THE

CHEMICAL
Studies of Thermobifida fusca Plant Cell
RECORD Wall Degrading Enzymes

DAVID B. WILSON
Department of Molecular Biology & Genetics, Cornell University, 458 Biotechnology Building,
Ithaca, NY 14853, USA

SYNOPSIS: I have been studying the Thermobifida fusca cellulose degrading proteins for the past
25 years. In this period, we have purified and characterized the six extracellular cellulases and an
intracellular b- glucosidase used by T. fusca for cellulose degradation, cloned and sequenced the struc-
tural genes encoding these enzymes, and helped to determine the 3-dimensional structures of two
of the cellulase catalytic domains. This research determined the mechanism of a novel class of cel-
lulase, family 9 processive endoglucanases, and helped to show that there were two types of exocel-
lulases, ones that attacked the non-reducing ends of cellulose and ones that attacked the reducing
ends. It also led to the sequencing of the T. fusca genome by the DOE Joint Genome Institute. We
have studied the mechanisms that regulate T. fusca cellulases and have shown that cellobiose is the
inducer and that cellulase synthesis is repressed by any good carbon source. A regulatory protein
(CelR) that functions in the induction control has been purified, characterized, and its structural
gene cloned and expressed in E. coli. I have also carried out research on two rumen bacteria, Pre-
votella ruminicola and Fibrobacter succinogenes, in collaboration with Professor James Russell, helping
to arrange for the genomes of these two organisms to be sequenced by TIGR, funded by a USDA
grant to the North American Consortium for Genomics of Fibrolytic Ruminal Biology. © 2004 The
Japan Chemical Journal Forum and Wiley Periodicals, Inc. Chem Rec 4: 72–82; 2004: Published
online in Wiley InterScience (www.interscience.wiley.com) DOI 10.1002/tcr.20002

Key words: Thermobifida fusca; cellulose binding domain; endocellulase; exocellulase; xylanase;
xyloglucanase; Fibrobacter succinogenes; Prevotella ruminicola; protein engineering; regulation;
site directed mutagenesis; structure; synergism.

I first started studying microbial cellulases in 1982, when
Dr. Dexter Bellamy came to Cornell from General Electric
and brought a set of thermophilic cellulolytic bacteria, which
he had isolated1. On his recommendation, I chose to study
Thermobifida fusca YX (then called Thermomonospora fusca)
and set out to purify and characterize the set of extracellular
cellulases that it produced. T. fusca (Fig. 1) is a filamentous soil
bacteria that grows at 55°C in defined medium on cellulose as
a carbon source. It appears to be a major degrader of plant cell
walls in heated organic material such as compost piles and
rotting hay2.
T. fusca YX produced a very active extracellular protease
that created many proteolytic artifacts. This enzyme3 was puri-
fied, characterized, and its structural gene was cloned and
sequenced4. This enzyme resembles a lytic protease in having
an essential chaperon in an N-terminal pre-sequence and the
entire amino acid sequence shows 46% identity to the a lytic
protease precurser5. This enzyme has high activity on most pro-
 Correspondence to: Professor David B. Wilson; Phone: 607-255-5706;
Fax: 607-255-6249; Email: dbw3@cornell.edu

The Chemical Record, Vol. 4, 72–82 (2004)

72 © 2004 The Japan Chemical Journal Forum and Wiley Periodicals, Inc.
 Microbial Cellulases

sive studies of binding and synergism by T. fusca and T. reesei

Fig. 1. Microscopic image of T. fusca micelia at a magnification of 60X.
teins including hair and feathers. We were able to isolate a
stable strain, ER1, which lacked the protease that we have used
for most of our research6. Using standard purification methods,
we were able to purify and characterize six different cellulases,
a low molecular weight cellulose binding protein and a
xylanase from the culture supernatant of T. fusca ER1 grown
on cellulose7–10(Table I–II). In addition we also cloned puri-
fied and characterized an intracellular b-glucosidase11 and an
extra cellular xyloglucanase12. A major contributor to all the
research discussed in this article is my long time coworker,
Diana Irwin.
An important influence on our studies of the properties
of T. fusca cellulases is a long standing collaboration with
Dr. Larry Walker in the Department of Biological and Envi-
ronmental Engineering at Cornell. He has carried out exten-
cellulases and their catalytic and binding domains13–19. He also
developed a method for determining the rate of fragmentation
of Avicel particles using a particle counter20 and used it to show
that fragmentation is linear with time and enzyme concentra-
tion unlike the rate of reducing sugar production, which is
nonlinear with both time and enzyme concentration. He also
studied the effect of particle size on fragmentation21.
We also cloned and sequenced four endocellulase genes
from T. fusca by screening E. coli colonies containing plasmid
and phage libraries with T. fusca DNA inserts using a CMC
overlay assay22–24 and we cloned a xylanase gene using a xylan
overlay25. Since the four endocellulase genes that we cloned
encoded the four endoglucanases we had purified, it seemed
probable that we had identified all of the endocellulases that
are produced by T. fusca when it grows on cellulose. It is puz-
zling that a second family 5 cellulase gene, which we did not
find, is present in T. fusca and recently it was shown to be
expressed in E. coli and have good activity on CMC26.
We succeeded in cloning the genes that encoded the two
T. fusca exocellulases, but we had to screen plasmid libraries
with labeled oligonucleotides designed to code for the N-
terminal sequence of each protein, which we had determined,
as the enzymes had too low an activity on CMC to detect
colonies containing their genes by a CMC overlay9,10. All of
the cloned genes were sequenced and the sequences showed
that every gene encodes a family 2 cellulose binding domain
(CBD), which is at the N-terminus of Cel5A, Cel6B and
Cel48A and at the C-terminus of Cel6A, Cel9A, Cel9B and
Xyn11A (Fig. 2)27–29.
The T. fusca cellulase catalytic domains belong to four dif-
ferent families: 5, 6, 9 and 48. In the two families that con-
tained two cellulases, 6 and 9, the enzymes in each family are
very different, as one family 6 cellulase is an endocellulase
(Cel6A) while the other is an exocellulase (Cel6B) and the cat-

 David B. Wilson was born in 1940 in Cambridge, MA, USA. He received his B.A. from
Harvard in 1961 and his Ph.D. from Stanford in 1966. He was a postdoctoral fellow in the
Biophysics Department at Johns Hopkins Medical School from 1966 to 1967. He joined Cornell
University as an Assistant Professor of Biochemistry in 1967, was promoted to Associate
Professor in 1973 and to full Professor in 1984. His current research continues to focus on
the mechanisms of plant cell wall degradation by bacterial enzymes. He was elected to the
American Academy of Microbiology in 2003. 

