www.almacgroup.com
“Rational Design of Enzymes tofit the process chemist
timelines”
Dr. Derek Quinn / ALMAC Group
Organic Process Research &DevelopmentPasadena, 6-8th March 2017
Founded in 1968
Global Headquarters in Craigavon, Northern Ireland
Turnover >$500m
~ 4000 Personnel Globally
Unique ownership – charitable foundation
All Profit Re-invested
Arran Chemical Company - purchased 2015
ALMAC Group Overview
Advantages from using enzymes
Enzyme Hit Process Development options with economic assessment
Phaseappropriate
fit-for-purposeprocess
OrCommercialEconomic
Process withconstant pricing
assessmentEnzymeKits
Bioinformaticselection & insilico design
Enzymeengineering
1. ProcessOptimisation
2. SubstrateEngineering
3. Ultrasound& reactionengineering
4. Resinapplication forinhibition and
extraction
5. EnzymeImmobilisation
6. Enzymeengineering
Constant economic assessment
Enzyme process design
• Selectivity• Specificity• Stability• Activity
• Cost• Productivity• Quality• Sustainability
Product isolation, purification, removal of bioburden/catalyst
IRED no.
% HPLC peak area @ 260
nm
% HPLC peak area of
product enantiomersee
Product SM (S) (R)
3 9 91 54 46 8
5 2 98 0 100 100
8 77 23 89 11 78
9 2 98 13 87 74
12 3 97 100 0 100
16 3 97 77 23 54
20 100 0 100 0 100
26 1 99 89 11 78
27 31 69 92 8 84
28 8 92 0 100 100
SM
Product
• In silico selection of 50 IRED enzyme library• Gene synthesised, cloned and expressed• Screened against client substrate – 20% hit rate
IRED found by in-silico screening
• IRED-20 selected for development(100% ee and best activity in screening)
• Enzyme form defined: Spray-dried cells suitable
• Challenges encountered during PRD:• Product inhibition limits product titre 15g/L• Product highly water soluble difficult recovery
by extraction• Evolution???
• Solution implemented: Ion exchange column inrecirculation mode (54g/L) Solves inhibition problem in aqueous
reaction medium Enables efficient product recovery with
ammonia in methanol
IRED process and challenges
Enzyme evolution overview
• Classical mutagenesis techniquesslow, undirected, high screening burden
• Rational mutagenesis:Crystal structures/Modelling, computationally ,hungry.
• Random mutagenesis:Mutant generation simple, but very high screening burden
• Directed mutagenesis/combinatorial approach:Directed evolution, high screening burden reduced by recombination,very powerful, mimics nature
• Machine learning:Mathematical algorithms which predict mutations and learns fromboth positives and negatives
• Rational mutagenesis revisitied, computational power has increasedexponentially, new powerful algorithms can now exploit this.
• Awarding of Nobel prize in 2013 to Karplus, Levitt and Warshel forQM/MM
Rational Design / Modelling Methods
Classical mechanics Hybrid methods Quantum mechanics
• Newtonian mechanics andempirical
• Atom is the smallest unit• Force fields• Computationally fast• Less accurate
• QM/MM methods• EVB method
• Electronic structure• Based on approx. to the
time indep Schrödingereq
• HF, pos-HF, DFT andsemiempiricals
• Computationallyexpensive
• Size limitations
Molecular dynamics
E = Kr
r -req( )
2+
bonds
å Kq
q-qeq( )
2+
angl es
åV
n
21 +cos nj -g[ ]( ) +
Aij
Rij
12-
Bij
Rij
6
æ
èçç
ö
ø÷÷
i< j
atoms
åtors ions
å +qi qj
e Rij
æ
èçç
ö
ø÷÷
i< j
atoms
å [1]
Docking methods
Rational Design/Protein engineering
InitialStructuralanalysis
In silicopredictionof mutantEnzymes
Wet labvalidation
Docking/Mechanism
determination
Pointmutations
Typically 30-50rationally
designed mutantenzymesper round
Wet labvalidation
Rational Design
Hydrolase CarbonylReductase
AmidaseNitrileHydratase
Transaminase
How to make chiral amines
Chiral amine target
Transaminase+
ketone
Aminoacid-DH/Amine-DH
+a-ketoacid/ketone
Ketoreductase+
ketone
Amine oxidase+
racemate
Imine reductase+
imine
Hydrolase/Nitrilase+
ester/nitrile
Classical resolution,Auxilliary chemistry
Asymm. hydrogenation,Asymm. PTC
WT (S)-TAM Vibrio fluvialis
Bulky substrates – typically not wellaccepted by wildtype TAm enzymes
Rational engineering approach taken toopen up pocket and improve interactions
Substrate docked into active site andenergy minimisation carried out todetermine most likely pose
Non-catalytic poseDistance between the amino group of PMPand the substrate carbonyl is large (5.1 Å).
