Enzyme Optimization and Process Development for a Scalable Synthesis of (R)-2-Methoxymandelic Acid

Article abstract of DOI:10.1021/acs.oprd.1c00250

The rational protein engineering of wild type BCJ2315 nitrilase and its use in the development of a one-pot, enantioselective, dynamic kinetic resolution for (R)-2-methoxymandelic acid is reported. Through a combination of molecular docking and B-factor a

Full text of DOI:10.1021/acs.oprd.1c00250

pubs.acs.org/OPRD
Article
Enzyme Optimization and Process Development for a Scalable
Synthesis of (R)‑2-Methoxymandelic Acid
Mark E. Scott, Xiaotian Wang, Luke D. Humphreys,* Michael J. Geier, Balamurali Kannan, Johann Chan,
Gareth Brown, Daniel F. A. R. Dourado, Darren Gray, Stefan Mix, and Aliaksei Pukin
Cite This: https://doi.org/10.1021/acs.oprd.1c00250
Read Online
sı
*
Metrics & More
Article Recommendations
Supporting Information
ACCESS
ABSTRACT: The rational protein engineering of wild type BCJ2315 nitrilase and its use in the development of a one-pot,
enantioselective, dynamic kinetic resolution for (R)-2-methoxymandelic acid is reported. Through a combination of molecular
docking and B-factor analyses, a focused library of mutants was identified and screened to improve nitrilase selectivity, activity, and
stability. Initial optimization revealed that the addition of sodium bisulfite prevented enzyme deactivation by the aldehyde starting
material, removing the need to isolate the cyanohydrin and allowing for development of a one-pot process for mutant screening.
Further optimization of the process with the preferred mutants revealed subtle interactions between temperature, pH, substrate
loading, and enzyme source which ultimately led to development of a suitable lyophilized whole cell process for scale-up, affording
(R)-2-methoxymandelic acid in 97% ee and 70% isolated yield on multigram scale.
KEYWORDS: nitrilase, mandelic acid, enantioselective, biocatalysis, dynamic kinetic resolution, protein engineering
derivatives,7 we anticipated several challenges with this
approach: avoiding enzyme deactivation from cyanide or
INTRODUCTION
■
In the past two decades the application of biocatalysis for the
aldehyde while achieving high enantioselectivity and identify-
preparation of pharmaceuticals, their intermediates, and
ing a process friendly isolation, including a product
commodity chemicals has undergone significant growth and
crystallization which avoided a concentration to dryness. To
development.1 Biocatalysts have been more widely adopted by
this end, this manuscript highlights our approach to addressing
industry due to their selectivity and ability to catalyze a broad
these challenges via the concurrent development of a one-pot
range of chemical transformations under mild reaction
process and enzyme engineering to address the anticipated
conditions in green solvents (e.g., water). In addition,
enantioselectivity and process robustness challenges, followed
biocatalysts can be readily engineered to optimize properties
by our subsequent downstream process optimization isolation
such as stability, activity, and selectivity which allows for the
development.
design of robust and scalable processes.
Biocatalysts have been broadly categorized using enzyme
commission (EC) numbers.2 The nitrilase superfamily are
hydrolase enzymes (EC3) that catalyze the hydrolysis of
nitriles, acid amides, secondary amides, ureas, and carbamates
via their catalytic cysteine, glutamate, and lysine triad.3,4 While
each of these nitrilase-mediated synthetic transformations are
useful, arguably the most appealing from a process chemistry
perspective is nitrile hydrolysis given the ease of introducing
nitrile functionalities within synthetic building blocks as a
masked carboxylic acid surrogate.
RESULTS AND DISCUSSION
■
Our initial approach was two-pronged, whereby we con-
currently developed suitable conditions for initial nitrilase
screening, coupled with a rational in silico design of a 96
mutant library that focused on improved enantioselectivity and
enzyme stability over the wild-type. Further process refinement
was planned to enable robust evaluation of more promising
mutants from the first round of enzyme engineering on scale.
In turn, this would facilitate a second round of mutants via
combination of the best performing mutants from round one.
Initial process development was performed using a
previously reported novel nitrilase BCJ2315 isolated from
Burkholderia cenocepacia J2315.6 This nitrilase was selected
Recently, as part of our process chemistry efforts for a
clinical compound, we sought to develop an enantioselective
route to the process intermediate (R)-2-methoxymandelic acid.
Based upon our initial survey of the literature we were
immediately intrigued at the possibility of designing an
enantioselective one-pot enzymatic nitrilase approach toward
(R)-2-methoxymandelic acid, based on prior successful reports
for the parent mandelic acid compound.5,6 Our initial approach
was to leverage a dynamic kinetic resolution (DKR) whereby
racemic cyanohydrin 2 is selectively hydrolyzed to the requisite
(R)-2-methoxymandelic acid 3 (Scheme 1). Based on related
literature precedent for ortho substituted mandelic acid
Special Issue: Excellence in Industrial Organic Syn-
thesis 2021
Received: June 23, 2021
https://doi.org/10.1021/acs.oprd.1c00250
Org. Process Res. Dev. XXXX, XXX, XXX−XXX
© XXXX American Chemical Society
A

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Scheme 1. Proposed One-Pot DKR Route to (R)-2-Methoxymandelic Acid (3)
a
Table 1. Effect of Organic Cosolvents on BCJ2315 Lyophilized Whole Cell Activity
b
b
Entry Solvent (10% v/v) Substrate concentration (g/L 2) Enzyme Loading (%, w/w) Conversion at 3.5 h (%)
Conversion at 16 h (%)
% ee
1
2
3
4
5
6
7
Methanol
Ethanol
Isopropanol
DMSO
MTBE
Toluene
10
10
10
10
10
10
10
10
10
10
10
10
10
10
0
0
0
48
5
3
0
0
0
67
62
27
85
−
−
−
92
75
85
90
Ethyl acetate
60
a
To 0.1 mL of a solution of 2 in organic solvent was added BCJ2315 (freeze-dried cells) in 0.9 mL of phosphate buffer (100 mM, pH 7.4) followed
b
by shaking at 30 °C. Determined using HPLC at 283 nm. Conversion was determined by [% AN 3 + % AN ent-3]/[% AN 1 + % AN 2 + % AN 3
+ % AN ent-3].