© 2004 The Japan Chemical Journal Forum and Wiley Periodicals, Inc. 73
 THE CHEMICAL RECORD

Table I. Cellulases of Thermobifida fusca, activity of whole and truncated enzymes.

Activities (mmol CB/0 min-mmol)

Enzyme MW (kD) CMC SW FP BMCC Refs.

7
Cel 5A endo 46.3 2840 90 0.83 15
Cel 5Acd 34.5 2477 85.3 0.57* —
7,46
Cel 6A endo 43.0 355 631 0.85 6.5
Cel 6Acd 30.4 330 560 0.50* 3.2
9
Cel 6B exo-NR 59.6 0.3* 1.5 0.13* 2.0
Cel 6Bcd 45.7 0.5 1.1 0.05* 1.0
41
Cel 9A processive endo 90.4 475 202 1.03 19.1
Cel 9Acd + CBD3 68.0 488 54 0.24* 6.1
Cel 9Acd 51.4 108 6.3* 0.13* 0.15*
Cel 9Acd + CBD2 74.0 121 23.2 0.27* 0.29*
Cel 9A (D Fnll) 79.7 234 45.6 0.65 16.3
7
Cel 9B endo 101.2 5410 363 0.18* 0.42*
10
Cel 48A exo-R 104.0 0.3* 0.4* 0.07* 0.19*
Cel 48A cd 71.9 0.2* 0.2* 0.05* 0.09*
*target % digestion could not be achieved, assayed using 1.5 nmol/ml enzyme.
exo-NR = exocellulase, nonreducing end directed, exo-R = exocellulase, reducing end directed
endo = endocellulase

Table II. Other characterized carbohydrate active enzymes of T. fusca.

Enzyme MW (kD) Substrate Product Activity Km Ref.

mmol red. sugar/ min-mmol mg/ml

Xyn10B 42.3 Xylan—Birchwood mixed xylobiose 3,724 1.4 unpub
and higher MW
28
Xyn11A 31.9 xylan xylobiose and 15,600 1.1
higher MW
Xyn11Acd 19.9 xylan 14,200 2.3

mmol glu/min-mmol mM
11
Bgl1C 53.4 cellobiose pNp-glu glucose 1,497 0.35
719 0.24

mmol XGOs/min-mmol mg/ml

12
Xeg74 94.7 tamarind XG XGO’s mixed 578 3.2
Xeg74cd 79.5 875 3.9
XGO = xyloglucan oligomers
red. sugar = reducing sugar
pNp-glu = 4-nitrophenyl b-D-glucopyranoside

alytic domains have only 31% sequence identity. For family 9, all six T. fusca cellulases that are used to degrade cellulose are
one enzyme is an endocellulase (Cel9B), the other is a novel different from each other and appear to have been obtained
type of cellulase, a processive endoglucanase (Cel9A), and the from different organisms, rather than having been produced in
two catalytic domains have only 27% sequence identity. Thus, T. fusca by gene duplication.

74 © 2004 The Japan Chemical Journal Forum and Wiley Periodicals, Inc.
 Microbial Cellulases

Fig. 2. Cartoon showing the arrangement of the domains of T. fusca glycohydrolases listed in Table I and II.

It is interesting that an unrelated mesophilic bacterium,
Cellulomonas fimi, appears to utilize the same set of cellulases.
Its six cellulases have similar activities to those of T. fusca, and
their catalytic and CBD domains are in the same families30.
However, the two organisms probably did not obtain their
cellulase genes from a common relative, as the location of the
family 2 CBD differs for five of the six enzymes (Table III). In
contrast, Streptomyces coelicolor, which grows weakly on cellu-
lose and lacks one of the six T. fusca cellulase genes, has the
family 2 CBD encoded in the same location as in the corre-

