Evolving Enzymes for Highly Efficient Chemical Processes

Pascal Dnkelmann

Talk Outline
Introduction to Codexis and our biocatalysis paradigm
What do we do & how do we do it? or What is directed evolution and what are the benefits?

Real examples of process improvement on pharmaceutical targets

Atorvastatin Panel Plates as an investigational tool Ezetimibe Montelukast

Summary

Codexis Biocatalysis Paradigm

Design the desired conceptual biocatalytic chemical process Make it green-by-design High substrate loadings / low organic solvent Minimise / eliminate nasty reagents High selectivity / low by-products & less purification

Evolve the biocatalyst for fitness to enable the desired process Iteratively improve the performance of the biocatalyst via genetic modification (Molecular BreedingTM)

Codexis Biocatalysis Paradigm

Bring together several scientific specialties to design and evolve a proprietary biocatalyst and a chemical process in tandem
Molecular Biology Process Chemistry Analytical Science

Combinatorial / HTP techniques Informatics

HITS & Information

Optimised Biocatalyst & Process

Iteration(s)

Allows for unconventional biocatalytic processes which solve specific problems

How MolecularBreedingTM Improves Genes

Genes/Enzymes Random fragmentation provides many potential crossover points Recombination in correct order provides a high % active enzymes
Shuffle, Screen

mutations good neutral bad

Population based shuffling combines fragments from multiple parents Screening identifies genes with novel combinations of beneficial mutations and/or fewer deleterious mutations Rapid large improvements are obtained

etc.
Shuffle, Screen

The ProSAR-Driven Methodology

At any point there are approximately 50 mutations under investigation in the hopper These are tested in combinatorial libraries and the function and the sequences of the resulting proteins are analyzed Mutations are parsed into four classes: 1. Beneficial are fixed into the population 2. Potentially beneficial are sent back into the hopper for retesting 3. Deleterious are discarded 4. Neutral, which have no effect on protein function, are discarded The amount of diversity under investigation is maintained by adding novel diversity discovered in rational design, homologous sequences, saturation or PCR mutagenesis libraries

State-of-the-Art Enzyme Evolution

a) Obtain sequenceactivity data

b) Statistical Analysis (ProSAR)

c)

Beneficial Deleterious Neutral

d) e) f)

New backbone Existing diversity New diversity

New libraries

Directed Evolution - Benefits

Substrate Specificity
Relative activity
100 10 1 .01 .001
S1 S2

Selectivity
(impurities)

Enantioselectivity

best of 2 parents best shufflant S R R S

S3

S4

S5

S6

Expression

Substrate/Product Tolerance
(conversion)
In vitro activity (rel) control WT Rd1 Rd2 Rd3

30 20 10

Activity

Stability - pH, solvent

10 5 Best parent Best shufflant

Start Gen1Gen2 Gen3 Gen4

pH5.5 pH5.5/10 thermal solvent stability

Real Examples (i) Atorvastatin

4 Enzymes developed to provide a new, environmentally friendlier and more efficient / cost-effective process

Hydroxynitrile Starting Material for Atorvastatin

O N H OH N OH OH CO2 Ca0.5 F NC CO 2Et

Existing processes
Forcing conditions for cyanation resulted in base-catalyzed side reactions
carbohydrate malic acid O diketene Cl HO O O
HBr, Ethanol

OH Br CO2 Et
NaCN

OH
asymmetric hydrogenation; bioreduction

, high pH
OH CO2 Et

NC

CO2 Et

+ byproducts

CO2 Et

Cl

NaCN

Product is a high-boiling oil

By-Products in Uncatalyzed Cyanation

CN NC O O NC CN HO O O HO O O O O HO O NC O HO O O NC O O O O O O

OH Cl

O O

KCN NC alkaline 80o C 85.5%

OH

O O

US 5,908,953

High vacuum fractional distillation from close-boiling by-products is required. Cleanup does not address the root of the problem. A catalyst for cyanation at neutral pH was needed to solve this problem