based on prior engineering success to enable the synthesis of
(R)-2-chloromandelic acid along with a variety of related
substituted mandelic acid analogues.7 We began initial
screening studies using isolated cyanohydrin 2 and BCJ2315
wild-type nitrilase from soluble expression in E. coli and
isolated as a lyophilized cell-free extract (CFE), since this
preparation was favored from a process perspective due to its
purity and anticipated minimal downstream isolation/workup
complications. However, trial reactions using 10% by weight
CFE at pH 7.4 with 25% methanol gave only ca. 1% AN of 3
after 63.5 h. Screening was then performed with the CFE to
further evaluate cosolvent impact on reaction conversion using
10, 20, 30% (v/v) of methanol or ethanol, or 10% (v/v) of
toluene via overnight shaking with 10% by weight enzyme
loading relative to 2. None of the reaction mixtures with
alcohol cosolvents contained the product after 16 h of shaking
at 30 or 37 °C. On the contrary, reactions that were conducted
under biphasic conditions with toluene as a cosolvent afforded
minor formation of 3 in all conditions evaluated. From these
initial studies the best conditions led to 16% product formation
by HPLC using 10% (v/v) in toluene (10 g/L BCJ2315 CFE,
100 g/L of 2, 100 mM potassium phosphate (pH 7.4), shaken
for 16 h at 30 °C).
Based on our initial limited success using lyophilized
BCJ2315 CFE enzyme, we re-evaluated the enzyme prepara-
tion used for our studies. Previous reports for the production
of BCJ2315 nitrilase from E. coli employed dithiothreitol
(DTT) reducing agent during purification and eventual storage
of the isolated enzyme6 despite the fact that the final large scale
biotransformations employed wet cells.7 The use of reducing
agents such as DTT is proposed to protect the catalytic
cysteine residue from oxidative disulfide bond formation which
would render the enzyme inactive. Indeed the requirement for
reducing reagents has also been observed with other purified
nitrilases, although high concentrations of DTT have been
found to be detrimental to certain nitrilases.8,9 DTT should
also prevent the formation of additional intra- or intermo-
lecular disulfide bonds which could inactivate the enzyme.
Nitrilases have been shown to have structures made up from
homodimers, and this quaternary structure, which could be
destabilized through disulfide bond formation, is also
responsible for higher observed activity.10 In wet whole cells
however, the nitrilase enzyme is protected by the intracellular
cytoplasmic reducing environment. This preparation has
limitations in terms of handling, stability, and storage,11 and
therefore a dried enzyme formulation is often preferable.
In our initial screening studies, the lyophilized cell-free
extract without additional additives showed some activity but
was significantly lower compared to that achieved with the
same equivalent charge of whole cell biomass, presumably due
to oxidative degradation or deactivation due to thermal
exposure during sonication. Chemical methods of cell lysis
which do not generate heat such as BugBuster were not
investigated due to the cost of their use on scale being
prohibitive. In an effort to reduce any potential effect of
oxidative degradation, a pretreatment of the CFE with DTT
was evaluated. While our initial studies showed improved
reactivity, achieving 50% conversion of cyanohydrin 2 in 3.5 h
(10 g/L of 2, phosphate buffer pH 7.4, 30 °C with 10% v/v
DMSO), the effect was found to be variable in subsequent
replicate studies. Due to this lack of robustness, we decided to
focus instead on both wet and lyophilized whole cell
preparations, the latter of which was more appealing due to
its ease of handling, storage, and reproducibility. In addition,
we investigated water immiscible solvents which would make a
subsequent workup more straightforward and we switched to a
higher pH buffer to minimize the potential to form HCN
under the reaction conditions. After 3.5 h of reaction (67 g/L
of 2, Tris-HCl buffer pH 8.5, 30 °C with 10% v/v ethyl
acetate), the product content as determined by HPLC area in
the reaction mixture was 28% and 42% for wet and freeze-dried
cells respectively, using 5% enzyme loading. In these studies,
wet cell loadings were adjusted to ensure equal enzyme charge
based on their dry weight relative to 2. Based upon these
results, the use of lyophilized whole cells was selected for
further process optimization experiments via parallel assess-
ment of single reaction parameters and process productivity.
Initial single parameter optimization efforts (reaction
medium, temperature, pH, and solvent tolerance) using
lyophilized whole cells were conducted at low substrate
concentrations (5−10 g/L of 2) so as to avoid any interfering
inhibition effects and to omit the potential need for pH
adjustment during the course of the reaction due to the
formation of the carboxylic acid product. In addition, low
B
https://doi.org/10.1021/acs.oprd.1c00250
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
enzyme loadings were used to prevent the reaction from
achieving 100% conversion, thereby enabling the impact of
each process parameter to be assessed.
optical purity of the product. Raising the temperature from 30
to 40 °C did not have any detectable impact on product ee but
did lead to increased enzyme deactivation in ethyl acetate as a
cosolvent, and in turn incomplete reaction. This effect was
observed across all ethyl acetate containing reactions, with
poor conversion being obtained, even after 16 h.
Reassessment of organic solvents on the activity of BCJ2315
nitrilase from lyophilized whole cells was performed using both
water miscible and immiscible solvents (Table 1). Cosolvent
volumes were kept at 10% v/v in order to maintain the
solubility of the substrate within the reaction mixture. As
observed in earlier optimization efforts for CFE BCJ2315,
alcohols (entries 1−3) were once again found to be deleterious
to the reaction, shutting down all activity while nonprotic
solvents (entries 4−7) gave much better results. Although a
good result was observed with DMSO, a water immiscible
cosolvent was preferred for ease of workup and EtOAc
provided the best combined outcomes for conversion and
optical purity, with toluene not far behind.
Until now, process optimization had been developed using
the preformed 2. However, a more cost-effective approach
would be a one-pot process using readily available 2-
methoxybenzaldehyde (1) as starting material. Given the
promising results obtained thus far, it was decided to explore
this one-pot reaction under biphasic conditions starting with
the aldehyde and NaCN. Previous reports of similar one-pot
processes had been limited due to enzyme deactivation by the
benzaldehyde starting material which had led to the develop-
ment of more complex fed-batch processes.7b,12 Enzyme
deactivation using BCJ2315 lyophilized whole cells had been
observed early on in our reaction optimization efforts. While
typical reaction conditions are expected to favor cyanohydrin
formation over the aldehyde, this reaction is in equilibrium and
thus some amount of enzyme deactivating aldehyde would
always be present. This issue was expected to impact process
robustness and cost due to the subsequent need for higher
enzyme loadings to overcome enzyme deactivation and achieve
full conversion. In addition, higher enzyme loadings would also
complicate downstream product extraction and isolation
operations. To avoid these potential issues, we hypothesized
that sodium bisulfite in the reaction would sequester any free
aldehyde as the corresponding bisulfite adduct, thereby
reducing its ability to deactivate the enzyme. In this proposed
process, the aldehyde would first react with bisulfite to form 4.