© 2004 The Japan Chemical Journal Forum and Wiley Periodicals, Inc. 75
 THE CHEMICAL RECORD
Table III. Cellulase families found in similar bacterial strains.
Thermobifida fusca
CBD
Enzyme Type Location Type
Cel 5A Endo N-term Endo
Cel 6A Endo C-term Endo
Cel 6B Exo N-term Exo
Cel 9A Processive C-term Processive
Cel 9B Endo C-term Endo
Cel 48A Exo N-term Exo
sponding five T. fusca genes (Table III). Thus, these related
organisms appear to have obtained their cellulase genes from
a common ancestor.
Studies of the activity of modified T. fusca cellulases,
lacking the family 2 CBD, showed that this domain is not
required for activity on the soluble substrate carboxymethyl
cellulose (CMC) and it has only a small role in activity on
amorphous cellulose, except for Cel9A. However, it is impor-
tant for activity on crystalline substrates, such as filter paper or
bacterial microcrystalline cellulose (BMCC) (Table I). Fur-
thermore, studies of the effect of the CBD on synergism
showed that it is much more important for exocellulases than
for endocellulases (Table IV). This work confirmed the impor-
tant role that CBDs play in crystalline cellulose degradation
and the presence of a binding domain appears to be a general
adaptation for enzymes that degrade insoluble substrates, as
substrate binding domains have now been found on most
enzymes that catalyze the hydrolysis of insoluble substrates31.
Another important finding from our cellulase research is
the existence of two classes of exocellulases. One, with
members in family 7 and family 48, attacks the reducing ends
of cellulose molecules and the other, with members so far only
in family 6, attacks the non-reducing ends32. Furthermore, we
showed that two exocellulases only give synergism when they
are from different classes. Our research also confirmed the
existence of bacterial exocellulases, which was first shown for
C. fimi, Cel6B33. The T. fusca exocellulase that attacks the non-
reducing end of cellulose chains, Cel6B, is very similar in its
activity to the fungal exocellulases in this class. The main dif-
ference is that the T. fusca enzyme is more thermostable and
has a broad pH optimum centered at pH 7, while the fungal
enzymes have narrow pH optima centered at pH 434. In con-
trast, family 48 exocellulases have a very different structure
from the comparable fungal enzyme, T. reesei Cel7A35. Fur-
thermore, Cel48A has very different activity, being much less
active than Cel7A, both by itself and in synergistic mixtures.
Cel7A is responsible for the higher activity of T. reesei crude
cellulase, as compared to T. fusca crude cellulase, as addition
76
Cellulomonas fimi Streptomyces coelicolor
CBD % identity of CBD % identity of
Location cd to T. fusca Location cd to T. fusca
C-term 21 N-term 50
N-term 42 C-term 49
C-term 55 N-term 65
C-term 80 — —
N-term 59 C-term 60
C-term 33 N-term 57
of Cel7A to T. fusca crude cellulase doubles its specific activ-
ity on crystalline cellulose8.
At this time, it is not known why aerobic bacteria and
aerobic fungi use completely different enzymes to attack the
reducing end of cellulose. One possibility is that aerobic cel-
lulolytic bacteria attack different types of cell walls than aerobic
fungi and that the family 48 enzymes are more active on the
cell walls that bacteria attack than the family 7 enzymes. We
have started to test this idea by studying the digestion of
tomato cell walls by T. fusca hydrolases. Preliminary studies
have shown that a mixture of cellulases plus a xyloglucanase
have low activity on plant cell walls, but that T. fusca crude
cellulase, which contains these enzymes along with other pro-
teins, can degrade the cell walls12. Once we have identified the
minimal set of enzymes needed to degrade the cellulose in
plant cell walls, we can test the effect of replacing T. fusca
Cel48A with T. reesei Cel7A in the hydrolysis of different types
of plant cell walls.
Detailed studies of synergism by T. fusca cellulases, and by
T. fusca cellulases plus T. reesei Cel6A and Cel7B have provided
strong evidence that synergism does not require specific inter-
actions between the individual cellulases. It appears that two
cellulases give synergism if and only if they attack insoluble
cellulose at different sites and if they create new sites for each
other as a result of their activity. It is interesting that pretreat-
ment of crystalline cellulose with an endocellulase makes it a
better substrate for exocellulases36,37, but pretreatment with an
exocellulase does not increase the activity of endocellulases on
the pretreated cellulose. We have shown that in synergistic mix-
tures of endocellulases and exocellulases, the activity of the
endocellulase is increased8 so that the stimulation of the endo-
cellulase must result from transient changes in the cellulose
structure, such as displacement of a segment of a cellulose
strand, which returns to the original inaccessible state if the
endocellulase does not bind to it quickly. A careful study of
the reason for the non-linearity of the rate of cellulose degra-
dation by T. fusca cellulases with both time and amount of
enzyme38 showed that it was the heterogeneity of the substrate
© 2004 The Japan Chemical Journal Forum and Wiley Periodicals, Inc.
 Microbial Cellulases

Table IV. Synergistic activity of binary mixtures of T. fusca enzymes on filter paper.

Molar Ratio Filter PaperActivity

Enzymes of Enzymes (mmol CB/min-mmol) References

Endo + Exo
8
Cel5A + Cel6B 1:4 2.85
Cel5Acd + Cel6B 1:4 2.15 ≤
Cel5A + Cel6Bcd 1:4 0.62* ≤
Cel5Acd + Cel6Bcd 1:4 0.63* ≤
7
Cel6A + Cel6B 2:1 3.8
Cel6Acd + Cel6B 2:1 2.0 ≤
9
Cel5A + Cel48A 1:4 0.76*
Cel6A + Cel48A 1:4 0.75* ≤

Exo + Exo
9
Cel6B + Cel48A 1:1 0.52*

Processive Endo + Endo

7
Cel9A + Cel5A 2:1 2.1
Cel9A + Cel6A 3:1 2.2 ≤

Processive Endo + Exo

9
Cel9A + Cel48A 1:4 1.8
7
Cel9A + Cel6B 1:1 2.0

Multiple Enzyme Mixtures

9
Cel9A + Cel5A + Cel48A 1:1:8 3.3
Cel6B + Cel9A + Cel5A + Cel48A 4:1:1:4 7.5 ≤
Cel9B + Cel6A + Cel6B + Cel9A + Cel5A 1:1:8:1:1 10.2 ≤
Cel9B + Cel6A + Cel6B + Cel9A + Cel5A + Cel48A 1:1:4:1:1:4 11.6 ≤
*5% digestion could not be achieved by these mixtures.

