Focused Directed Evolution of Pentaerythritol Tetranitrate Reductase by Using Automated Anaerobic Kinetic Screening of Site-Saturated Libraries

DOI: 10.1002/cbic.201000527

Focused Directed Evolution of Pentaerythritol Tetranitrate

Reductase by Using Automated Anaerobic Kinetic

Screening of Site-Saturated Libraries

Martyn E. Hulley,^[a]Helen S. Toogood,^[a]Anna Fryszkowska,^[b]David Mansell,^[b]

Gill M. Stephens,^[c]John M. Gardiner,^[b]and Nigel S. Scrutton*^[a]

This work describes the development of an automated robotic

platform for the rapid screening of enzyme variants generated

from directed evolution studies of pentraerythritol tetranitrate

(PETN) reductase, a target for industrial biocatalysis. By using a

96-well format, near pure enzyme was recovered and was suit-

able for high throughput kinetic assays; this enabled rapid

screening for improved and new activities from libraries of

enzyme variants. Initial characterisation of several single site-

saturation libraries targeted at active site residues of PETN

reductase, are described. Two mutants (T26S and W102F) were

shown to have switched in substrate enantiopreference

against substrates (E)-2-aryl-1-nitropropene and a-methyl-

trans-cinnamaldehyde, respectively, with an increase in ee

(62% (R) for W102F). In addition, the detection of mutants

with weak activity against a,b-unsaturated carboxylic acid sub-

strates showed progress in the expansion of the substrate

range of PETN reductase. These methods can readily be adapt-

ed for rapid evolution of enzyme variants with other oxido-

reductase enzymes.

Introduction

The use of enzymes in asymmetric synthesis has become an

increasingly cost-effective and established technique to manu-

facture fine chemicals, pharmaceuticals and agrochemical in-

termediates.^[1]This is due to the often unmatched efficiency,

precision, stereo- and enantioselectivity of enzyme-catalysed

reactions.^[1–4]However, naturally occurring biocatalysts might

lack some properties required for large-scale chemical produc-

tion, such as high activity and selectivity for non-natural sub-

strates.^[5]To overcome the limitations in the functioning of en-

zymes in industrial biotransformations, directed evolution tech-

niques in combination with high-throughput screening have

been developed to improve the characteristics of enzymes in a

targeted manner.^[2,4]

the Old Yellow Enzyme family (OYE; E.C. 1.6.99.1) in the reduc-

tion of activated a,b-unsaturated alkenes.^[9]These FMN-con-

taining, NAD(P)H-dependent oxidoreductases catalyse the

reduction of a,b-unsaturated ketones, aldehydes, nitroalkenes,

carboxylic acids and derivatives, to yield products with a

variety of biotechnological and pharmaceutical applications

(Scheme 1).^[9,10]Members of this family include the enzymes

OYE1 (brewer’s yeast),^[11]pentaerythritol tetranitrate reductase

(PETN reductase; Enterobacter cloacae PB2),^[12]YqjM (Bacillus

subtillis),^[13]and a recently described thermostable member

TOYE (Thermo-anaerobacter pseudethanolicus E39).^[14]

The OYE family of enzymes are classic examples of exten-

sively characterised biocatalysts capable of reducing many in-

dustrially relevant substrates. However, family members also

lack sufficient activity for formation of important industrial syn-

thons, have low stereo- and/or enantioselectivity or can form

the wrong enantiomer. Recent site-saturation mutagenesis

studies of OYE1 at residue W116 generated mutants with re-

versed stereochemical outcomes compared to the wild type.^[15]

These techniques involve the generation of libraries of

mutant enzymes (single to multi-amino acid changes) by tech-

niques such as random and site-directed mutagenesis of

known key residues.^[1,2,6]Such techniques can be applied to

improving key characteristics of enzymes, such as substrate

binding, stereo- and enantioselectivity of substrates and/or

products, and even increasing the biocatalyst stability in indus-

trial reaction conditions (e.g., high organic solvent content).^[1,7]

One drawback with such methods is the requirement of

screening large numbers of library mutants to find improved

biocatalysts, although recently developed methods of iterative

saturation mutagenesis combined with structural- and func-

tional-based residue targeting can dramatically reduce the li-

brary size.^[6]