Any remaining aldehyde may alternatively undergo cyanohy-
drin formation directly with HCN (Scheme 2) thereby
working with the sodium bisulfite to lower the effective
concentration of the inhibiting benzaldehyde.
Initial proof of principle studies for this one-pot process
were successfully carried out on gram scale using 5% by weight
lyophilized whole cell BCJ2315 in 9:1 water/ethyl acetate
(0.49 M final substrate concentration). The reaction was
performed using 1.1 equiv of NaHSO3 and 1 equiv of NaCN,
affording 3 in 93% ee and in 90% isolated yield after 48 h at 22
°C. Repeat studies were subsequently performed to evaluate
the impact of temperature for this one-pot process under more
challenging reaction conditions employing a lower 2.5 wt %
lyophilized whole cell BCJ2315 loading. Reaction mixtures
were prepared by mixing 1 volume of ethyl acetate with 7.5
volumes of an aqueous aldehyde−bisulfite adduct solution
(prepared via dissolution of 1 in 0.979 M aqueous NaHSO3).
A series of reactions were next performed to assess the
influence of pH and temperature on the reaction outcome
using 10% by weight lyophilized whole cell BCJ2315 nitrilase
for the preferred cosolvents ethyl acetate and toluene (Table
2). The results reveal that higher pH had a positive effect on
Table 2. Influence of pH, Temperature, and Reaction
Cosolvent on BCJ2315 Lyophilized Whole Cell Activity
a
b
Solvent (10%
v/v)
Temp
(°C)
Conversion after 8 h
(%)
Entry
pH
% ee
c
1
2
3
4
5
6
7
8
Ethyl acetate
Ethyl acetate
Toluene
30
40
30
40
30
40
30
40
30
40
30
40
8.5
8.5
8.5
97
83
97
97
97
90
90
86
86
93
93
89
89
94
94
90
90
c
c
c
Toluene
8.5
d
d
d
d
d
d
d
d
Ethyl acetate
Ethyl acetate
Toluene
10.1
10.1
10.1
10.1
10.7
10.7
10.7
10.7
e
77
96
95
97
Toluene
9
Ethyl acetate
Ethyl acetate
Toluene
e
10
11
12
55
95
95
Toluene
a
2 was solubilized in either 0.5 mL of ethyl acetate or toluene and
added to BCJ2315 (20 mg of freeze-dried cells) in 4.5 mL of an
appropriate pH buffer to a final concentration of 40 g/L of 2.
Reaction mixtures were shaken at the indicated temperature for 8 h.
b
Determined by HPLC at 283 nm. Conversion was determined by [%
AN 3 + % AN ent-3]/[% AN 1 + % AN 2 + % AN 3 + % AN ent-3].
c
d
100 mM Tris-HCl buffer. 100 mM Bicarbonate-carbonate buffer.
e
Conversion reported here is % AN 3 + % AN ent-3 due to formation
of significant impurity formation.
Scheme 2. Proposed One-Pot DKR Route to (R)-2-Methoxymandelic Acid (3) via Bisulfite Adduct Formation
C
https://doi.org/10.1021/acs.oprd.1c00250
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
To this was added 1.5 volumes of 4.9 M aqueous NaCN
solution dropwise (total reaction volume concentration = 100
g/L of 1). These studies evaluated the reaction rate at 21, 30,
and 37 °C by HPLC analysis at 1, 2.5, 4.5, 7, and 24 h.13 This
study revealed that while both reactions at 21 °C and at 30 °C
showed complete consumption of the starting materials over
24 h, the reaction at 37 °C did not reach completion,
presumably due to enzyme deactivation from both the high
temperature and substrate concentration. Interestingly, lower
reaction concentrations of 68 g/L of 1 (0.5 M) at this higher
37 °C temperature were tolerated and allowed the reaction to
be complete, further highlighting this subtle interplay between
substrate concentration and temperature for this process.
Based on this observation, this effect was further investigated
using more challenging lower whole cell enzyme loadings (2%
by weight) to identify the preferred reaction conditions that
enabled robust performance while maximizing throughput
(Table 3). These results showed incomplete reactions for the
enzyme.16 This approach was preferred since it often can
explain the in vitro results while enabling exploration of the
enzyme’s multidimensional fitness landscapes, thereby allowing
evolutionary studies to focus on smaller data sets of mutants at
a fraction of the cost versus typical directed evolution
approaches.
We began our rational engineering of BCJ2315 by
constructing a homology model of the enzyme and submitting
it for extensive molecular dynamics (MD) simulations to better
sample the positions of all atoms in explicit solvent (Figure 1).
These studies found the active center is predominately
comprised of the anticipated catalytic triad (E48, K130 and
C164), with the following residues that comprise the binding
pocket: Y54, L56, W59, T134, E137, W165, W188, P189,
S190, F191, S192, A197, A198, G202, P203, N206, L246,
Q247, A248, G249, and G250.
The nitrilase reaction mechanism can be divided into three
steps:17 (1) The thiol group of the catalytic C164 which
undergoes nucleophilic attack on the electrophilic carbon of
the substrate nitrile group, forming a thioimidate intermediate;
(2) the thioimidate intermediate is hydrolyzed forming a
second acyl enzyme intermediate, with release of ammonia;
(3) and finally the acyl enzyme intermediate is hydrolyzed to
produce the carboxylic acid. In this overall process, the first
step of the reaction mechanism should be rate limiting and was
therefore chosen as the initial enzyme state for docking of the
(R)- and (S)-substrates.
Homology model validation was conducted by docking (R)-
2-chloromandelonitrile and (S)-2-chloromandelonitrile to WT
BCJ2315 and comparing computation findings with previously
reported experimental findings.7 This analysis found the
binding of (R)-2-chloromandelonitrile was better than that of
its enantiomer, as the coordination between the substrate
nitrile carbon and the C164 thiol group is shorter for (R)-2-
chloromandelonitrile.13 Docking of our substrates ((R)-2 and
(S)-2) in this validated enzyme model revealed that the
reaction coordinate for the first step is shorter and the angle of
the nucleophilic attack is better for (R)-2 and therefore
potentially closer to the Michaelis−Menten complex (Figure
2). Moreover, (R)-2 is able to stabilize stronger intermolecular
interactions within the active site via hydrogen bonding of the
S192 side chain with the oxygen of the methoxy group and also
by hydrogen bonding of W165 with the hydroxyl group of (R)-
2. Both interactions are not present in the docking pose of the
(S)-2. These findings again correlate to the experimentally
observed selective formation of 3 from (R)-2 by BCJ2315.