Biocatalytic Cyanation
Halohydrin Dehalogenase (HHDH)
OH Cl O OEt HCl +HCl O O OEt +HCN NC OH O OEt

Replaced natural reagent for the enzyme CN for Cl

pKa~9 HCNaq NaClaq

HClaq pKa<0 NaCNaq

neutral pH stat two phase, no solvent

100 80 % Conversion 60 40 20 0 0 100 200 300 Time (mins)

130 g/L product

Improvements in: Activity / Productivity Relieved product inhibition Stability in process ~4000x improvement in volumetric productivity per biocatalyst loading

400

500

600

substrate:biocatalyst = 100:1 wt:wt

Expression mutant of wild-type enzyme

HN Process Improvements
Shuffling also solves processing / scale-up issues often associated with natural enzymes:

natural enzymes

organic layer (product) aqueous layer

shuffled biocatalysts

7 g/L KRED/GDH; 80 g/L Substrate Reaction time: 24 hrs LAB Phase separation: >1 hour Isolated yield: ~80%

<1 g/L KRED/GDH; 160 g/L Substrate Reaction time: 10 hrs. PLANT Phase separation: ~1 minute Isolated yield: >95%

Science & Technology July 10, 2006 ; Volume 84, Number 28; pp. 24-27

TBIN - Key Intermediate for Atorvastatin

OH NC CO2Et NC OH O CO2tBu NC OH OH CO2tBu NC O O CO 2tBu

OO HN O

O
F

O N O N H O CO2tBu F F N OH

N H

O H 2N

O CO2tBu

OH CO2Ca0.5

TBIN is the first crystalline intermediate in the manufacture of Atorvastatin Diastereomeric purity of the statin side-chain is upgraded at this point

TBIN Original Process

OH NC CO2Et
OLi O

OH O NC CO2tBu

NaBH4 MeOBEt2 cr yogenic NC

OH OH CO2tBu

O NC

O CO 2tBu

HN

chiral diol"

TBIN (ATS-8) first crystalline intermediate

Original Reduction Step:

Stoichiometric borohydride reduction of stoichiometric boronate complex Cryogenic conditions are required for adequate diastereoselectivity. Imperfect diastereoselectivity (3-4% diastereomer at 70C) Wrong diastereomer must be purged downstream; yield losses likely Cumbersome quench & vacuum distillation(s) to remove/recover MeOBEt2 Stoichiometric borate waste stream.
US 6,596,879 to Warner-Lambert

Codexis Ketone Reduction Alternative

OLi

OH NC
HN

OH NC

O CO 2tBu

CO 2Et

KRED (keto reductase) Co-factor [NAD(P)H] Glucose / GDH

OH NC

OH CO 2tBu NC

O CO2 tBu

"Chiral Diol"

TBIN

Desired reduction step: Catalytic transfer hydrogenation under ambient conditions. Perfect diastereoselectivity at the incipient chiral center Hydride source is cheap and renewable High yield of crystalline TBIN Simple extractive work-up Benign waste stream

Identification of Starting Enzyme

Diastereoselectivity
Enzyme Substrate
OH O NC CO2tBu

003 nd

006 >99% trans

008 >99% cis

009 nd

010 nd

011 >99% cis

012 nd

*nd: not detected

Activity CDX008 as starting point required ~200x improvement in activity/stability to enable commercial process
Activity (umol/h/mg)
3

008

011

011-XP

011/NADH

TBIN - Enzyme Evolution Strategy

Libraries of enzyme variants are expressed in E. coli and evaluated in HTP using a tiered approach:
1st tier to identify active and stable variants (fluorescence NADPH / 340nm) 2nd tier to confirm activity (on HK) and selectivity (LCMS differentiate products) 3rd tier to confirm in process improvements (20 ml scale - chemistry)
Note: HK contains 10-20% tert-butyl acetoacetate (BAA) which is a substrate for some ketoreductases, hence 2nd tier screen
OH NC + O CO2 tBu NADPH NADP OH CO2 tBu O CO2 tBu

