cells
for stable expression, 129130, 130t, 146
stoichiometric ratio and metabolic stress in, 161, 161t
N
NADH
balance of, 6869, 69f
carrier system for, 6869
in electron transfer chain, 6364, 66
in glycolysis, 6064, 66, 6869, 69f
in lactate metabolism, 7980
in lipid metabolism, 86
in tricarboxylic acid cycle, 6163, 6869, 69f
NADPH
in pentose phosphate pathway, 6465
Necrosis, 51
Neomycin
in gene transfer, 139140
in medium, 117
Next generation sequencing technologies, 304305
Nickel, in medium, 112
NimbleGen microarrays, 295296, 295t
Nitrogen
in cultured cell metabolism, 67
metabolic flux analysis of, 182f, 183
N-linked glycosylation, 8992, 90f, 91f, 174, 174f
Non-essential amino acids, 81, 108, 108t
Non-nutritional components, of medium, 113
NSO cells, 8t, 19t, 20
doubling time of, 154t
gene expression in, 307f
for stable expression, 130t
Nuclear envelope, 29
Nuclear membrane, 29
Nuclear pores, 29
Nucleoid, 28
Nucleoside(s), in medium, 109
Nucleosomes, 290
Nucleotides, precursors of glycans, 9394, 93f, 94f
Nucleus, 2729, 28f
Nunc Cell Factories, 201
Nutrient consumption curve, 154155
Nutrient consumption rate, specific, 156157
Nutrients, transport of, 43
Nutrient starving time, 271272, 271t
Nystatin, in medium, 117t
O
OAA. See Oxaloacetate
Off-line data, 160, 160f
INDEX |321
Oligonucleotide-based microarrays (oligoDNA), 295296, 298t