In silico model analysis focused predominately on identifying
favorable mutations of two general types: first, active site
mutations that could improve selectivity by (1) increasing the
binding affinity for the (R)-2 and (2) decrease the step 1
reaction coordinate of the (R)-2 while subsequently increasing
it for the (S)-2, and second, through B-factor analysis using
four replicate 50 ns MD simulations of BCJ2315 to identify
flexible residues/regions of the enzyme that could be modified
to improve structural stability.13 Additional mutants were also
selected from previous literature reports and through
coevolution analysis of nitrilase sequences.7a In total, 96
mutants were assessed experimentally in the first round of in
vitro screening which revealed that a substantial number of
mutants were able to improve selectivity and/or activity/
stability.13 A list of mutants which gave either increased
selectivity and conversion or an increase in both, relative to
WT, are detailed in Table 4. All mutants were expressed at the
Table 3. Effect of Aldehyde and Temperature on Reaction
a
Profile using Lyophilized Whole Cell BCJ2315
Aldehyde
Loading (g/L)
Temp % AN 3 + % AN ent-3 after
b
Entry
(°C)
24 h (%)
% ee
1
2
3
4
5
6
7
8
68
68
45
37
37
30
21
30
21
19−24
64
95
84
95
96
17
94
93
94
94
94
93
93
N/A
100
100
100
150
150
200
48−89
c
−
a
The reaction mixtures were prepared by mixing 7.5 volumesof
aqueous aldehyde-bisulfite adduct solution, prepared via reaction of
appropriate amount of 1 with 1.1 equiv of NaHSO3 in water, with 1
volume of ethyl acetate, followed by dropwise addition of 1 equiv of
NaCN in 1.5 volumes of water. BCJ2315 (2% by weight relative to 1)
was added, and the reaction mixture was stirred at the appropriate
b
temperature. Determined by HPLC. Full conversion was deemed
achieved when no 2 was detected at 283 nm. Conversion was
determined by [% AN 3 + % AN ent-3]/[% AN 1 + % AN 2 + % AN
c
3 + % AN ent-3].15 No 3 or ent-3 was detected.
process using ethyl acetate cosolvent at higher reaction
temperature and/or substrate concentrations. A further
complicating factor was that at ambient temperatures and
100 g/L of 1 loading, the aqueous solution of the bisulfite-
aldehyde adduct 4 used to prepare cyanohydrin 2 has a
concentration of 0.979 M, which is near its saturation point of
1.02 M.14 This was believed to be a contributing reason for the
observed inconsistent results when increasing the aldehyde 1
loading to 150 g/L (Table 3, entry 7). For this reason, the
target temperature and substrate concentration chosen were 21
°C and 100 g/L of 1 for the process, as it was expected to give
the most consistent results regarding reaction rate, total
turnover, and selectivity. As shown in Table 3, the ee of the
product was not affected by changes to either substrate loading
or temperature under these reaction conditions.
With suitable reaction conditions in hand, we began enzyme
engineering nitrilase BCJ23156 to fit the process with the goal
of improving substrate loading, enzyme robustness, and the
enantioselectivity of 3. A rational in silico approach was used to
identify key enzyme residues for modification based on our
previous success in efficiently re-engineering a transaminase
D
https://doi.org/10.1021/acs.oprd.1c00250
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Figure 1. MD Reference structure of nitrilase BCJ2315.
Figure 2. NIT BCJ2315 docking for (S)-2 (left) and (R)-2 (right). Distances are shown in Å.
same time under the same conditions in 96 well plate format.
SDS-PAGE analysis was performed on the soluble cell lysate
(freeze−thaw lysis) to ascertain if major variations in enzyme
expression were responsible for the increased conversion
performance noted in the best performing mutants.13 This
analysis showed that the mutants which resulted in higher
activity during screening did not have a significantly higher
level of soluble NIT expression compared to the WT. Based on
these initial results the best performing mutants from small
scale testing (M24-S192P, M25-S192G, M43-Y199Q, and
M65-M232V) were grown at 1 L fermentation scale for
subsequent gram scale evaluation.
procedure detailed above in order to assess their robustness
(Figure 3). These reactions were run for longer than the initial
screen (Table 4) with samples taken at multiple time points
and analyzed by HPLC to determine conversion. Variants
M65-M232V and M43-Y199Q were found to have very similar
activity to WT, with the remaining mutants less active.
Confirming the screening results, variants M24-S192P and
M25-S192G showed excellent selectivity (98% and 97% ee,
respectively); however, M24-S192P showed rapid enzyme
deactivation resulting in poor conversion. Similarly, M25-
S192G failed to fully convert the substrate, though its
deactivation was much slower compared to M24-S192P.
Mutants M65-M232V and M43-Y199Q did not show
selectivity improvement versus the WT but did reach full
conversion of 2.
Gram scale time course studies of the best performing
mutant nitrilases from Table 4 versus WT BCJ2315 were
conducted with 5% by weight enzyme loading relative to 1
(total concentration 100 g/L of 1) using the optimized one-pot
E
https://doi.org/10.1021/acs.oprd.1c00250
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Table 4. Best Performing First Generation Mutants of
a
BCJ2315
b
b
Mutant ID
Enzyme Mutation
% ee
Conversion (%)
WT
WT
93.3
90.8
91.4
91.9
92.8
93.5
93.9
94.2
93.7
94.3
97.0
97.7
25.5
33.9
38.9
44.0
33.3
37.2
31.6
36.9
37.3
37.5
27.6
26.5
M77
M55
M65
M49
M10
M93
M42
M43
M4
A323S
A108N
M232V
A64E
W59F
L201I
Y199C
Y199Q
T49L
M25
M24
S192G
S192P
a
0.5 mg of freeze-dried whole cells enzymes were incubated in a 96-
Figure 4. Time course study for M100, M25, and WT BCJ2315
enzymes. Determined by HPLC at 283 nm. Conversion was
determined by [%AN 3 + % AN ent-3]/[% AN 1 + % AN 2 + %
AN 3 + % AN ent-3].
well plate with 10 mg of 2 at a concentration of 10 mg/mL in a 10:1
mixture of 0.1 M sodium carbonate buffer (pH 9) and EtOAc at room
temperature in a microplates shaker at 1500 rpm for 2 h. Determined
by HPLC at 283 nm. Conversion was determined by [%AN 3 + % AN
ent-3]/[% AN 1 + % AN 2 + % AN 3 + % AN ent-3].
b
substantial change on the substrate active site posed imparted
via the S192G mutation. In mutant S192G, the T134 side
chain forms a stabilizing hydrogen bond with the (R)-2
methoxy group, while in the WT the substrate methoxy group
instead forms a stabilizing hydrogen bond with the S192 side
chain. As a result, for the first reaction step (nucleophilic attack
of the SH group of the active center Cys on the electrophilic
carbon of the substrate nitrile group), the reaction coordinate
of (R)-2 is shorter (2.9 Å) and the angle of the attack is better
than in the WT ((R)-2 (3.5 Å) (Figure 2 and Figure 5).