[a] M. E. Hulley, Dr. H. S. Toogood, Prof. N. S. Scrutton

Faculty of Life Sciences, University of Manchester

131 Princess Street, Manchester M1 7DN (UK)

Fax: (+44)161-306-8918

E-mail: nigel.scrutton@manchester.ac.uk

[b] Dr. A. Fryszkowska, Dr. D. Mansell, Dr. J. M. Gardiner

School of Chemistry, University of Manchester

131 Princess Street, Manchester M1 7DN (UK)

[c] Prof. G. M. Stephens

This approach was exemplified by applying it to the asym-

metric reduction of activated C=C bonds; an important synthe-

sis step as up to two stereogenic centres can be formed.^[8]

Recent work has demonstrated the biocatalytic potential of

Department of Chemical and Environmental Engineering

University of Nottingham, University Park, Nottingham NG7 2RD (UK)

Supporting information for this article is available on the WWW under

http://dx.doi.org/10.1002/cbic.201000527.

ChemBioChem 2010, 11, 2433 – 2447

ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

2433

N. S. Scrutton et al.

Results and Discussion

Library selection

A C-terminal His-tag was incorporated into an existing highly

expressing PETN reductase construct^[17,18]to enable a rapid,

highly efficient and high-throughput method of biocatalyst

purification. The apparent kinetic constants of both the steady

state and reductive/oxidative half reactions were essentially

unchanged with the substrates NADPH and 2-cyclohexen-1-

one (1a; Table 1; Figure S2 in the Supporting Information). This

is in contrast to the OYE family member YqjM structural integ-

rity and catalytic efficiency of which was severely compromised

by the addition of a C-terminal His₆-tag.^[19]

Scheme 1. General mechanism and stereochemistry of a,b-unsaturated

cyclic ketone substrate reduction by PETN reductase. R¹, R², R³, R⁴=CH₃, H,

(CH₃)₂, (H)₂or H, CH₃, (H)₂, (CH₃)₂for ketoisophorone bound in two confor-

mations.

The steady-state kinetics and biocatalytic potential (yield

and enantiopreference) of wild-type PETN reductase has been

studied previously by using a variety of a,b-unsaturated alde-

hydes, ketones and nitro-olefins as substrates.^[20–22]The high-

throughput screening protocols were tested against libraries of

single generation site-saturated mutants of PETN reductase for

improvements in the steady state reaction rate of NADPH oxi-

dation in the presence of a variety of known poor substrates,

and screened against compounds not reduced by the wild-

type enzyme. Mutants deemed “hits” by our protocols were

produced on a larger scale (1 L) and subjected to further

steady state analysis and biotransformation reactions to con-

firm alkene reduction had occurred, and to see whether the

mutations had any effect on the product enantiopurity.

A more extensive iterative saturation mutagenesis study was

conducted with the OYE homologue, YqjM, at 20 active site

residue positions including C26 and Y28.^[16]The screening pro-

cedures involved the use of crude cell preparations, with prod-

uct detection and stereochemistry monitored by GC.

A

number of single and multiple residue mutants were generat-

ed with improved performance, such as enhanced steady state

kinetics against selected substrates, better enantiopurity of the

products, and the generation of mutants with opposite enan-

tiopreferences.