Oligopeptides, transport of, 43
O-linked glycosylation, 89
Omnitrope, 910
On-line data, 160, 160f
On-line measurement, for fedbatch culture, 243
Organelle(s), 2324, 2736, 28f. See also specific organelles
Organic nutrients, in medium, optimal concentration of, 110,
110f
Osmolarity
of interstitial fluid, 100
of medium, 111112, 125
OTR. See Oxygen transfer rate
OUR. See Oxygen uptake rate
Oxaloacetate (OAA), in glucose metabolism, 61, 68
Oxidative phosphorylation pathway, 61
Oxygen
cellular demand for, 219
concentration in medium, 217218
consumption of, 219
dissolved concentration of, 214
optimal concentration of, 219
transport of, 39
Oxygen balance, in reactor, 220, 276280
on gas phase, 276, 278279
on liquid phase, 277279
Oxygen starvation time, 271272, 271t
Oxygen supply, for bioreactors, 220228
by medium recirculation, 221f
by silicon tubing/membrane, 220
by sparging, 220228
direct, 221
by surface aeration, 220
Oxygen transfer, in bioreactors, 206207, 213231
driving force for, 214216, 220, 277, 279280
enhancing or improving, 216, 220, 221
experimental measurement of, 224225, 224f, 225f
hydrostatic pressure and, 225
in immobilization reactor, 229, 229f
mass transfer coefficient in, 216
objective of, 219
in plug-flow reactors, 229230, 230f
rate of, 215216, 220
scaling up and, 275280, 280f, 280t
surface/interfacial area and, 216
through gas-liquid interface, 212219, 213f, 214f
Oxygen transfer rate (OTR), 215216, 220
balance with oxygen uptake rate, 220, 276280
scaling up and, 276280
Oxygen uptake rate (OUR), 160, 160f, 219, 220, 224225, 225f
balance with oxygen transfer rate, 220, 276280
scaling up and, 276280
P
p53 tumor suppressor, in regulation of glucose metabolism, 77
Pancreas, beta cells of, 31
Passage, 5355
PAT. See Process analytical technology
Patents, expiration of, 910
Pathway-related data, 173, 173f. See also Metabolic flux analysis
PCR. See Polymerase chain reaction
PDI. See Protein disulfide isomerase
PDQuest, 299
Penicillin(s)
declining price of, 23
development and production of, 24, 2f
discovery of, 2
Penicillin G
in medium, 117t
production outside U.S., 23
Pentose phosphate pathway (PPP), 60, 62f, 6465
in culture, 149
molecular transformation in, 6465
oxidative segment of, 64
Peptides, in medium, 108, 123t
Peptone, in medium, 124
Perfusion culture, 249261
analysis of, 251254
cell culture products from, 249t
cell retention in, 251, 254261
external vs. internal recovery device in, 252
material balance on, 251253, 252f, 253f
recycling factor in, 254, 254f
Permeability, of lipid bilayer, 23, 39
Permeases, 39
Peroxisomes, 27, 36
lipid metabolism in, 83, 85, 87, 88f
PFK. See Phosphofructokinase
PFKFB. See Phosphofructokinase/fructose biphosphate
PFR. See Plug-flow reactor
pH
of bioreactors, scale translation and, 283, 283f
of fedbatch culture, 244
of medium, 113115
of mitochondria, 2930
Phenotype, epigentic changes in, 291293
Phosphate
intracellular and extracellular concentrations of, 100, 100t,
153, 153t
in medium, 104, 111, 111t
transport of, 4345
Phosphatidyl choline, in medium, 125
Phosphatidyl ethanolamine, in medium, 125
Phosphofructokinase (PFK), in regulation of glucose metabolism,
75, 75f
Phosphofructokinase/fructose biphosphate (PFKFB), in
regulation of glucose metabolism, 7576, 75f
INDEX |322
Phospholipids
types of, 22, 22f
Phosphorylation
of histones, 292293, 292t
in protein therapeutics, 14, 16
Photolithographic synthesis, in microarray analysis, 295, 295t,
296f
Physiological state, of culture, 158
Pichia (yeast) cell culture, 1213, 13t
Pinocytosis, 3940
Piston-flow reactor. See Plug-flow reactor
PK. See Pyruvate kinase
PK cells, 8t
Plants, transgenic, 12
Plasma cells, 20, 33, 33f, 130, 151
Plasma membrane. See Cell membrane
Plasmid(s). See also Gene transfer
basic elements on, 134137, 134f
free, 135
method of delivery, 133134, 134t
selectable marker for, 135137, 137t
Plastic, for microcarriers, 204, 205t
Plug-flow reactor (PFR), 193196, 193f
Pluronic F68, in medium, 116117, 116t, 123
Pluronic F77, in medium, 117
Pluronic F88, in medium, 116t, 117
PolyA tail, 286
Poly-d-lysine, in medium, 123t
Polyethylene glycol (PEG), in medium, 116t
Poly-L-lysine, in medium, 123
Polymerase chain reaction (PCR), 304
Polymyxin B, in medium, 117
Polypropylene microcarriers, 205t
Polystyrene, for microcarriers, 203204, 205t
Polyvinyl alcohol (PVA), in medium, 116t
Polyvinylpyrrolidone (PVP), in medium, 116t
Post-process calculations, 169
Post-translational modification
analysis of, 299
in endoplasmic reticulum, 31
in protein therapeutics, 5, 1217
Potassium chloride, in medium, 111112
Potassium ions
intracellular and extracellular concentrations of, 2122, 100
101, 100t, 153, 153t
in medium, 111112, 111t
transport of, 4345
PPP. See Pentose phosphate pathway
pRB. See Retinoblastoma protein
Process analytical technology (PAT), 167168

Product accumulation rate, 156, 159
Product concentration profile, 154155
Product formation. See also specific products
kinetic model of, 162165
lactate metabolism and, 58f, 59
quantitative description of, 156161
specific rate of, 156157, 159, 165166
Programmed cell death. See Apoptosis
Proline, in medium, 108
Promoter
for gene transfer, 134135
in mammalian genes, 286, 288
Pro-oncogenic genes, in regulation of glucose metabolism, 77,
77f
Proportional feeding
with base addition, 244
with oxygen uptake rate, 245
with turbidity, 244
Protease inhibitors, in serum, 118
Proteases, for dissociation, 53
Proteasome, 36
Protective agents, in medium, 116117, 116t
Protein(s)
carrier
in medium, 117, 118, 122
transport by, 122, 122t
in cytoplasm, 27, 27t
in cytoskeleton, 37
folding of, 3233, 8991
genes coding for, 285286, 288
in interstitial fluid, 100
secretion of
and cell membrane, 26
in endoplasmic reticulum, 3135, 35f, 83
in Golgi apparatus, 3235, 35f, 83
time of, 32, 34, 34f, 34t, 35, 35f
Protein C, in-process structural alterations to, 16
Protein disulfide isomerase (PDI), 33
Protein-free medium, 103
Protein hydrolysates, in medium, 124, 240
Protein molecules, as therapeutics, 78. See also Protein
therapeutics
Protein therapeutics, 48
alternative technologies for, 1214
biosimilar or follow-up biologics, 910
g-carboxylation in, 12, 14, 16
disulfide-bond formation in, 5, 14
fedbatch culture for, 233, 235f. See also Fedbatch culture
from fusion proteins, 8
INDEX |323
glycosylation in, 5, 1216, 8996, 90f