Selective conversion of the (R)-substrate should therefore
occur at a higher rate for the S192G mutant than the WT
BCJ2315.
Based on B-factors analysis, M232 is predicted to be an
unstable residue.13 Mutation M232V removes the methionine
residue from the surface, potentially preventing associated
oxidation issues and therefore increasing the intermolecular
interactions within a hydrophobic pocket which contribute to
the increased structural stability of the enzyme (Figure 6). The
combination of both mutations (S192G and M232V) was
required to overcome the reduced stability observed with the
selectivity enhancing S192G mutation alone.
With a robust and highly selective reaction process in hand,
we began optimization of the downstream process. At the end
of the hydrolysis we noted that the reaction was a suspension
of cells at pH ≈ 8.4. We envisioned the initial process workup
operations to involve a Celite filtration to remove cellular
debris followed by adding an additional 1 volume of ethyl
acetate to generate a layer cut and remove unreacted
benzaldehyde from the aqueous workup stream containing 3.
Additional stress studies were also carried out to investigate the
downstream impact of removing the ethyl acetate addition and
layer cut altogether. It was found that without this unit
operation any remaining benzaldehyde could still be
completely purged to the current in-process control limit of
<3% 1, in the final ethyl acetate/diisopropyl ether/heptanes
crystallization of 3. Ultimately, the minimal yield losses
(<0.25% 3) along with downstream process impurity concerns
led us to retain this layer cut operation. While our initial Celite
filtrations were slow, they were successful in removing the bulk
of the cellular debris, enabling visualization of the resulting
cloudy white water and clear ethyl acetate layers after an
Figure 3. Time course study for select best performing first
generation mutants. Determined by HPLC at 283 nm. Conversion
was determined by [%AN 3 + % AN ent-3]/[% AN 1 + % AN 2 + %
AN 3 + % AN ent-3]
Although the first generation of mutants failed to identify a
mutant NIT that addressed both selectivity and activity, a
second generation series of double mutants combining the
increased selectivity of M25-S192G with the improved activity
of M43-Y199Q and M65-M232V was deemed worthwhile for
further evaluation. Evaluation of these double mutant
combinations on gram scale using the one-pot conditions
(5% by weight lyophilized whole cell, 22 °C) revealed that the
double mutant M100-S192G-M232V gave the best perform-
ance, achieving improved stability compared to the single
mutation variant M25-S192G while maintaining the selectivity
of the M65-M232V variant (97.0% ee and 72% conversion vs
97.0% ee and 64% conversion for M25-S192G over 4 h, Figure
4), and with only slightly lower activity than the wild type
enzyme. Additionally, we note that analysis of the crude
reaction mixture did not show any presence of the
corresponding mandelic amide, a typical impurity observed
in nitrilase processes, simplifying our downstream isolation
processes.
The observed selectivity improvement of M100-S192G-
M232V versus WT BCJ2315 can be rationalized through the
F
https://doi.org/10.1021/acs.oprd.1c00250
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
Figure 5. NIT BCJ2315_S192G docked to (S)-2 (left) and (R)-2 (right). Distances shown are in Å.
Figure 6. BCJ2315 (left) vs BCJ2315_M232V (right). Distances shown are in Å.
additional 1 volume of both ethyl acetate and water were
added. Further improvement in both the filtration time and
initial layer cut at pH ≈ 8 was also possible through the use of
an Akta Flux S cross-flow filtration system using a hollow fiber
cartridge containing a 0.2 μm polysulfone membrane.
very clear layer cut. Additional optimization found that
increasing the MgSO4 loading from 1.5× to 2× more
effectively resulted in collapse of the problematic protein
emulsion (Figure 7).
In the next step of the isolation process the acidification to
pH ≈ 1 with conc. HCl and extraction of 3 using ethyl acetate
presented a significant operational challenge due to the
formation of a white emulsion in the ethyl acetate layer due
to emulsification of solubilized protein carryover at this low
pH. Spiking studies for different organic comixtures using ethyl
acetate/M100 lyophilized cell supernatant revealed a signifi-
cant increase in solubility of 3 versus unspiked ethyl acetate,
leading to concerns about protein carryover and its impact on
mother liquor losses in the crystallization. As such, efforts were
focused on enzyme removal and a resolution to the emulsion
challenge. Initial studies focused on protein removal either
through precipitation by aging for 2 h at 60 and 90 °C prior to
Celite filtration or through treatment at pH 10 overnight in an
effort to digest or break down proteins. In all cases no
significant improvement was observed after acidification to pH
≈ 1. Several salt additives were then evaluated in an effort to
break the emulsion. While charging 1.5× NaCl, CaCl2 or
Na2SO4 after pH adjustment did not lead to any significant
change in the emulsion formation, fortuitously, MgSO4 was
found to break the emulsion forming a thin rag layer at the
interface that could be subsequently polish filtered to reveal a
Following layer cut of the resulting clear ethyl acetate/water
layers, the pH ≈ 1 aqueous layer was further back extracted
Figure 7. Impact of select salt additives and their loading on emulsion
formation at pH ≈ 1.
G
https://doi.org/10.1021/acs.oprd.1c00250
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
with ethyl acetate twice in order to minimize loss of 3 to the
aqueous layer (<0.25% 3).
nitrilase process. Rational protein engineering to improve
enzyme selectivity and robustness using in silico analysis
identified a number of favorable mutations via active site and
B-factor analysis for initial evaluation. This efficient and low-
cost approach led to the identification of two beneficial
mutations over the WT nitrilase with respect to enantiose-
lectivity and robustness. A plug-and-play enzyme change was
feasible without problems; the improved variant enzyme gave
similar activity and higher product enantioselectivity in
conditions developed for maximized enzymatic turnover with
the wild type. Reaction design identified enzyme deactivation
by the aldehyde as a key obstacle for achieving a robust one-
pot process. Ultimately, this observation led to the discovery
that the addition of sodium bisulfite could prevent this
deactivation thereby allowing for a simplified high substrate
loading process that avoided a fed-batch approach. Down-
stream process optimization activities also identified magne-
sium sulfate as a critical workup additive to avoid emulsion
formation, allowing for the isolation of the desired final
product in good yield and in excellent enantioselectivity with
high control of residual protein content.