KRED

OH NC

OH CO2 tBu

General KRED activity in tier 1 detects improvements for both ketone substrates

Activity Improvement of CDX008

%conversion of the best hit variants at RT
50 40 %Conversion 30 20 10 0 008 008-1 008-21 008-3 008-4 008-5

~20-fold improvement over Wild Type by 5 rounds of shuffling

CDX008 Thermostability Improvement

% Remaining activity after O/N, 40oC
100 % Remaining activity 80 60 40 20 0 008 008-1 008-21 008-3 008-41

~85% residual activity after O/N heat treatment (40C)

Active Site - Candidate Positions for Randomization

KRED Evolution for TBIN

80
Nearest structure, 25% homology TBIN KRED homology model

Activity relative to wild-type enzyme

60

predicted binding pocket

40

20

0 WT 1st 2nd 3rd 4th 5th 6th 7th

Best biocatalyst in each generation

7th Round evolvant met productivity targets

Diastereopurity of Codexis C7 diol

LC/MS/MS chromatograms of t-butyl 6-cyano-(3R,5R)-dihydroxyhexanoate samples
A. Oil prepared by Codexis KRED

>99.9 de, or 99.95% R,R

T im e

B. Oil prepared from commercial crystalline tbutyl (6R)-(2-cyanoethyl)-2,2 dimethyl1,3-dioxane-(4R)-acetate

0.10%
0 1 5.00 15 .50 16.0 0

3S,5R
1 6.50 17 .00

3R,5R
17.50 18.0 0 1 8.50 19 .00 19.5 0 2 0.00 T im e

CodexTM Panel Plates as an Investigational Tool

Previously selected recombinant diversity allows identification of starting points for evolution, potentially short-cutting project timelines by months

Case Study CodexTM Panel Result

Mutations that are beneficial for enantioselectivity. Mutations that are beneficial for rate.
20 40 60 80 100

100

Stereomeric Excess (%)

S
50

Conversion (%)
-50

Positions of interest for enantioselectivity.

R
-100

Based on the sequence-activity data from the pre-tuned panel, a fine-tuned panel was designed and screened.

Case Study Fine-tuned Panel Result

CodexTM
100

KRED Panel

hit used for + controls on the fine-tuned panel plate

100

Fine-tuned Custom Panel

S
Stereomeric Excess (%)
50

OH A B
50

>5-fold activity of pre-tuned hit. Perfected stereoselectivity.

20 40 60 80 100

20

40

60

80

100

Conversion (%)
-50 -50

Conversion (%)

OH A B

screened in presence of 5x higher substrate loading.

R
-100

-100

Utility of KRED CodexTM

The CodexTM KRED Panel provided starting enzymes for the development of new, biocatalytic processes for the following intermediates:
O
OH Cl N O OMe

OH Ph

N O

For montelukast: Natural enzyme: no activity Current: commercialized

HO S H N R
OH O
(R) (S)

For ezetimibe: Natural enzyme: no activity Current: tech. transfer

OH
OMe

For duloxetine: Natural enzyme: no activity Current: commercialized

NHCOPh

For penem antibiotics: Natural enzyme: 80% d.e. Current: >99.9% d.e. tech. transfer

Chiral Raw Material:: Natural enzyme: inactivated by substrate Current: tech. transfer

Merit of Codexis KRED Processes

Chemical Processes
Stoichiometric or Catalytic:
DIP-Cl/BH3 (H.C. Brown) Me-CBS and BH3 (E.J. Corey) Ru-BINAP (R. Noyori) Resolution

Codexis KRED Processes

All Catalytic:
-Straightforward manufacture and use -Run at >100 g/L -Economically superior. -Scaleable.

Stereopurity: require upgrades (yield loss) Hazardous and toxic:

- These reagents are typically, toxic, sensitive to water and air, - Expensive.