growth and advances in, 1
host cells for, 127129
immunogenicity of, 9596
industrial cell lines for, 8t, 9
in-process structural alterations to, 1517, 16t
instability in production of, 144
from mammalian cells, 6t
from non-mammalian host, 6t
phosphorylation in, 14, 16
post-translational modification of, 5, 1217
product quality and process robustness for, 1417
recombinant technology for, 5
stable cell lines for, 127, 129146
stoichiometric ratio in, 11
tissue-derived, 5
transient expression of, 127129
Proteoglycans, in extracellular matrix, 4546
Proteome profiling, 299302, 299f
iTRAQ labeling in, 300301, 300f, 301f
SILAC labeling in, 300302, 301f
2D liquid chromatography in, 300302
Proton pumps, 36
Pseudogenes, 286
Pulsatile flow, in microfiltration, 260, 260f
Pumping, in bioreactor, 268270
Purines, in medium, 109, 109t
Puromycin, as selection marker, 137t
Pyridoxine, in medium, 109
Pyruvate
as controlling node, 67
in glycolysis, 60, 61, 66, 67, 70
lactate conversion to, 7980
in medium, 107
NADH balance and, 6869, 69f
transport of, 43, 7273, 72f
in tricarboxylic acid cycle, 60, 61
Pyruvate kinase (PK), in regulation of glucose metabolism, 7576
Q
Quality, of cell culture products, 1417
Quality by Design, 162
Quantitative description, of growth, 156161
Quasi-steady state, 179
R
Radial flow impellers, 206, 265266, 265t, 266f
Rare genes, 293294, 293t
Rate-limiting enzymes, 74
Rate-limiting nutrient, 165
Reaction systems
cellular vs. chemical, 175176
material balance for, 175180

Reactive oxygen species (ROS)
glutathione and, 64, 116
in medium, 116
Reactor state parameters, 160f
Read, in DNA sequencing, 297
Recessive marker, 136
Recombinant technology, 5, 192. See also specific applications
and products
Recovery process, 1112
Recycling factor, in perfusion culture, 249261, 254, 254f
Remicade, expiration of patent, 9
Repetitive sequences, in genome, 287288
Reporter gene, 136
Respiratory quotient (RQ), 219, 281
Retention, cell
methods of, 254261
in perfusion culture, 249261
Retinoblastoma protein (pRB), 4849
Retrograde transport, 34, 35, 35f
Reynolds number, 267268, 267f
RGD peptide, in medium, 123t
Riboflavin, in medium, 109
Ribonucleic acid. See RNA
Ribose
in medium, 107
synthesis of, 67
Ribosomal RNA (rRNA), 21, 27
removal, in microarray analysis, 296297
Ribosomes, 27, 31
Rice hydrolysate, 124
RNA
abundant, intermediate, and rare, 293294, 293t
genes coding for, 285286
location in cell, 28
non-protein coding, 285286
synthesis of, 28
RNA-seq, 296297, 298t
Robustness, of cell culture processes, 1417
Roche sequencer, 302t
Roller bottles, 199201, 200f, 212
ROS. See Reactive oxygen species
Rotational cage, 258, 258f
Rough endoplasmic reticulum, 31
RQ. See Respiratory quotient
rRNA. See Ribosomal RNA
Rubidium, in medium, 112
Rushton impellers, 206, 265266, 265t, 266f
S
Sacchromyces cerevisiae systems, 12
Saturated fatty acids, 85
Scale translation
and chemical environment, 283, 283f
INDEX |324
constant parameters in, 269270, 269t