Our initial crystallization efforts for 3 focused on a
streamlined process whereby the organic layers were combined
and concentrated to 1.5 volumes under vacuum and heated to
50 °C to ensure full dissolution followed by ramp cooling to
25−30 °C, seeding, and further cooling to 0 °C prior to
filtration. While this process gave a mobile slurry and ee
upgrade to >99.0%, mother liquor losses were unacceptably
high (15−25%), owing to the high solubility of 3 in ethyl
acetate (ca. 6 mL ethyl acetate/g 3). Solubility screening of 3
in a wide range of solvents identified a small handful of
potential antisolvents for the crystallization of 3: heptane(s),
toluene, xylenes, and ethereal solvents such as dibutyl ether
and diisopropyl ether. After extensive screening of various
solvent combinations, the preferred crystallization process
distilled the combined ethyl acetate layers to 1.5 volumes
under vacuum followed by addition of 1.5× diisopropyl ether
antisolvent. The mixture was then heated to 65 °C for 1 h to
ensure full dissolution and then cooled to 50 °C and held at 50
°C for 3 h. During this holding process a hazy reaction mixture
was observed, at which point seeding was performed to
produce a mobile slurry. In order to minimize mother liquor
losses, heptanes (1.0×) was then slowly added at 50 °C over
ca. 2 h to prevent oiling out of 3. Cooling and filtration at −20
°C afforded 3 in 60−75% yield and in high (96−100%) ee and
good assay purity with acceptable mother liquor losses (ca.
10% yield).
EXPERIMENTAL SECTION
■
Solvents and reagents were obtained from commercial sources
and used without further purification. 1H NMR was performed
on a Bruker Avance III HD 400 MHz spectrometer (1H NMR
at 400 MHz, 13C NMR at 101 MHz) with solvent resonances
as the internal standard (1H NMR: DMSO-d6 at 2.50 ppm;
13C NMR: DMSO-d6 at 39.52 ppm). 1H NMR data are
reported as follows: chemical shift, integration, multiplicity (s
= singlet, d = doublet, t = triplet, dd = doublet of doublets, dt =
doublet of triplets), and coupling constants (Hz). HPLC/
UPLC was performed on an Agilent 1260 system equipped
with a PDA detector. Bradford assay analysis was performed
using an Agilent Cary 60 UV−vis spectrophotometer equipped
with Cary Win UV software using BSA (lyophilized powder,
Sigma) and Coomassie Plus Assay Reagent (ThermoFisher
Scientific).
Enzyme analysis using the Bradford assay for the finalized
process waste streams and final solid was also evaluated to
determine effective purge points and to assess downstream
process risk. As shown in Table 5, several effective protein
Table 5. Enzyme Content in Waste Streams for Finalized
Process for 3
Protein
Concentration
(ppm)
a
Weight % of Total
Biomass Charged
Analysis Point
Initial ethyl acetate layer waste
stream after Celite filtration
(pH ≈ 8)
Aqueous layer waste stream after final Not detected
ethyl acetate extraction (pH ≈ 1)
Mother liquor waste stream (ethyl
acetate/diisopropyl ether/
heptanes)
553
1.7
Preparation of (R)-2-Methoxymandelic Acid (3). To a
solution of 2-methoxybenzaldehyde (40.00 g, 293.8 mmol,
1.00 equiv) in ethyl acetate (40.03 g) was added solid NaHSO3
(30.58 g, 293.9 mmol, 1.00 equiv) and water (300.04 g). After
the addition was complete, the reaction mixture was agitated
for an additional ca. 60 min, resulting in a colorless hazy
solution. To this reaction mixture was then slowly added a
solution of NaCN (14.69 g, 299.7 mmol, 1.02 equiv) in water
(60.03 g) while maintaining the pot temperature below 25 °C.
After the addition was complete, the reaction mixture was
agitated for an additional ca. 60 min. To this reaction mixture
was then added M100 dried whole cell enzyme (2.04 g, 5% w/
w relative to the aldehyde). After the addition was complete,
the reaction mixture was agitated for an additional 24 to 48 h
until the reaction was deemed complete by HPLC. The
reaction mixture was then filtered through a layer of Celite, and
the reactor and Celite were rinsed forward with water (40.02
g). The combined filtrate (caution: trace NaCN may still be
present in the filtrate) were washed with ethyl acetate (40.03
g). The organic layer was discarded, and the aqueous layer was
acidified to pH 1 with conc. HCl resulting in formation of an
emulsion. MgSO4 (80.05 g) was then added to the aqueous
layer, and the resulting solution was stirred for ca. 15 min to
break the emulsion. Ethyl acetate (100.04 g) was then added,
and the resulting rag layer at the interface was removed via a
b
0.0
2.0
391
Isolated 3
91
0.2
a
Determined using Bradford assay. See Supporting Information for
b
details. Detection limit = 10 ppm.
purge points were built into the process, the main ones being
the initial ethyl acetate wash and final isolation. Together,
these two points enable 95% purge of the protein carried over
after the initial Celite filtration, which was deemed suitable for
our intended downstream process use of 3.
CONCLUSION
■
The concurrent process development and enzyme engineering
activities resulted in the development of a superior process
versus what could be achieved by performing these activities
individually without considering their respective impacts on
each other. This approach helped identify initial protein
preparation limitations and enabled the concurrent adjustment
of the process chemistry conditions to produce a more robust
H
https://doi.org/10.1021/acs.oprd.1c00250
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
polish filtration. The organic layer was removed, and the
aqueous layer was extracted with ethyl acetate (2 × 100.05 g).
The combined organic layers were concentrated under vacuum
to approximately 60 mL (ca. 1.5 volumes) using a reactor
jacket temperature of approximately 55 °C. Diisopropyl ether
(60.03 g) was then added in one portion to the resulting pale
yellow hazy solution and agitated at about 65 °C for
approximately 1 h. The reaction mixture was then cooled
over 2 h to approximately 50 °C, and then seed (0.1216 g) was
charged resulting in slurry formation. Once a slurry was
confirmed, heptanes (40.05 g) was charged over not less than 2
h. After addition, the slurry was then cooled over about 7 h to
approximately −20 °C and agitated at −20 °C for not less than
3 h. The final product was filtered and washed with precooled
diisopropyl ether (40.05 g) at about −10 °C. The final solids
were dried at 45 °C in a vacuum oven with nitrogen sweep for
not less than 4 h to provide 37.27 g of the title compound
(70% yield, 99.10% assay purity, 96.89 ee% determined by
UPLC at 283 nm) as a white solid. Average yields ranged from
60% to 75% with a purity range of 95−103 assay% by qNMR
and 96−100% ee by UPLC at 283 nm.