that caused the non-linearity, rather than product inhibition
or cellulase instability.
In a collaboration with Dr. Andrew Karplus at Cornell,
the 3-dimensional structure of the T. fusca Cel6A catalytic
domain was determined by X-ray crystallography39. The struc-
ture was ultimately refined to 1.18Å and it was the highest res-
olution structure of a 30,000 MW protein when published
(Fig. 3). The structure showed that Cel6A, an endocellulase,
has an open active site cleft as had been predicted by Rouvi-
nen et al.40, who had published the first cellulase structure, that
of Trichoderma reesei Cel6A, an exocellulase, with its active site
in a tunnel. At this time, these structural differences have been
found for all true endocellulases and exocellulases, although in
some exocellulases only part of the active site cleft is enclosed35.
A major finding from our research is the identification of
a new type of cellulase, Cel9A, which is a processive endoglu-
canase. This enzyme has relatively high activity on CMC, like
an endocellulase but it also produces 87% soluble products
from filter paper and only 13% insoluble products, while true
endocellulases produce between 30 and 40% insoluble prod-
ucts from filter paper8. The structure of Cel9Acd, determined
by Dr. Sakon in Dr. Karplus’ laboratory, contains a family 9
catalytic domain rigidly attached to a family 3c CBD, with the
active site cleft in the catalytic domain aligned with the poten-
tial cellulose binding site on the CBD (Fig. 4)41. The structure
of Cel9Acd was also determined with several oligosaccharides
bound in the active site. Comparison of liganded structures
with the free structure identified the catalytic water in the free
structure and showed that several residues moved upon sub-
strate binding41. The enzyme appears to make an initial endo-
cellulolytic cleavage of a cellulose molecule, and then the
fragment bound to the CBD slides into the active site and
cellotetraose units are cleaved processively from the non-
reducing end of the bound chain. This mechanism was con-
firmed by constructing two other forms of Cel9A using genetic
engineering. One lacked the family 3 CBD, while the other
contained only the catalytic domain. The activities of the
constructs (Table I) show that removal of the family 3 CBD