dimensionless variables in, 267270
effect of scale on physical behavior, 269270
geometric nonsimilarity in, 264
geometric similarity in, 264
major effects of scale, 264
mixing time in, 271274, 271t
nutrient starvation time in, 271274, 271t
objective of, 264
physical and mechanical parameters of, 264
Reynolds number in, 267268, 267f
Scaling down, 263264. See also Scale translation
Scaling up
Secretion time, 32, 34, 34f, 34t, 35, 35f
Sedimentation, 254255, 254f
Selectable marker, in gene transfer, 135137, 137t
Selenate, in medium, 111
Selenium
in medium, 112, 116
Senescence, 5355
Separation methods, 254261
Sequencing, DNA. See DNA sequencing
Serine
as backbone of phospholipid, 22, 24
in medium, 108
Serum
disadvantages of use, 119
in medium, 102, 117119, 240
Serum-free medium, 102103, 108, 109
Settling cyclone, 254255, 254f
SGLT transporters, 7172
Shotgun liquid chromatography, 300
Sialic acid, in glycan biosynthesis, 9394
Signaling, cellular, 4546
Signaling pathways, in regulation of glucose metabolism, 77, 77f
Signal recognition particles (SRPs), 3233, 35, 35f
SILAC, 300302, 301f
Silicon tubing/membrane, for oxygen supply, 220
Simple stirred tank bioreactor, 206207
Single-cell cloning, 131133
Single molecule detection, in DNA sequencing, 304
Single-use bioreactor, 199
Smooth endoplasmic reticulum, 31
Sodium beta-glycero-phosphate buffer, 115
Sodium bicarbonate buffer, 113114, 115f, 115t
Sodium butyrate, 293
Sodium chloride, in medium, 111112

Sodium/glucose transporter, 42
Sodium ions, 2122, 21t
intracellular and extracellular concentrations of, 2122, 21t,
100101, 101t, 153, 153t
in medium, 104, 111112, 111t
transport of, 4345
Sodium/potassium ATPase transporter, 4445, 44f
Software, for data visualization, 172
Solid microcarriers, 203204, 203f
Solid phase, in bioreactor, 196
SOLiD sequencing technology, 302t
Solutes, cellular transport of, 40
Soybean hydrolysate, 124
SP2/0 cells, 19t, 49
for stable expression, 130t
Sparging, 220228
direct, 221
Specific death rate, 157
Specific growth rate, 156157
Specific nutrient consumption rate, 156157
Specific product formation rate, 156157, 159, 165166
Specific rate, 156157, 169
Spectinomycin, in medium, 117t
S phase of cell cycle, 4748, 48f
Sphingomyelin, in medium, 125
Spin filter separation, 258, 258f
Spin filter stirred tank, 210211, 211f
Spreadsheets, 168, 170171
SRPs. See Signal recognition particles
Stable cell lines, 127, 129146
adaptation of cells in, 131, 142, 142f, 143f
amplification for, 131132, 138141, 141f
automation and high throughput technology for, 145146,
145f
genomics and, 146
screening for, 131
selection of cells for, 131132
single-cell cloning for, 131132
stability of clones selected for, 143145
stability of product quality in, 144145
transfection for, 131
Stable Isotope Labeling by Amino Acid in Cell Culture (SILAC),
300302, 301f
Standardized templates, for data logging and processing, 168
169
Starvation time, in bioreactors, 271272, 271t
State of cultures, 158
Stationary phase, of cell growth, 151f, 154155, 154f
ST cells, 8t
Stem cell(s), 4
INDEX |325
differentiation in culture, 143