Daniel F. A. R. Dourado − Almac Sciences, Craigavon BT63
5QD, U.K.
Darren Gray − Almac Sciences, Craigavon BT63 5QD, U.K.
Stefan Mix − Almac Sciences, Craigavon BT63 5QD, U.K.
Aliaksei Pukin − Almac Sciences, Craigavon BT63 5QD, U.K.
Complete contact information is available at:
https://pubs.acs.org/10.1021/acs.oprd.1c00250
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The authors thank Mark Davis, Lars Heumann, Jeff Ng, and
Richard Yu for their helpful discussions, input, and feedback
during the review of the manuscript.
ABBREVIATIONS
% AN percent area normalization
CFE cell free enzyme
■
EC
HPLC high performance liquid chromatography
MD molecular dynamics
enzyme commission
1H NMR (DMSO-d6, 400 MHz) δ: 12.34 (1H, broad s),
7.35 (1H, dd, J = 1.8, 7.4 Hz), 7.27 (1H, dt, J = 2.0, 7.8 Hz),
6.97 (1H, d, J = 8.4 Hz), 6.94 (1H, dt, J = 0.8, 3.6 Hz), 5.30
(1H, s), 3.77 (3H, s); 13C NMR (DMSO-d6, 100 MHz) δ:
174.4, 156.6, 129.1, 128.8, 128.0, 120.4, 111.2, 67.0, 55.6.
NIT nitrilase
UPLC ultra high performance liquid chromatography
WT
X
wild type
weight ratio
ASSOCIATED CONTENT
REFERENCES
■
■
sı
* Supporting Information
The Supporting Information is available free of charge at
https://pubs.acs.org/doi/10.1021/acs.oprd.1c00250.
(1) (a) Wu, S.; Snajdrova, R.; Moore, J. C.; Baldenius, K.;
Bornscheuer, U. T. Biocatalysis: Enzymatic Synthesis for Industrial
Applications. Angew. Chem., Int. Ed. 2021, 60, 88−119. (b) Abdelra-
heem, E. M. M.; Busch, H.; Hanefeld, U.; Tonin, F. Biocatalysis
Explained: From Pharmaceutical to Bulk Chemical Production. React.
Chem. Eng. 2019, 4, 1878−1894. (c) Pharmaceutical Biocatalysis
Chemoenzymatic Synthesis of Active Pharmaceutical Ingredients;
Grunwald, P., Ed.; Jenny Stanford Publishing: Singapore, 2020.
(2) (a) Enzyme Nomenclature 1992: Recommendations of the
Nomenclature Committee of the International Union of Biochemistry and
Molecular Biology on the nomenclature and classification of enzymes;
Webb, E., Ed.; Academic Press: New York, 1992. (b) Enzyme
Nomenclature. https://www.qmul.ac.uk/sbcs/iubmb/enzyme/ (ac-
cessed 2021-06-03).
(3) Shen, J.-D.; Cai, X.; Liu, Z.-Q.; Zheng, Y.-G. Nitrilase: A
Promising Biocatalyst in Industrial Applications for Green Chemistry.
Crit. Rev. Biotechnol. 2021, 41, 72−93.
(4) (a) Pace, H. C.; Brenner, C. The Nitrilase Superfamily:
Classification, Structure and Function. Genome Biol. 2001, 2,
reviews0001.1. (b) Brenner, C. Catalysis in the Nitrilase Superfamily.
Curr. Opin. Struct. Biol. 2002, 12, 775−782.
Complete experimental procedures and characterization
data including: biocatalyst preparation, computational
methods of homology model building and MD
simulation, nitrilase mutant preparation and evaluation
(rounds 1 and 2), preparation of bisulfite adduct 4,
Bradford assay method, HPLC method, NMR spectra
(PDF)
AUTHOR INFORMATION
■
Corresponding Author
Luke D. Humphreys − Gilead Alberta ULC, Edmonton,
Alberta, Canada T6S 1A1; orcid.org/0000-0001-6497-
4339; Email: Luke.Humphreys@Gilead.com
Authors
(5) See for example: (a) Baum, S.; van Rantwijk, F.; Stolz, A.
Application of a Recombinant Escherichia coli Whole-Cell Catalyst
Expressing Hydroxynitrile Lyase and Nitrilase Activities in Ionic
Liquids for the Production of (S)-Mandelic Acid and (S)-
Mandeloamide. Adv. Synth. Catal. 2012, 354, 113−122. (b) Bhatia,
S. K.; Mehta, P. K.; Bhatia, R. K.; Bhalla, T. C. Optimization of
Arylacetonitrilase Production from Alcaligenes sp. MTCC 10675 and
its Application in Mandelic Acid Synthesis. Appl. Microbiol. Biotechnol.
2014, 98, 83−94. (c) He, Y.-C.; Zhang, Z.-J.; Xu, J.-H.; Liu, Y.-Y.
Biocatalytic Synthesis of (R)-(2)-Mandelic Acid from Racemic
Mandelonitrile by Cetyltrimethylammonium Bromidepermeabilized
Cells of Alcaligenes Faecalis ECU0401. J. Ind. Microbiol. Biotechnol.
2010, 37, 741−750. (d) Ni, K.; Wang, H.; Zhao, L.; Zhang, M.;
Zhang, S.; Ren, Y.; Wei, D. Efficient Production of (R)-(−)-Mandelic
Acid in Biphasic System by Immobilized Recombinant E. coli. J.
Mark E. Scott − Gilead Alberta ULC, Edmonton, Alberta,
Canada T6S 1A1; Present Address: Flare Therapeutics,
215 First Street, Suite 150, Cambridge, Massachusetts,
02142, United States
Xiaotian Wang − Gilead Alberta ULC, Edmonton, Alberta,
Canada T6S 1A1
Michael J. Geier − Gilead Alberta ULC, Edmonton, Alberta,
Canada T6S 1A1
Balamurali Kannan − Gilead Alberta ULC, Edmonton,
Alberta, Canada T6S 1A1
Johann Chan − Gilead Sciences, Inc, Foster City, California
94404, United States; Present Address: Praxis Precision
Medicines, One Broadway, Cambridge, Massachusetts,
02142, United States
̌
̌
Biotechnol. 2013, 167, 433−440. (e) Petrícková, A.; Sosedov, O.;
Gareth Brown − Almac Sciences, Craigavon BT63 5QD, U.K.