© 2004 The Japan Chemical Journal Forum and Wiley Periodicals, Inc. 77
 THE CHEMICAL RECORD
Fig. 3. Cel 6A structure made using PDB file 1tml with cellotetraose (red)
modeled into the open cleft active site. Asp 117 (black) is the catalytic acid.
Other important catalytic residues are Asp 79 (cyan), Arg 78 (pink), Tyr 73
(purple), Asp 156 (gray), and Asp 265 (blue)48,49.
converts Cel9A into a true endocellulase, completely abolish-
ing its processive activity as predicted by our mechanism42.
This type of enzyme has been found in both anaerobic and
aerobic bacteria, but not in eucaryotic organisms43,44. Cel9A
can give synergism with endocellulases and both classes of exo-
cellulases, providing further evidence that Cel9A is a unique
type of cellulase (Table IV).
We have carried out extensive site-directed mutagenesis
studies on Cel6A, both to determine its catalytic mechanism
and to try to determine how it hydrolyzes crystalline cellulose.
This enzyme is ideal for these studies as we have a high-
resolution structure of the catalytic domain as well as a struc-
ture with cellobiose bound in subsites Glc(-2) and Glc(-1).
Furthermore, Cel6Acd is expressed in E. coli at a high level
(170 mg/l) so that it is relatively simple to produce the mutant
enzymes. Finally it has been possible to determine the Kd for
both methylumbelliferyl glycoside and oligosaccharide binding
to Cel6A by fluorescence quenching45. At this time, we have
constructed fifty-eight mutants in 33 of the 306 total residues
in this protein46–48. In addition to determining the activities of
the mutant enzymes on different types of cellulose and some
oligosaccharides, we determined their binding affinities to
several oligosaccharides and methylumbelliferyl glycosides.
The results from all these studies combined with molecular
dynamics modeling of Cel6A bound to cellotetraose allowed
us to propose a detailed mechanism for cellulose hydrolysis by
Cel6A (Fig. 3)49. Surprisingly, our model differs slightly from
the model proposed for T. reesei Cel6A50, as we find that mutat-
78
Fig. 4. Cel9A structure using PDB file 4tf4 showing six glucose residues in
the active cleft (red) and the beta strand 2 (Thr475 to Gln 486) of the CBD3c
in solid 3D (gold). The region deleted for the block mutant, Thr245- Arg252
is colored black. The calcium atoms are shown in black; Glu424, the catalytic
acid, is blue. Asp55 and Asp58, the putative catalytic bases, are shown in
purple. An extended cellulose chain could interact well with the conserved
residues of beta strand 2 and the active cleft of the catalytic domain. The
product is released from the Glc(-1) to Glc(-4) sites41.
ing Ser85 to Ala has no effect on activity, yet in their model
the equivalent residue is hydrogen bonded to the catalytic
water molecule. This result shows that either the two enzymes
use somewhat different mechanisms or their model, which is
based on detailed structural information, is wrong, which
seems unlikely.
Some of the Cel6A mutations reduce activity on CMC
and amorphous cellulose to a much greater extent than activ-
ity on crystalline cellulose. This shows that the rate limiting
step for crystalline cellulose hydrolysis differs from that of the
other substrates. This is not surprising since Cel6A has a spe-
cific activity on crystalline cellulose that is 1/500 that on the
other substrates. I believe that the rate limiting step on crys-
talline cellulose is probably the binding of a cellulose chain into
the active site and that once it is bound it is rapidly cleaved.
A key question is which residues function in the rate limiting
step? At this time, we only have succeeded in finding one
mutation that reduces activity on crystalline substrates to a
greater extent than it reduces activity on other substrates,
which is the result expected for a residue that only participates
in the rate limiting step. Even with this residue, only one of
© 2004 The Japan Chemical Journal Forum and Wiley Periodicals, Inc.
 Microbial Cellulases
Table V. Cel6A mutants resulting in higher CMC activity.
Cel6A mutant % CMC Activity46
wild type 100
Arg 237 Ala 176
Glu 263 Asp 236
Glu 263 Gln 162
Glu 263 Gly 110
Lys 259 His 373
Arg 237 Ala, Glu 263 Asp 290
Arg 237 Ala, Glu 263 Gln 363
His 159 Ser 168
Lys 259 His 485
three amino acid substitutions showed this effect. One very
interesting mutation, Trp16 to Ile, lowered activity on amor-
phous cellulose more than activity on CMC or filter paper.
This differential effect was increased in the catalytic domain.
The structure of this mutant cd was determined by X-ray crys-
tallography and the mutation created minor shifts in a number
of regions (A. Larsson, personal communication).
We identified five residues where mutations produce
enzymes with significantly higher activity on the soluble
substrate carboxymethyl cellulose (Table V)47. Every mutant
enzyme, which has higher activity on CMC, has weaker
binding to oligosaccharides. Unfortunately, none of the
mutant enzymes have increased activity on insoluble
substrates.
There are a number of experiments that show that the
loop in Cel6Acd from residue 73 to 87 undergoes a major con-
formational shift during or after substrate binding and that this
shift is essential for catalysis. Furthermore, the position of the
loop in the X-ray structure is not that of the active conforma-
tion. The evidence that the structure is not the active structure
of Cel6A is the fact that mutation of Asp79 to Asn causes a
30–50 fold reduction in activity on CMC or amorphous cel-
lulose48, even though in the X-ray structure, Asp79 is 11.3Å
from the site of bond cleavage. In T. reesei Cel6A, an exocel-
lulase, the corresponding residue is only 7Å from the site of
cleavage40.
One experiment that shows that the conformational
change is essential for activity is mutation of either of two adja-
cent Gly residues (86, 87) at one end of the loop to Ala, as
both mutations significantly decrease Cel6A activity (16 and
4.5% residual activity, respectively)46. Since these residues are
not buried and are not in the active site, the main effect of
either mutation is to reduce the ease of loop movement by
increasing the size of the side chain. Removal of the disulfide
bond containing another loop residue, by mutating both
Cys80 and Cys125 to Ser also reduced activity (12% residual
activity), possibly by changing the final position of the loop46.
© 2004 The Japan Chemical Journal Forum and Wiley Periodicals, Inc.
Further evidence for loop movement comes from mea-
surements of the thermodynamics of MU(Glc)2 binding to
Cel6Acd and two family 6 cellulases where the loop is less
mobile. For Cel6A, DS = –46.8 (J/mol K)(41), while for T.
fusca Cel6B and T. reesei Cel6A the values are 53.4 and 63,
respectively51. This dramatic difference can be explained if sub-
strate binding restricted the mobility of the residues in the
Cel6A loop, thus decreasing the entropy while there is only a
small change in the mobility of the loop in the exocellulases.
Dr. Anna Larsson in Alwyn Jones’ laboratory has deter-
mined several new crystal structures for T. fusca Cel6A WT
and the mutant enzymes (personal communication). The wild-
type structure differs from our original structure in that most
of the loop (residues 81–87) is not visible as it is too mobile,
but the rest of the molecule has a structure that is the same as
the original one. A second structure is of a mutant enzyme
Tyr73Ser that has extremely low activity. The structure of this
mutant enzyme was also determined with a cellotetraose ana-
logue bound in the active site. This structure also lacks residues
81–87, possibly because the mutation prevents the conforma-
tional change, explaining the loss of activity. A structure of the
Asp117Ala mutant enzyme crystallized in the presence of cel-
lotriose shows a glucose residue bound in the Glc(–3) subsite
which is the first evidence for this subsite in Cel6A.
A similar conformational change has been shown to occur