medium for, 9799, 117
mouse embryonic, doubling time of, 154t
size of, 21, 151
Stirred tank bioreactor, 193, 193f
agitation in, 265266
purpose of, 265
continuous, 193196, 193f, 212
heterogeneous, 196, 196f
impellers in, 206, 265268, 265t, 266f
membrane, 210, 210f
mixing time in, 206, 268274
power consumption in, 266270, 266f
simple, 206207
spin filter, 210211, 211f
velocity of, 268
volumetric flow rate in, 268270
well-mixed, 193196
Stock solutions, amino acid, 108
Stoichiometric components, of medium
vs. catalytic macromolecular components, 105106
vs. habitation-conducive components, 103104
Stoichiometric-limiting nutrient, 165
Stoichiometric matrix, 179180
Stoichiometric ratio, 11, 156, 158159
in data processing, 169, 171, 171f
in fedbatch culture, 160f, 236240, 236f, 236t, 239f, 239t,
246247
as indicator of metabolic stress, 161, 161t
of lactate to glucose, 158, 236237, 236f, 236t
of product to substrate, 158
Stoichiometry, of cell cultivation, 147166
Streptomycin, in medium, 117
Stress, metabolic, stoichiometric ratio as indicator of, 161, 161t
Substrate
stoichiometric ratio of product to, 158
yield of biomass on, 158
yield of product on, 158
Sugars, in medium, 106107
Superoxide dismutase, in medium, 116
Superoxide radical, 116
Support systems, cellular, 202206
Surface aeration, 220
Surface area, and scale translation, 264
Surfactants, in medium, 116117
Suspension culture, 202, 212
SV40 late promoter, 135
Symporters, 4142, 42f
Syrian hamster cells (BHK), 8t, 9, 19t
adhesion of, 142
aggregates of, 209
veterinarian vaccines from, 8t

T
Tangential flow, in microfiltration, 260, 260f
Taurine, in medium, 116
TCA. See Tricarboxylic acid cycle
Telomerase, 54
Telomeres, 5455, 54f
Temperature, and lipid bilayer, 24
T-flasks, 199
Thiamine pyrophosphate, in medium, 109
3T3 cells, 54
Thymidine, in medium, 109, 109t
TIGAR, 77
Tin, in medium, 112
Tissue culture systems, 199202
Tissue plasminogen activator (tPA)
development of, 5
in-process structural alterations to, 16
product quality and process robustness, 1516
Titers, increases in, 23, 2f
TNFa. See Tumor necrosis factor a (TNFa)
tPA. See Tissue plasminogen activator
Trace elements, in medium, 111112, 112t
Transcription, 28
Transcription factors, 28, 291
Transcriptome analysis, 293297
in cell culture processing, 306308, 306f, 307f
DNA microarrays for, 295297, 295t, 298t
Transfection. See also Gene transfer
Transferrin
iron chelators as alternative to, 121, 121t
iron transport by, 45, 122t
in medium, 105, 116, 118, 121, 240
recombinant, 121
secretion time of, 34, 34t
Transgenic animals, 12, 14, 14t
Transgenic plants, 12
Trans-Golgi network (TGN), 32, 33, 35, 35f
Transient expression, 127129
host cells for, 129
production life of system, 129
Translation of scale. See Scale translation; Scaling down; Scaling
up
Transport
classes of processes, 40
mechanisms of, 3947. See also specific mechanisms
in metabolism, 7174
nutrient, 43
Transporters, 4042. See also specific types
Tricarboxylic acid cycle (TCA), 6064, 62f
amino acid metabolism and, 8182, 82f
in culture, 149
INDEX |326
glutamine and, 80, 80f

lipid metabolism and, 86
NADH balance in, 6869, 69f
regulation of, 7478
transport and, 7174
TrichostabinA, 293
TRICINE buffer, 115t
Trypsin
for dissociation, 53, 212
for proteolysis, 300
Tubular bioreactors, 193196, 193f
Tumor necrosis factor a (TNFa), in fusion protein, 8
Turbidity, proportional feeding with, 244
2D liquid chromatography, 300302
U
Ubiquitin, 36
Ubiquitination of histones, 292293, 292t
UDP-galactose, 67
UDP-glucose, 67
Unfolded protein response (UPR), 33
Uniporters, 41, 42f, 71
Untranslated regions (UTRs), 286, 288
UPR. See Unfolded protein response
Uridine, in medium, 109t
Urokinase, as tissue-derived protein therapeutic, 5
UTRs. See Untranslated regions
V
Vaccine(s)
veterinary, 8t, 9
viral. See Viral vaccines
Vanadium, in medium, 112
Vector. See Gene transfer
Velocity, in bioreactors, 268, 278, 278f
Vero cells, 8t, 9, 19t, 201
Vesicle diffusion model, 33
Veterinary vaccines, 8t, 9
Vibromixer, 212, 212f
Viral vaccines, 45
cell sources for, 20
inactivated, 4, 9
industrial cell lines for, 8t, 9
live attenuated, 45
principal, in prevention of human disease, 5t
Viscosity, of medium, 116117
Vitamin(s), in medium, 109, 240
Vitamin A, in medium, 109
Vitamin B6, in medium, 109
Vitamin B12, in medium, 109
Vitamin C, in medium, 109
Vitamin D, in medium, 109