Produces >99.9% e.e. alcohol. Easy to use

- Ambient conditions, run in aqueous solvents, - Degradable in waste water treatment plant.

Generally, tedious work-up procedures required.

Straightforward work-up.

Ene Reductase CodexTM

The ERED Codex is useful for:
Chiral reduction of ,-carbonyls:
-Esters, -Nitriles -Ketones/aldehydes, -Nitro compounds. Co-factor recycling with either IPA/KRED or glucose/GDH, Variants are highly stable to organic solvents (e.g. IPA), and elevated temperatures.
O ERED NADPH NADP+ O or O

Transaminase CodexTM Panel

O Transaminase NH2

NH2

Chemical Process
Diastereomeric salt resolution, Reductive amination, Reduction of eneamines, Simulated Moving Bed.

Codex TA Panel
Versatile catalyst for a broad range of substrates. Standard methods of use. High volumetric through-put. Solutions:
- Economically attractive processes, - High yielding, - Standard catalyst manufacturing

Issues:
- Low yielding, - Expensive catalysts, - Need for dedicated equipment.

Halohydrin Dehalogenase CodexTM

O R1 R2 SCNR1 R2 OH X R1 R2 HCO2R1 R2 OH OH NO2R1 R2 OH NO2 S R1 OCNO O R2 R1 R2 R1 R2 OH N3 OH CN NH

N3CN
-

Chemical Process
Sharpless: Epoxidation of allylic alcohols Jacobsen HKR: no , -disubstitution Issues:
Hetero atom substitution problematic Regioselectivity issues Expensive TMS-CN, TMS-N3 Metals used: Osmium, Cobalt, Chromium

Codex HHDH Panel

Broad range nucleophiles , -disubstitution HHDH Opportunity:
Hetero atom substitution allowed Regioselectivity tuneable Cheap nucleophiles

Precedent: HN for atorvastatin

Real Examples (ii) Ezetimibe

Enzyme & process developed to avoid use of Me-CBS

Ezetimibe Process - Differentiation

O O Ph N O F O BH3 THF (R)-Me-CBS THF; 25C Codex KRED Glucose/GDH F OH Ph O N O O

Chemical Process
Uses Me-CBS and BH3
- Both are toxic and hazardous. - Me-CBS is expensive and used at high loading Stereoselectivity is inadequate and highly touchy (over-reduction). Sensitive to moisture and reagent quality Stereopurity upgrade required

Codexis KRED Process

Catalytic: Runs at 100 g/L:
- Environmentally friendly - Greatly reduced hazard (toluene at RT) - Economically superior Produces >99.9% e.e. alcohol Eliminates over reduction Isolated yield >95%

Complicated Work-up

Straightforward work-up

Real Examples (iii) Montelukast

Enzyme & process developed to avoid use of DIP-Cl

Montelukast Process - Differentiation

O Cl N O OMe Merck: DIP-Cl CDX: IPA/KRED OH Cl N O OMe

Neither SM nor product are soluble - run as a slurry-to-slurry process No need for co-factor re-cycle or distillation
Chemical Process
Uses more than stoichiometric DIP-Cl/BH3
Corrosive, acidic solid; causes burns; moisture-sensitive Expensive

Codexis KRED Process

Catalytic - Runs at 100 g/L with S/C >30:
Environmentally friendly Greatly reduced hazard (IPA at RT) Economically superior

Produced 97% e.e. alcohol. Isolated 87.1% yield, 99.5% e.e. Tedious work-up to remove borate salts

Produces >99.9% e.e. alcohol. Isolated yield >90%. Straightforward work-up via filtration

Summary
Biocatalytic options are real alternatives for process chemistry, Modern molecular biology tools combined with HTP screening allows for catalyst optimization on an as needed basis, Enzymes can be optimized to function in desired process conditions and to meet desired economic targets, New process and catalyst composition-of-matter intellectual property is developed. Combination of Molecular Biology with.. Process Chemistry Analytical Science Combinatorial Technology ... provides a powerful tool for engineering of green processes