in two family 5 endoglucanases, Clostridium thermocellum
CelC52 and Humicola insolens endoglucanase 553. In each
enzyme, the loop in the unliganded structure is less ordered
than in the liganded structure and the loop moves to partially
cover the substrate in the active site cleft. Family 5 and family
6 enzymes are similar in structure and both loops contain two
adjacent Gly residues at one end.
We also have made site-directed mutants in the exocellu-
lase Cel6B, using a model structure based on the structures
of the related enzymes, T. fusca Cel6A and T. reesei Cel6A.
Twenty-three mutant genes in eleven different residues were
constructed and the expressed proteins were purified and char-
acterized51. Two pairs of mutations were constructed to create
new disulfide bonds that joined the two loops, which form the
active site tunnel (N233C/0506C and 230C/0512C). Both
mutant enzymes contained the new bond and both retained
activity on all substrates tested, showing that opening of the
loops is not required for activity as had been suggested54.
Unfortunately, neither mutant enzyme showed increased ther-
mostability, as has been seen when a new disulfide bond was
created in some proteins. Four Gly residues were mutated to
try to increase thermostability51, but none of them did. Sur-
prisingly, two of these mutations increased activity on FP and
BMCC and a double mutant containing both mutations had
even higher activity on FP (twice wild type (WT)). Unfortu-
nately, this increase in activity did not show up in synergistic
mixtures with an endoglucanase. Several mutants had signifi-
79
 THE CHEMICAL RECORD
cantly reduced inhibition by cellobiose, but they also had
slightly lower activity than the WT enzyme.
Site-directed mutagenesis has been used to determine the
role of eight key residues in Cel9A and two deletion mutants
were constructed55. One removed a FnIII domain (Fig. 2),
while the other removed seven residues that block one end of
the active site cleft (Fig. 4). The properties of the point mutants
confirmed that Glu424 functions as the catalytic acid and that
both Asp55 and Asp58, which bind the catalytic water, are
required for the catalytic base function. These residues are con-
served in all members of family 9 and this is the only hydro-
lase family in which two acidic residues bind the catalytic
water. When Tyr206 was replaced by Phe, Cel9A retained
some activity and binding, but when it was replaced by Ser,
activity and binding were lost. Mutation of Tyr318 greatly
reduced binding, but only partially reduced activity. However,
both of these mutant enzymes were non-processive, showing
that high affinity binding in the Glc(-3) to Glc(-1) subsites is
essential for processivity. Removal of the FnIII domain
decreased activity and processivity to 65% of wild type and
decreased binding to BMCC slightly. When the two FnIII
domains were removed from Clostridium thermocellum CbhA,
a family 9 exocellulase, activity was reduced to 50%56. It is not
clear how this domain affects activity but its primary role seems
to be in the digestion of crystalline cellulose. Evidence for
FnIII domain activity disrupting crystalline cellulose was also
presented by Nidetzky et al.57 Deletion of the block at the end
of the T. fusca Cel9A active site cleft did not change activity
or products but did increase binding to BMCC. This result
shows that production of cellotetraose during processive
hydrolysis does not result from the peptide block.
T. fusca regulates cellulase synthesis in two ways, induc-
tion by cellobiose and carbon source repression58–60. Induction
increases the cellulase level about 12 fold and derepression
increases the level about nine fold. Growth on cellulose gives
the highest cellulase level as it simultaneously induces and
derepresses, since the products of cellulose degradation are
consumed by T. fusca as rapidly as they are produced. Both
mechanisms act at the level of transcription for Cel5A, the only
enzyme that was tested59. All of the six T. fusca cellulase genes
contain the same upstream fourteen base inverted repeat
sequence TGGGAGCGCTCCCA. In the four endocellulase
genes, there is a single copy of this sequence, which starts
between 36 and 63 bases upstream of the start codon, while
in the two exocellulase genes there are two copies that are from
200 to 350 bases from the start codon60. This sequence is the
binding site for a regulatory protein, CelR, that has been
cloned, expressed, purified and characterized61,62. This protein
is a member of the lactose repressor family and it binds cel-
lobiose with a mM Kd62. Furthermore, cellobiose lowered the
affinity of CelR to DNA containing the 14 base sequence. A
mutant that was constitutive for cellulase synthesis had a muta-
80
tion in the CelR gene that lowered its affinity for the 14 base
sequence providing conclusive evidence that CelR functions in
the regulation of cellulase synthesis63. There are two findings
that suggest that CelR is not just a repressor. One is the fact
that the two exocellulase genes contain multiple copies of the
CelR binding sequence, which are located at different
upstream sites from the single endocellulase sites and the exo-
cellulases are made in equal amounts while each of the endo-
cellulases is present at 1/3 to 1/6 the level of an exocellulase.
There are no strong similarities in the potential cellulase pro-
moter sequences, so it is possible that CelR regulates the level
of synthesis of the cellulases. The other very surprising finding
is that the level of CelR is much lower in T. fusca grown on
glucose or xylan than in T. fusca grown on cellobiose or cellu-
lose. We are still working to understand exactly how CelR
functions and if it is participating in both controls.
The T. fusca genome was sequenced in 2000 by the
U.S. Department of Energy Joint Genome Institute
(http://www.jgi.doe.gov/JGI_microbial/html/index.html). It
is 3.7 MB in size and in addition to the six cellulase genes we
had identified, it contained a second family 5 cellulase gene
and a family 74 gene, but neither of these genes contain a
perfect upstream CelR binding site. We have expressed and
characterized the family 74 enzyme and it is a xyloglucanase
(Xeg74) with low activity on CMC12. It is produced by cells
grown on cellulose, xylan or corn fiber, but not by cells grown
on glucose or cellobiose. Xeg74 contains a family 2 CBD and
it appears to function to remove xyloglucan from cellulose in
plant cell walls, allowing the cellulases better access to the cel-
lulose (Fig. 5)12. T. fusca can not grow on xyloglucan as a
carbon source, even though it completely degrades it to
oligosaccharides. It seems likely that a xylanase cloned from
T. alba64 which has a family 2 CBD, has a similar function,
removing xylan to allow the cellulases better access to the cel-
lulose in plant cell walls, as its synthesis is induced by cellobiose
and not by xylan. T. fusca contains a number of other xylanase
genes that are induced by xylan and these probably function
to allow T. fusca to grow on xylan65.
While much of my cellulase research has focused on T.
fusca cellulases, I have collaborated with Dr. James Russell in
the Department of Microbiology at Cornell to study cellulose
digestion by rumen bacteria. We started using genetic engi-
neering to try to construct a cellulolytic rumen bacterium
that could degrade cellulose at pH values below 5. Prevotella
ruminicola was chosen as the starting organism as it grows
readily on cellulodextrins at low pH. We cloned, expressed and
characterized a family 5 endoglucanase gene from P. rumini-
cola66, which did not encode a CBD and had low activity on
insoluble cellulose. Further work showed that the enzyme
expressed in E. coli was only the C-terminal domain of the
P. ruminicola protein and that P. ruminicola produced two
endoglucanases that only differed in their N-terminal
© 2004 The Japan Chemical Journal Forum and Wiley Periodicals, Inc.
 Microbial Cellulases