This study describes the development of an automated, ro-

botic anaerobic high-throughput procedure for screening libra-

ries of enzymes against industrially useful substrates. Protocols

include fully automated colony picking, extraction and micro-

scale purification of the enzymes by using Ni-magnetic agarose

beads, followed by steady state reactions with up to three sub-

strates per library extraction. The approach taken was to

screen for improvements in the steady state catalysis rather

than directly monitoring for enhancement in yield and ee. This

gives the advantages of decreased reaction times, faster

screening of potential substrates, and normalisation of biocata-

lyst concentration. The use of purified enzymes rather than

whole cell extracts allows us to remove the majority of con-

taminating Escherichia coli en-

Our selection criteria for mutation positions was based on

targeting residues known or predicted to be involved in sub-

strate/inhibitor binding and/or catalysis, or due to their prox-

imity within the active site. Previous studies have shown or

predicted that residues H181, H184 and Y351 are involved in

substrate/inhibitor binding in OYEs.^[16,21–23]Residues W102 and

T26 are positioned close to, and potentially clash or interact

with picric acid and progesterone.^{[16,21,23,24]}In addition, T26 is

known to interact with the FMN cofactor and influence the

redox potential of OYEs.^[16,22,25]Residue Y186 was targeted; it is

zymes, such as keto reductases,

which have previously been

Table 1. Apparent: A) steady state kinetic constants, and B) rate constants for the oxidative and reductive half

reactions of PETN reductase with 2-cyclohexen-1-one 1a. Comparison of the apparent kinetic parameters of

native and His₈-tagged PETNR.

found to catalyse undesirable

side reactions in the screening

of YqjM mutant libraries.^[16]Posi-

tive “hits” are later subjected to

screening for improved yields

and enantiopurity of the prod-

ucts. This study describes the ini-

tial high-throughput screening

of libraries of single generation

site-saturated mutants of PETN

reductase for improvements in

steady state reaction rates

against a variety of typical and

potential OYE substrates.

A) Steady state^[a]

B) Half reactions^[b]

Substrate

Enzyme

k_cat

[s^ꢀ1

K_M

[mm]

K_i

[mm]

k_oxorred

[s^ꢀ1

K_d

[mm]

]

NADPH

wild type

His₈

wild type^[c]

His₈

8.48ꢁ0.77

8.72ꢁ0.39

6.12ꢁ0.17

5.86ꢁ0.38

128.0ꢁ24.5

115.2ꢁ11.3

1239ꢁ81

n.d.

23.3ꢁ1.7

21.0ꢁ3.5

34.9ꢁ0.4

33.1ꢁ0.4

33.3ꢁ1.5

24.7ꢁ0.8

70.9ꢁ4.6

81.8ꢁ4.7

8100ꢁ110

7600ꢁ810

alkene 1a

1200ꢁ210

[a] Reactions (1 mL) were performed in buffer (50 mm KH₂PO₄/K₂HPO₄, pH 7.0) containing NADPH (10–250 mm),

2-cyclohexen-1-one (1a; 0.05–50 mm added as concentrated stocks in ethanol), and PETNR (100 nm). The reac-

tions were followed continuously by monitoring NADPH oxidation at 340 nm for 2 min at 258C. Concentration

dependence of NADPH and 1a was determined with a constant concentration of 1a (6.5 mm) and NADPH

(150 mm), respectively. [b] Reactions (1 mL) were performed in buffer (50 mm KH₂PO₄/K₂HPO₄, pH 7.0) contain-

ing PETNR (100 mm) and FMN reduction/oxidation was monitored at 464 nm for 1 s at 258C for the reductive

and oxidative half reactions, respectively. In the oxidative half reaction, PETNR FMN was previously reduced to

the hydroquinone species by dithionite titration. [c] Previously published data;^[20]n.d.: not determined.

2434

www.chembiochem.org

ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ChemBioChem 2010, 11, 2433 – 2447

Site-Saturated Library Screening of PETN Reductase

the proton donor in a number of OYEs, although the role of

this residue is less clear in PETN reductase.^[26]Two other resi-

dues (Y68 and Q241) were targeted due to their close proximi-

ty within the substrate binding site, the latter of which is locat-

ed in a highly flexible loop. We generated eight single site-

saturated libraries of mutants (T26X, Y68X, W102X, H181X,

H184X, Y186X, Q241X and Y351X) by PCR with degenerate oli-

gonucleotides (NNK) at the required amino acid positions. To

ensure 95% coverage of all possible amino acids per mutation

site, a minimum of 94 colonies needed to be screened when

using NNK degeneracy.^[16]One library of 96 clones was

screened per amino acid position against up to eight alkene

substrates (total screening effort: 4896 individual activity rates,

excluding repeats and wild-type controls).

fied protein levels of ten randomly selected clones were moni-

tored by SDS-PAGE (Figure 1A and B; Table S2 in the Support-

ing Information) and the protein concentration of the eluates

was determined (Table S3 in the Supporting Information).