•Nucleophilic attack of PMP not possibledue to steric clash between the phenylgroup of substrate and PMP.
ACS Catalysis, DOI: 10.1021/acscatal.6b02380
TAm
Structural analysis
(S)-TAM Vibrio fluvialis
Am
ino
acid
sdis
trib
uti
on
Co-evolution network
Homologous Sequences analysis (~30000 seq.)
ACS Catalysis, DOI: 10.1021/acscatal.6b02380
Best variant of (S)-TAMVibrio fluvialis
Catalytic pose
•Distance between the substrate carbonylcarbon and the PMP amino group (3.1 Å)quite close to the distance reported forthe PMP: acetophenone intermediate(2.65 Å) (Cassimjee et al. 2015).
•Strong T-shaped π-stacking and S/π interactions introduced.
•V153S mutation may increase the subunitinterface affinity. Calculated
ΔΔGV153S =-2.01 kcal/mol.
•After 18h the best variant retains 98 %and 90% of activity at 40°C and 50°C,respectively, while the other variantssuffer a significant decrease in activity at50°C
ACS Catalysis, DOI: 10.1021/acscatal.6b02380
General Mechanism of Transaminase Enzymes
Complex mechanism: Planar quinonoid orientation and catalytic lysineresidue defines chirality of product
Best variant: Analysis of the quinonoid intermediate
• The spatial arrangement of the catalytic lysine with respect to the planar quinonoid commands theorientation of the amino group of the catalytic lysine, which undergoes nucleophilic attack towards thenitrogen binding carbon of the quinonoid intermediate.
• The spatial arrangement favours the formation of pro-(S) external aldimine and consequently theformation of the end S-amine product
ACS Catalysis, DOI: 10.1021/acscatal.6b02380
TAM-WT
ConversionSubstrate: phenyl-acetophenone
Reaction conditions: 5 mM substrate, 1M IPA, 0.01 mM enzyme, 0.1 M K·phosphate buffer (pH 8.0), 10% DMSO (v/v), 40°C, 18 h. Allreactions were performed in triplicates.
TAM- best variantConversion: ND Conversion: 42.1%
ACS Catalysis, DOI: 10.1021/acscatal.6b02380
• 113 rational designed variants, 2 in vitro screening rounds
• Best variant had > 1700 fold-improvement in the reaction rate
a Reaction conditions: 5 mM substrate, 1M IPA, 0.01 mM of purified enzyme, 0.1 M potassium phosphate buffer (pH 8.0), 10 %DMSO (v/v), 40 °C, 18 h. All reactions were performed in triplicates.b Kinetic parameters are the apparent rate constants calculated from the initial reaction rates at a fixed concentration ofamine donor (1 M).c Fold change with W57G/R415A as referenced Reaction rates are initial reaction rates. Reaction conditions: 5 mM substrate, 1 M IPA, 0.01 mM of purified enzyme, 0.1 MK·phosphate buffer (pH 8.0), 10 % DMSO (v/v), 40 °C. Reactions were performed in duplicates.e ND - not detectable. Reaction rate for WT was no more than 0.1 μM/h, considering the detection limit.