Baum, S.; Stolz, A.; Martínková, L. Influence of Point Mutations Near
I
https://doi.org/10.1021/acs.oprd.1c00250
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
pubs.acs.org/OPRD
Article
the Active Site on the Catalytic Properties of Fungal Arylacetoni-
trilases from Aspergillus niger and Neurospora crassa. J. Mol. Catal. B:
Enzym. 2012, 77, 74−80. (f) Sosedov, O.; Stolz, A. Random
Mutagenesis of the Arylacetonitrilase from Pseudomonas fluorescens
EBC191 and Identification of Variants, Which Form Increased
Amounts of Mandeloamide from Mandelonitrile. Appl. Microbiol.
Biotechnol. 2014, 98, 1595−1607. (g) Wang, S.-Z.; Wang, Z.-K.;
Gong, J.-S.; Qin, J.; Dong, T.-T.; Xu, Z.-H.; Shi, J.-S. Improving the
Biocatalytic Performance of Co-immobilized Cells Harboring
Nitrilase via Addition of Silica and Calcium Carbonate. Bioprocess
Biosyst. Eng. 2020, 43, 2201−2207. (h) Zhang, X.-H.; Liu, Z.-Q.; Xue,
Y.-P.; Wang, Y.-S.; Yang, B.; Zheng, Y.-G. Production of R-Mandelic
Acid Using Nitrilase from Recombinant E. coli Cells Immobilized with
Tris(Hydroxymethyl)Phosphine. Appl. Biochem. Biotechnol. 2018, 184,
1024−1035. (i) Zhang, X.-H.; Wang, C.-Y.; Cai, X.; Xue, Y.-P.; Liu,
Z.-Q.; Zheng, Y.-G. Upscale Production of (R)-Mandelic Acid with a
Stereospecific Nitrilase in an Aqueous System. Bioprocess Biosyst. Eng.
2020, 43, 1299−1307.
(16) Dourado, D. F. A. R.; Pohle, S.; Carvalho, A. T. P.; Dheeman,
D. S.; Caswell, J. M.; Skvortsov, T.; Miskelly, I.; Brown, R. T.; Quinn,
D. J.; Allen, C. C. R.; Kulakov, L.; Huang, M.; Moody, T. S. Rational
Design of a (S)-Selective-Transaminase for Asymmetric Synthesis of
(1S)-1-(1,1′-biphenyl-2-yl)ethanamine. ACS Catal. 2016, 6, 7749−
7759.
(17) Kobayashi, M.; Goda, M.; Shimizu, S. Nitrilase Catalyzes
Amide Hydrolysis as well as Nitrile Hydrolysis. Biochem. Biophys. Res.
Commun. 1998, 253, 662−666.
(6) Wang, H.; Sun, H.; Wei, D. Discovery and Characterization of a
Highly Efficient Enantioselective Mandelonitrile Hydrolase from
Burkholderia Cenocepacia J2315 by Phylogeny-Based Enzymatic
Substrate Specificity Prediction. BMC Biotechnol. 2013, 13, 14.
(7) (a) Wang, H.; Gao, W.; Sun, H.; Chen, L.; Zhang, L.; Wang, X.;
Wei, D. Protein Engineering of a Nitrilase from Burkholderia
Cenocepacia J2315 for Efficient and Enantioselective Production of
(R)-2-Chloromandelic acid. Appl. Environ. Microbiol. 2015, 81, 8469−
8477. (b) Wang, H.; Sun, H.; Gao, W.; Wei, D. Efficient Production
of (R)-o-Chloromandelic Acid by Recombinant Escherichia coli Cells
Harboring Nitrilase from Burkholderiacenocepacia J2315. Org. Process
Res. Dev. 2014, 18, 767−773.
(8) Kobayashi, M.; Nagasawa, T.; Yamada, H. Nitrilase of
Rhodococcus Rhodochrous J1. Purification and Characterization.
Eur. J. Biochem. 1989, 182, 349−356.
(9) (a) Dennett, G. V.; Blamey, J. M. A New Thermophilic Nitrilase
from an Antarctic Hyperthermophilic Microorganism. Front. Bioeng.
Biotechnol. 2016, 4, 5. (b) Khandelwal, A. K.; Nigam, V. K.; Vidyarthi,
A. S.; Ghosh, P. Evaluation of Various Ions and Compounds on
Nitrilase Produced from Streptomyces sp. Artif. Cells, Blood Subst.
Immobil. Biotechnol. 2010, 38, 13−18.
(10) Thuku, R. N.; Brady, D.; Benedik, M. J.; Sewell, B. T. Microbial
Nitrilases: Versatile, Spril Forming, Industrial Enzymes. J. Appl.
Microbiol. 2009, 106, 703−727.
(11) (a) Kiziak, C.; Conradt, D.; Stolz, A.; Mattes, R.; Klein, J.
Nitrilase from Pseudomonas fluorescens EBC191: Cloning and
Heterologous Expression of the Gene and Biochemical Character-
ization of the Recombinant Enzyme. Microbiology 2005, 151, 3639−
3648. (b) Banerjee, A.; Kaul, P.; Banerjee, U. C. Enhancing the
Catalytic Potential of Nitrilase from Pseudomonas putida for
Stereoselective Nitrile Hydrolysis. Appl. Microbiol. Biotechnol. 2006,
72, 77−87. (c) Banerjee, A.; Kaul, P.; Banerjee, U. C. Erratum to:
Enhancing the Catalytic Potential of Nitrilase from Pseudomonas
putida for Stereoselective Nitrile Hydrolysis. Appl. Microbiol.
Biotechnol. 2017, 101, 6299. (d) Howden, A. J.; Preston, G. M.
Nitrilase Enzymes and their role in Plant-microbe Interactions.
Microb. Biotechnol. 2009, 2, 441−451.
(12) Wang, H.; Fan, H.; Sun, H.; Zhao, Li; Wei, D. Process
Development for the Production of (R)-(−)-Mandelic Acid by
Recombinant Escherichia coli Cells Harboring Nitrilase from
Burkholderia cenocepacia J2315. Org. Process Res. Dev. 2015, 19,
2012−2016.
(13) See Supporting Information for more details.
(14) Determined by preparing 4 and measuring the amount of water
needed to fully dissolve a known amount of compound with shaking.
See Supporting Information for more details.
(15) Conversion calculations did not include the % AN of the
bisulfite adduct 4 as it was found to be unstable under the sample
preparation/method analysis, converting to the aldehyde which was
measured instead.
J
https://doi.org/10.1021/acs.oprd.1c00250
Org. Process Res. Dev. XXXX, XXX, XXX−XXX