In all libraries (Figure 1A and B for Y68X), variable protein

expression levels were seen in whole cell extracts, yet after

robotic protein microscale purification remarkably similar levels

Robotic screening optimisation and validation

To maximise the chance of detecting mutants with improved

activity over wild-type enzyme, several aspects of the robotic

protocols were optimised and the accuracy and reproducibility

tested. Firstly, all screening procedures, excluding culture

growth and centrifugation, were carried out under anaerobic

conditions (<5 ppm oxygen) to avoid significant false positive

enzyme activity detection (NADPH oxidation in the absence of

alkene reduction) due to unproductive flavin reoxidation by

oxygen.^[20]This combined with the relatively long reaction

times (10 min) dramatically decreased the background levels of

activity and enabled slow activities against poor substrates to

be detected (results not shown). It also enabled us to screen

each purified library of mutants with up to three potential sub-

strates, which considerably reduced the costs and overall

screening time.

To investigate the randomness of the mutagenesis, ten

clones from each library were fully sequenced (Table S1 in the

Supporting Information). Some libraries had higher randomisa-

tion rates of amino acid codons than others, and occasional

additional mutations were seen (insertions and deletions),

which were presumably caused by PCR error. The least rando-

mised library was found to be Q241X, which also showed the

highest frequency of premature stop codons. A more extensive

sequencing of libraries H181X and H184X was conducted, with

80 and 64% of each library sequenced, respectively. Only 14 of

the expected 21 outcomes (including wild type and stop

codon) were detected in library H181X, with H181I being the

most frequently observed mutation. In contrast, library H184X

was more highly randomised and showed 20 of the expected

21 outcomes, with fewer additional mutations and a more

even distribution of the mutations (results not shown).

Figure 1. SDS-PAGE analysis of ten randomly selected: A) cell lysates, and

B) robotically purified PETN reductase Y68X mutants. M: protein molecular

mass markers; P: purified PETN reductase standard; 1: Y68V; 2: Y68N; 3:

Y68G; 4: Y68L; 5: Y68K; 6: Y68R; 7: Y68Q; 8: Y68V; 9: Y68H; 10: Y68STOP.

C) Average (grey bars) and maximal (black bars) relative steady state activity

of each library of mutants compared to the average activity of wild-type

PETN reductase (96 clones per library). The error bars denote one standard

deviation of the data. Reactions (0.3 mL) were performed in buffer (50 mm

KH₂PO₄/K₂HPO₄, pH 7.0) containing NADPH (100 mm) and alkene 1a (1 mm

added as a stock in ethanol), and started by the addition of PETNR eluant

(50–150 mL). The reactions were continuously monitored at 340 nm for

10 min at 258C.

A potential problem in comparing mutant activity data from

high-throughput screening methods is the frequent variability

in expression levels of mutants and consequential differences

in enzyme concentrations within the assays. To avoid the ne-

cessity of determining the protein concentration of each prep-

aration, the robotic procedures included microscale protein

purification to normalise eluted enzyme concentration for the

subsequent assays. To test the consistency of the eluted pro-

tein concentrations, both the expression and robotically puri-

ChemBioChem 2010, 11, 2433 – 2447

ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chembiochem.org

2435

N. S. Scrutton et al.

of purified protein were produced (average of (126ꢁ

16) mgmL^ꢀ1across four libraries). Some libraries showed higher

variability in expression levels than others (83–100% expres-

sion of library members sampled), with the mutant clones

Y186X showing such consistently poor PETN reductase expres-

sion (10% clones expressed) that further screening was aban-

doned.

Table 2. Relative screening rates and steady state kinetics of PETN reduc-

tase mutants with ketone-, aldehyde- and acid-containing alkenes.