ACS Catalysis, DOI: 10.1021/acscatal.6b02380
Rational Design
Hydrolase CarbonylReductase
AmidaseNitrileHydratase
Transaminase
B/C:ter
300K341K
Molecular dynamics :B-factor calculated at two different temperatures
Amidase – Rational Design - Thermostability
Highly-flexible regions identified
Ni2
Ni1
Identify flexible regions: High B-factorMutate residues and test for reduction in B-Factor
SS-bond insertion criteria:
1 - Distance CB-CB 3Å-4.5Å2 – Distance C-SG 1.8 Å ; SG-SG ~2 Å3 – C-SG-SG-C dihedral angle ~90°4 - C-SG-SG angle ~103°
Amidase – Rational Design
Multimeric enzyme: 3 separately expressed proteinsStabilisation between protein interfacesStabilisation within subunits of proteinsS-S bonds, salt bridges, stabilising mutations
Amidase – Rational Design
SelectedMutants
Subst
rate
conc
ent
rati
on
(mM
)
• ~100 mutants generated in study for retention of activity at high ºC• Wild type and mutants were incubated at 70 ºC for 16h• Retention of activity measured (substrate depletion) at room temperature• Activity of mutants less than wild type, not surprising (less flexible enzyme)• Retention of mutant enzyme 4 greatly improved• >50 % of activity after 3 days, wild type activity absent
Rational Design
Hydrolase CarbonylReductase
AmidaseNitrileHydratase
Transaminase
Rational Design
Hydrolase CarbonylReductase
AmidaseNitrileHydratase
Transaminase
• Substrate inhibition: nM affinity• Molecular modelling, substrate docking, rational
mutant selection, gene synthesis, wet screening• Predicted affinity reduced to from nM to mM range
with 20 mutant enzymes
Wild type: 88.9 nM Variant: 12.7 mM
Addressing inhibition via CRED engineering
CRED-A231_F97A_L241A_W226A +NADPH docked
Carbonyl reductase - Rational evolution
Substrate
NADPH
• In silico analysis of WT CRED (Molecular Docking).
• 20 variants rationally designed.
• Best mutant increases conversion to >90% with aee of 98.7 % (S).
Result: Increased conversion of ketone 1 whilemaintaining the WT enzyme selectivity
1 2
Rational Design
Hydrolase CarbonylReductase
AmidaseNitrileHydratase
Transaminase
Background:• Client wanted enzymatic process to replace chiral Simulated Moving Bed
Chromotography process for late phase and commercial API
Problems:• Expensive racemate – so yield is important• Two enzyme process Hydrolase 1 = resolution Hydrolase 2 = Desymmetrisation• Hydrolase-1 had low E-value of 13 (R)-diester yield < 30%• Hydrolase-2 had low stability high enzyme loading required with resulting
isolation problems impact on yield
Solutions:• Improve E-value of resolution step via substrate engineering• Improve stability of desymmetrization enzyme via homologues panel
“Genomes, Biocatalysts and Pharmaceutical Mater ials”
Humboldt University, 13 November 2015, [email protected]
Desymmetrization for 2nd generation process
Problems solved:• Hydrolase-1 has E-value of 60 with dipropyl ester (R)-diester now at 45+%.• Hydrolase-2 accepted (R)-propyl diester with unchanged selectivity.• Hydrolase-2* found via homologues search, from a thermophilic organism. At
maintained selectivity the process now runs at 50 deg C with low enzymeloading.
Thinking forward:• Solutions so far kept registration schedule on track but could be improved.
• Protein engineering of hydrolase-2* to open the way to3rd generation single enzyme process.
Desymmetrization for 2nd generation process
• single enzyme approach preferred• Substrate docking and rational
mutagenesis: 2 rounds completed• Greatly increased selectivity, catalytic
rate also increased• Mutations combined to hit target
E value of >100
W.T.
W.T.
Hydrolase engineering for 3rd generation process
E-value
E-value
Rational Design/Protein engineering
InitialStructuralanalysis
In silicopredictionof mutantEnzymes
Wet labvalidation
Docking/Mechanism
determination
Pointmutations
Typically 30-50rationally
designed mutantenzymesper round
Wet labvalidation
InitialStructuralanalysis
molecular docking /Molecular Dynamics
simulations
Co-evolutionanalysis
Rational design ofmutants
Bibliographic search /Homology Modeling /Structural analysis
Molecular docking /Molecular Dynamics
Final list ofmutants for in
vitro screening
1 week 1-2 weeks
2-3 weeks computational, 3-5 weeks mutant generation
Screening
2 weeks
Rational Design: Typical Timelines
• Rational design in a matter of weeks canproduce major improvements to solve arange of issues
• It can deliver in its own right and/or be aplatform to build upon
34
• Lignocellulose, cell algaeweakening
• Cell permeation• Biocatalysis
enhancement• Surface attrition• Surface activation• Improved heat/mass
transfer• Emulsification
• Lignocellulose, cell,algae weakening
• (Bio)polymerdegradation
• Crystallization• Cell permeation• Emulsification
IN THE CAVITYextremeconditions oncollapse 5000oCand 2000atmospheres
IN THE BULKMEDIAintense shearforces
UNSYMMETRICCOLLAPSEInrush of liquid fromone sideof the collapsingbubbleproduces powerfuljet of liquidtargeted at surface
Acknowledgement Prof. T Mason, Uni. of Coventry
CelbiusUltrasound
• Lignocelluloseweakening
• Cell permeation• Biocatalysis
enhancement• Surface attrition• Surface activation• Improved heat/mass
transfer
Video courtesyof University of Twente, Netherlands.and Shimadzu Europa GmbH,Duisburg,Germany
• Lignocellulose weakening• (Bio)polymer degradation• Crystallization• Cell permeation
IN THE CAVITYextremeconditions oncollapse 5000oCand 2000atmospheres
IN THE BULKMEDIAintense shearforces
UNSYMMETRICCOLLAPSEInrush of liquid fromone sideof the collapsingbubbleproduces powerfuljet of liquidtargeted at surface
Acknowledgement Prof. T Mason, Uni. of Coventry
CelbiusUltrasound
Process Operation
Reactor 50 -5000 L
Recirculation10 – 100L/min
50W - ~ 5 kWDepending onmain reactorvol. & powerreqd.