TFSF-GBX crystalline cellulose, allowing it to be hydrolyzed by

TFSF-GBG endocellulases70.
Cel Mix - GBX
300 Cel Mix + XEG74 - GBX
The T. fusca research described in this paper has been
mg reducing sugar produced

250 supported throughout by the U.S. Department of Energy,

Basic Energy Sciences Program and additional support has
200
come from the D.O.E. National Renewable Energy
150 Laboratory and the USDA competitive Grants Program.

100

50
REFERENCES
0
[1] Bellamy, W.D. Dev Ind Microbiol 1977, 18, 249–254.
0 10 20 30 40 50 60 70
[2] Bachmann, S.L.; McCarthy, A.J. Appl Environ Microbiol
mg protein
1991, 57, (8) 2121–2130.
Fig. 5. Gluconoacetobacter xylinus synthesizes microcrystalline cellulose pelli- [3] Gusek, T.W.; Kinsella, J.E. Biochem J 1987, 246, 511–517.
cles (GBG) when grown on glucose. If tamarind xyloglucan is present in the [4] Lao, G.; Wilson, D.B. Appl Environ Microbiol 1996, 62, (11)
medium, cellulose/xyloglucan composite pellicles (GBX) are produced71. 4256–4259.
Although the mixture of Cel5A + Cel6A + Cel9A + Cel48A (Cel mix) was not [5] Epstein, D.M.; Wensink, P.C. J Biol Chem 1988, 263,
able to digest GBX, the combination of 4.7 mg of XEG74 and the Cel mix 16568–16590.
degraded this substrate as well as the concentrated crude T. fusca supernatant
[6] Wilson, D.B. Methods Enzymol 1988, 160, 314–323.
(TFSF)12.
[7] Calza, R.E.; Irwin, D.C.; Wilson, D.B. Biochem 1985, 24,
7797–7804.
[8] Irwin, D.; Walker, L.; Spezio, M.; Wilson, D. Biotech Bio-
67
engin 1993, 42, 1002–1013.
sequence . It is interesting that the larger enzyme appears to [9] Zhang, S.; Lao, G.; Wilson, D.B. Biochemistry 1995, 34, (10)
be produced by a ribosomal hop, which skips several stop 3386–3395.
codons67. A family 2 CBD was fused to the C-terminus of the [10] Irwin, D.C.; Zhang, S.; Wilson, D.B. Eur J Biochem 2000,
E. coli endoglucanase and the fusion protein had seven times 267, (16) 4988–4997.
higher activity on both ball milled cellulose and amorphous [11] Spiridonov, N.A.; Wilson, D.B. Curr Microbiol 2001, 42,
cellulose than the enzyme lacking the CBD68. The modified 295–301.
gene was mated into P. ruminicola, but it was not expressed. [12] Irwin, D.; Cheng, M.; Xiang, B.; Rose, J.K.C.; Wilson, D.B.
Another joint project is a study of the Fibrobacter suc- Eur J Biochem 2003, 270, 3083–3091.
cinogenes mechanism for cellulose degradation. Extensive [13] Bothwell, M.K.; Walker, L.P.; Wilson, D.B.; Irwin, D.C.;
Price, M. Biomass and Bioenergy 1993, 4, (4) 293–299.
cloning primarily by Dr. Cecil Forsberg has identified many
[14] Walker, L.P.; Belair, C.D.; Wilson, D.B.; Irwin, D.C. Biotech
endoglucanase genes, but none of them have activity on crys- Bioengin 1993, 42, 1019–1028.
talline cellulose69. Biochemical studies also have failed to [15] Bothwell, M.K.; Wilson, D.B.; Irwin, D.C.; Walker, L.P.
measure crystalline cellulose activity in F. succinogenes extracts. Micro Technol 1997, 20, 411–417.
The sequence of the F. succinogenes genome has been [16] Wilson, D.B. Crit Rev Biotech 1992, 12, 45–63.
determined by TIGR (http://www.tigr.org/tdb/mdb/ [17] Bothwell, M., Daughhetee, S., Chaua, G., Wilson, D.B.
mdbinprogress.html) financed by a USDA grant to the North Walker, L. Bioresour Technol 1997, 60, 169–178.
American Consortium for Genomics of Cellulolytic Rumen [18] Jeoh, T.; Wilson, D.B.; Walker, L.P. Biotechnol Prog 2002, 18,
Bacteria. The sequence confirms the cloning studies as there 760–769.
are many endoglucanase genes in the genome, but no genes [19] Jung, H.; Wilson, D.B.; Walker, L.P. Biotechnol Bioeng 2002,
are present in the families that contain the processive cellulases 80, (4) 380–392.
[20] Walker, L.P.; Wilson, D.B.; Irwin, D.C. Enzyme Microb
which are essential for cellulose degradation in other known
Technol 1990, 12, (May) 378–386.
cellulolytic microorganisms. Despite this fact, F. succinogenes [21] Peters, L.E.; Walker, L.P.; Wilson, D.B.; Irwin, D.C. Bioresour
grows more rapidly on cellulose as a sole carbon source Technol 1991, 35, 313–319.
than any known organism. This suggests that F. succinogenes [22] Collmer, A.; Wilson, D.B. Bio/Technology 1983, 1, 594–601.
uses a different mechanism for cellulose degradation than [23] Ghangas, G.S.; Wilson, D.B. Applied and Envir Micro 1988,
other known organisms. One possibility is that it produces 54, (No.10) 2521–2526.
proteins like the CX factor proposed by Dr. Reese that disrupt [24] Hu, Y.J.; Wilson, D.B. Gene 1988, 71, 331–337.

© 2004 The Japan Chemical Journal Forum and Wiley Periodicals, Inc. 81
 THE CHEMICAL RECORD