Substrate

Mutant

Potential hits^[a]Relative

(Z score)

Rate^[c]

]

rate^[b](fold) [s^ꢀ1

wild type

T26S

Y68L

W102Y

H184N

Q241T

Y351F

–

1.0

1.2

1.3

1.0

0.2

1.4

0.1

3.25ꢁ1.05

0.55ꢁ0.11

0.80ꢁ0.07

0.55ꢁ0.09

0.50ꢁ0.12

1.56ꢁ0.63

0.79ꢁ0.02

3.8 (3.0)

2.2 (1.7)

2.5 (1.5)

1.1 (4.2)

1.9

Screening hits

0.8 (3.9)

Two methods were employed for defining potential “hits” in

our screening protocols. Firstly, individual mutant library

member activity screening data were compared to the average

activity of 96 control wild-type clones that had been subjected

to identical robotic procedures as the mutants (Table 2 and

Table S4 in the Supporting Information; column 4). This ena-

bled us to directly compare the reaction rates of individual

mutants to wild-type activity. Due to the limitations in the

number of mutants screened per library, hits were conserva-

tively selected based on whether the activity was greater than

the mean activity plus one standard deviation of the wild-type

library (e.g., Figure 1C). In addition the Z scores of each

mutant were determined; this is a reflection of the variation of

individual mutant activity data against the average rate for the

96 mutants.^[27]Potential hits were defined as having screening

rates with a Z score of at least two (Table 2 and Table S4 in the

Supporting Information; column 3).

wild type

T26A

Y68F

W102F

W102I

Q241V

Q241W

Y351M

Y351P

–

1.0

0.8

1.4

2.7

2.4

0.9

0.6

0.11ꢁ0.01

0.20ꢁ0.01

0.28ꢁ0.00

0.54ꢁ0.01

0.66ꢁ0.03

0.20ꢁ0.01

0.21ꢁ0.04

0.11ꢁ0.01

0.15ꢁ0.01

2.9 (2.5)

4.0 (1.4)

3.2 (0)

3.3 (0)

2.6

2.7

2.6 (1.3)

2.4 (1.3)

wild type

T26S

Y68S

Q241L

Q241V

Y351F

–

1.0

0.9

1.0

1.5

1.3

0.1

1.86ꢁ0.14

1.98ꢁ0.10

0.92ꢁ0.02

2.17ꢁ0.07

2.15ꢁ0.06

0.23ꢁ0.02

2.8 (4.0)

1.9 (2.4)

2.0

1.6

0.6 (5.3)

wild type

T26A

Y68S

W102F

W102Q

Q241L

Q241V

Y351F

–

1.0

0.3

2.3

2.7

2.6

2.2

2.0

0.1

0.43ꢁ0.00

0.14ꢁ0.01

0.07ꢁ0.00

3.91ꢁ0.15

1.25ꢁ0.04

0.54ꢁ0.01

0.27ꢁ0.00

0.12ꢁ0.00

0.9 (3.6)

2.7 (1.1)

2.1 (0.5)

2.0 (0.5)

2.0

As additional controls, the mutant libraries were screened

with the oxidative substrates a,b-unsaturated ketone 1a and

aldehyde 3a^[20]to see whether the mutants had compromised

PETN reductase-like activity. All the mutants tested against

alkene 1a had lower specific activities than the wild type

(Table 2). Most of the hits with substrate 3a had essentially the

same specific activity as wild-type enzyme including T26S and

Q241L. Interestingly, mutation H184N gave PETN reductase the

highly conserved OYE-like active site substrate binding pair

H181/N184, yet showed only 15% activity with alkene 1a

(Table 2); this highlights the importance of the His/His couple

in this enzyme. On the whole, libraries Q241X and Y68X re-

tained relatively good levels of activity compared to wild-type

enzyme, while libraries H181X, H184X and Y351X showed typi-

cally very poor activity. As our assay method detects levels of

NADPH oxidation, it is not clear whether the changes in the

reactions rates with the mutation is at least partly due to

changes in its reactivity with and/or binding to NADPH.

1.6

0.7 (4.7)

wild type

T26A

–

1.0

5.4

0.01ꢁ0.00

0.02ꢁ0.00

1.7 (0.2)

wild type

H184F

–

n.a.

n.d.

0.01ꢁ0.00

2.5 (2.5)

wild type

H181I

H181S

–

2.9

4.2

n.a.

n.d.