Temperature 0 to 70oC
Process intensification
• Process being developed formulti-tonne,
• Step 2 of a 5 step process• Process when scale-up had
mixing issues• Poor performance for reaction
time >70 hr• Options? Enzyme evolution? or
Application of US? or both?
0
20
40
60
80
100
120
0 10 20 30 40 50 60 70 80
Convers
ion
(%)
Time (hrs)
Ultrasound assisted biotransformation
Series1
Series2
Process intensification
Current technologyConfigured in reactor / recirculation systems
Flanged cylindricalpipe
Transducers bondedto deliver ultrasound
Medium out
Medium in
Pictures courtesy of Celbius
40
Development of methods including:- ID (FTIR, NMR, MS, XRPD)- Assay- Related substances- Chiral purity- Microbial and biomass, residual protein- Residual solvents, proteins, biomass- Particle size- Genotoxic impurities- Additional specification tests as required, developed inparallel and in conjunction with chemical development
AU
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
1.10
1.20
Min ute s
0.0 0 0.2 0 0 .4 0 0 .60 0.80 1.0 0 1.2 0 1 .4 0 1 .60 1.80 2.0 0 2.2 0 2 .4 0 2 .60 2 .80 3.0 0 3.2 0 3 .4 0 3 .60 3 .80 4.00 4.2 0 4 .4 0 4 .60 4 .80 5.00
Analytical Development
4/24/2017
Assay Range Comment
NanoOrange® 10 ng/mL to 10 μg/mLLow protein-to-protein signal variability
Detection not influenced by reducing agents or nucleic acids
BCA 0.5 μg/mL to 1.5 mg/mLSamples must be read within 10 minutes
Not compatible with reducing agents
Bradford 1 μg/mL to 1.5 mg/mL
Proteins precipitate over time
High protein-to-protein signal variability
Not compatible with detergents
Lowry 1 μg/mL to 1.5 mg/mLLengthy, multistep procedure
Not compatible with detergents, carbohydrates or reducing agents
Absorbance at 280 nm 50 μg/mL to 2 mg/mLHigh protein-to-protein signal variability
Detection influenced by nucleic acids and other residues
4/24/2017
Example 3:
API 3
Example 4:
API 4
Example 5:
API5
Example 6:
API 6
Enzymatic stepPre RSM, C-C bond
formation
Pre RSM, ketone
reduction
Post RSM,Alcohol
oxidation
Pre RSM Ketone
Reduction
Stage in synthesisSeveral steps before API
Several steps beforeAPI
Penultimate step Several steps before API
Enzyme type
Enzyme (Cells from E.
coli fermentation)
KRED (liquid
formulation from in E.
coli fermentation)
KRED (whole cell
formulation produced in
Gluconobacter oxydans)
KRED (Dry powder
produced from E. coli
fermentation)
Control strategyTest for total proteins,
DNA, endotoxins and
microbiological residues;
demonstrate fate and
purge; no enzyme
residue specifications for
API
Test for total
proteins; no enzyme
residue specifications
for API
Test for total proteins,
DNA, endotoxins and
microbiological
residues; demonstrate
fate and purge; no
enzyme residue
specifications for API
Test for total
proteins;
in-process testing;
demonstrate fate and
purge;
no enzyme residue
specifications for API
43
Where we are…..
The oldest Biocatalysis Shop in Ireland!
Dr. Derek Quinn
Biocatalysis: Biology Team Leader Leader
Email: [email protected]
ALMACDepartment of Biocatalysis & Isotope Chemistry
Seagoe Industrial EstateCraigavon, N. IrelandBT63 5QD UK
www.almacgroup.com