Vitamin E, in medium, 109, 116
Vitamin K, in medium, 109
Vitronectin, in medium, 123t
Volume of cells, variation in, 150151, 150t, 151f
Volumetric flow rate, in bioreactor, 268270
Volumetric rates, 156157
W
W1-38 cells, 19t
Warburg effect, 6566
Water
as cellular component, 21, 148, 153
in medium, 106
Water for injection (WFI), 106
WaveTM, 201
Well-mixed stirred tank reactor, 193196, 193f
Westfalia disc centrifuge, 256
WFI. See Water for injection
X
XBP-1, 33
Y
Yeast (Pichia) cell culture, 1213, 13t
Yield coefficient, 158159
Z
Zeocin, as selection marker, 137t
Zinc
in medium, 111, 112
Zirconium, in medium, 112
Zwitterionic buffers, 115
INDEX |327

Cell Culture Bioprocess Engineering

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Cell Culture Bioprocess Engineering

Uploaded by

Copyright:

Available Formats

Cell Culture Bioprocess Engineering

With Contributions From:

Copyright 2012 by Wei-Shou Hu

Overview of Cell Culture

Cell Culture Engineering

the case of cell culture bioprocess technology, we

Fig. 1.1: Historical trend of penicillin titer and

As we look to the future of cell culture processing, it

Sir Alexander Flemings discovery of penicillin began

OVERVIEW OF CELL CULTURE TECHNOLOGY |2

dramatically lower in other parts of the world.

Now, two decades after the first introduction

A graph of historical data for cell culture products

Even for penicillin, there were tremendous process

OVERVIEW OF CELL CULTURE TECHNOLOGY |3

Cell Culture Products

production. In transforming cell culture products

Most viruses used as vaccines have been inactivated

OVERVIEW OF CELL CULTURE TECHNOLOGY |4

Table 1. Principal Viral Vaccines Used in Prevention of

Tissue culture (human

Tissue culture (chicken

Tissue culture (chicken

Tissue culture (duck

Lymph from calf or sheep

Tissue cultures and eggs

Highly purified subunit

Cell culture (MDCK, Vero) Attenuated

Duck embryo or human

Human diploid cell

Guinea pig heart cell

Chicken embryo cell

Chicken embryo cell

Russian spring summer

often by the prolonged cultivation in a non-human

Vaccine technology predated modern cell culture

The generation of recombinant DNA therapeutic

Initially, hybridoma cells were used. These are

OVERVIEW OF CELL CULTURE TECHNOLOGY |5

Table 2. Therapeutic Protein Biologics Produced in

White blood cell growth for

Anticancer, viral infections

Somatropin [human growth

Somatropin [human growth

Table 3. Non-Antibody Products Produced in Mammalian Cells

-glucocerebrosidase Gauchers disease

BioMarin Pharmaceutical 2005

Acute myocardial infraction

Relapsing multiple sclerosis

Relapsing multiple sclerosis

TNF receptor fusion

OVERVIEW OF CELL CULTURE TECHNOLOGY |6

Table 4. Therapeutic Antibody Products

Reversal of acute kidney

Johnson & Johnson

Prevention of blood clots

Humanized, anti-subunit T cell IL-2

Prevention of acute kidney

Chimeric, anti-chain T cell IL-2

Prophylaxis of acute organ

Prophylaxis of lowerrespiratory-tract disease

Active Crohns disease

Metastatic breast cancer

Acute myeloid leukemia