[25] Ghangas, G.S.; Hu, Y.J.; Wilson, D.B. J Bacteriol 1989, 171, [48] Wolfgang, D.W.; Wilson, D.B. Biochemistry 1999, 38, (30)
2963–2969. 9746–9751.
[26] Posta, K.; Beki, E.; Kukolya, J.; Hornok, L. Personal Com- [49] André, G.; Kanchanawong, P.; Palma, R.; Cho, H.; Deng, X.;
munication 2003. Irwin, D.; Himmel, M.E.; Wilson, D.B.; Brady, J.W. Protein
[27] Lao, G.; Ghangas, G.S.; Jung, E.D.; Wilson, D.B. J Bact 1991, Eng 2003, 16, 125–134.
173, (11) 3397–3407. [50] Koivula, A.; Ruohonen, L.; Wohlfahrt, G.; Reinikainen, T.;
[28] Jung, E.D.; Lao, G.; Irwin, D.; Barr, B.K.; Benjamin, A.; Teeri, T.T.; Piens, K.; Claeyssens, M.; Weber, M.; Vasella, A.;
Wilson, D.B. Appl Environ Microbiol 1993, 59, (9) Becker, D.; Sinnott, M.L.; Zou, J.Y.; Kleywegt, G.J.; Szarden-
3032–3043. ings, M.; Stahlberg, J.; Jones, T.A. J Am Chem Soc 2002, 124,
[29] Irwin, D.; Jung, E.; Wilson, D.B. Appl Environ Micrbiol (34) 10015–10024.
1994, 60, (3) 763–770. [51] Zhang, S.; Irwin, D.C.; Wilson, D.B. Eur J Biochem 2000,
[30] Warren, R.A.J. Annu Rev Microbiol 1996, 50, 183–212. 267, 3101–3115.
[31] Tomme, P.; Warren, A.J.; Miller Jr., R.C.; Kilburn, D.G.; [52] Dominguez, R.; Souchon, H.; Lascombe, M.; Alzari, P.M. J
Gilkes, N.R. Cellulose-binding domains: classification and Mol Biol 1996, 257, 1042–1051.
properties. In: Saddler, J.N., Penner, M.H., editors. Enzymatic [53] Davies, G.J.; Tolley, S.P.; Henrissat, B.; Hjort, C.; Schulein,
Degradation of Insoluble Carbohydrates. Washington, D.C.: M. Biochemistry 1995, 34, 16210–16220.
American Chemical Society, 1995. pp. 142–163. [54] Boisset, C.; Fraschini, C.; Schülein, M.; Henrissat, B.; Chanzy,
[32] Barr, B.K.; Hsieh, Y.-L.; Ganem, B.; Wilson, D.B. Biochem- H. Appl Environ Microbiol 2000, 66, (4) 1444–1452.
istry 1996, 35, 586–592. [55] Zhou, W.; Irwin, D.C.; Escovar-Kousen, J.; Wilson, D.B. in
[33] Meinke, A.; Gilkes, N.R.; Kwan, E.; Kilburn, D.G.; Warren, progress.
R.A.J.; Miller Jr., R.C. Mol Microbiol 1994, 12, (3) 413–422. [56] Kataeva, I.A.; Seidel III, R.D.; Shah, A.; West, L.T.; Li, X.-L.;
[34] Saloheimo, M.; Lehtovaara, P.; Penttila, M.; Teeri, T.T.; Ljungdahl, L.G. Aplied and Environmental Technology 2002,
Stahlberg, J.; Johansson, G.; Pettersson, G.; Claeyssens, M.; 68, (9) 4292–4300.
Tomme, P.; Knowles, J.K. Gene 1988, 63, 11–22. [57] Nidetzky, B.; Steiner, W.; Hayn, M.; Claeyssens, M. Biochem
[35] Parsiegla, G.; Reverbel-Leroy, C.; Tardif, C.; Belaich, J.P.; J 1994, 298, 705–710.
Driguez, H.; Haser, R. Biochemistry 2000, 39, 11238–11246. [58] Lin, E.; Wilson, D.B. Appl Environ Microbiol 1987, 53,
[36] Nidetzky, B.; Steiner, W.; Claeyssens, M. Biochem J 1994, 1352–1357.
303, 817–823. [59] Lin, E.; Wilson, D.B. J of Bacteriology 1988, 170, (9)
[37] Jeoh, T. Modeling and Analysis of Cellulose Hydrolysis by 3838–3842.
Thermobifida fusca Cel5A, Cel6B, and Cel9A and Binary Mix- [60] Spiridonov, N.A.; Wilson, D.B. J Bacteriol 1998, 180, (14)
tures of These Cellulases. Biological and Environmental Engi- 5463–5467.
neering. Ithaca: Cornell University, 2004. [61] Lin, E.; Wilson, D.B. J Bacteriol 1988, 170, 3843–3846.
[38] Zhang, S.; Wolfgang, D.E.; Wilson, D.B. Biotechnol Bioeng [62] Spiridonov, N.A.; Wilson, D.B. J Biol Chem 1999, 274, (19)
1999, 66, (1) 35–41. 13127–13132.
[39] Spezio, M.; Wilson, D.B.; Karplus, P.A. Biochemistry 1993, [63] Spiridonov, N.A.; Wilson, D.B. J Bact 2000, 182, (1)
32, 9906–9916. 252–255.
[40] Rouvinen, J.; Bergfors, T.; Teeri, T.; Knowles, J.K.C.; Jones, [64] Blanco, J.; Coque, J.J.; Velasco, J.; Martin, J.F. Appl Micro-
T.A. Science 1990, 249, 380–386. biol Biotechnol 1997, 48, (2) 208–217.
[41] Sakon, J.; Irwin, D.; Wilson, D.B.; Karplus, P.A. Nature Struc- [65] Kim, J.H.; Irwin, D.; Wilson, D.B. in progress.
tural Biology 1997, 4, (10) 810–817. [66] Matsushita, O.; Russell, J.B.; Wilson, D.B. J Bacteriol 1990,
[42] Irwin, D.; Shin, D.; Zhang, S.; Barr, B.K.; Sakon, J.; Karplus, 172, 3260–3630.
P.A.; Wilson, D.B. J Bacteriol 1998, 180, (7) 1709–1714. [67] Matsushita, O.; Russell, J.B.; Wilson, D.B. J Bacteriol 1991,
[43] Hazelwood, G.P.; Davidson, K.; Laurie, J.I.; Huskisson, N.S.; 173, 6919–6929.
Gilbert, H.J. J Gen Microbiol 1993, 139, (2) 307–316. [68] Maglione, G.; Matsushita, O.; Russell, J.B.; Wilson, D.B. Appl
[44] Meinke, A.; Schmuck, M.; Gilkes, N.R.; Kilburn, D.G.; Environ Microbiol 1992, 58, 3593–3597.
Miller, R.C.J.; Warren, R.A.J. Glycobiol 1992, 2, (4) 321–326. [69] Iyo, A.H.; Forsberg, C.W. Appl Environ Microbiol 1999, 65,
[45] Barr, B.K.; Wolfgang, D.E.; Piens, K.; Claeyssens, M.; Wilson, (3) 995–998.
D.B. Biochemistry 1998, 37, (26) 9220–9229. [70] Din, N.; Gilkes, N.R.; Tekant, B.; Miller, R.C.J.; Warren,
[46] Zhang, S.; Wilson, D.B. J Biotechnology 1997, 57, 101–113. R.A.J.; Kilburn, D.G. Bio/Tech 1991, 9, 1096–1099.
[47] Zhang, S.; Barr, B.K.; Wilson, D.B. Eur J Biochem. 2000, 267, [71] Whitney, S.E.C.; Brigham, J.E.; Darke, A.H.; Reid, J.S.G.;
244–252. Gidley, M.J. Plant J 1995, 8, (4) 491–504.

82 © 2004 The Japan Chemical Journal Forum and Wiley Periodicals, Inc.