0.01ꢁ0.00

[a] Values in parentheses are the average Z score from confirmed wild-

type members within each library. No confirmed wild-type PETN reduc-

tase clones were found in the Q241X library. [b] Reactions (0.3 mL) were

performed in buffer (50 mm KH₂PO₄/K₂HPO₄, pH 7.0) containing NADPH

(100 mm) and alkene (1 mm added as concentrated stocks in ethanol),

and started by the addition of robotically purified PETNR eluant (50–

150 mL). The reactions were followed continuously by monitoring NADPH

oxidation at 340 nm (365 nm for 4a) for 10 min at 258C. Relative rates

are expressed as the ratio of activity of each mutant relative to the mean

activity of 96 wild-type control clones in a separate library. [c] Steady

state kinetics of wild-type and mutant PETN reductases generated by

using conventional purification techniques. Reactions (0.3 mL) were per-

formed as above except for the PETN reductase concentration (50–

2000 nm) and reaction time (2 min); n.a.: not applicable; n.d.: no activity

detected.

The most dramatic increases in activity were seen with the

W102X library against alkenes 2a and 4a. Mutants W102F and

W102I showed five- and sixfold increases in steady state rate,

respectively, against the relatively poor substrate 2-methyl-

cyclopenten-1-one (2a). More moderate increases in activity

(about twofold) were seen with the mutations T26A and

Q241W. This is in comparison to mutagenesis studies of the

OYE morphinone reductase that showed mutant T32A (equiva-

lent to PETN reductase T26A) had a reduction in redox poten-

tial of FMN by ~50 mV, and changes in both the catalytic rate

(k_cat) and binding (K_M) of both the reductive and oxidative sub-

strates.^[25]

2436

www.chembiochem.org

ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ChemBioChem 2010, 11, 2433 – 2447

Site-Saturated Library Screening of PETN Reductase

Mutations W102Q and W102F showed 2.8- to ninefold in-

crease in rate against substrate 4a, while two other mutants

showed essentially the same rate as wild type (Q241L and

Q241V). Interestingly, enzymes W102F/I are equivalent to im-

proved mutants of OYE1 W116F/I, which showed a reversal in

stereochemistry against the substrate (S)-carvone due to the

substrate binding in a “flipped” orientation.^[15]Similarly, a

screen of the equivalent residue in YqjM generated four single

mutants (A104F/W/H/Y) with altered conversion rates and/or

enantioselectivity.^[17]PETN reductase residue W102 has previ-

ously been identified as potentially influencing the binding of

larger substrates and inhibitors.^{[16,21,23,24]}The substitution for

smaller hydrophobic residues, such as phenylalanine and iso-

leucine, might influence the degree of orientation of substrate

binding without significantly altering the localised hydropho-

bic environment.

in the formation of (R)-citronellal from neral compared to the

(S)-enantiomeric product with wild-type enzyme.^[15]In addition,

there was an increase in ee from 19 to 65%, similar to our

results with W102F reduction of 4a. The OYE1 mutant W116I

showed a switch in product enantiopreference in reactions

with (4S)-carvone, with a dramatic increase in ee. Previous

studies with YqjM showed that mutation of the equivalent resi-

due A104 to phenylalanine did not show a switch in product

enantiopreference, but did result in a small increase in conver-

sion and ee.^[16]

The co-crystal structures of wild-type PETN reductase and

mutants W102F and W102Y complexed with picric acid were

determined previously.^[30]A comparison of the structures

showed no major structural changes, however, the conversion

of W102 to phenylalanine removed the hydrogen bond be-

tween W102 NE1 and Q60 NE2 present in the wild-type struc-

ture.^[24]Solution studies indicated that picric acid was bound

more tightly to the active site of the two mutant enzymes,

with no significant impairment of FMN reduction by NADPH.

This was further supported by the co-crystal structures, which

showed both a higher occupancy of bound picric acid and a

shift in position of binding to the mutant enzymes due to the

removal of clashes with W102 atoms CZ3 and CH2.^[30]Taken to-

gether, these results suggest that residue W102 plays a critical

role in the orientation of the substrate in the active site. Muta-

tions at this position could impact on the relative occupancy

of the substrate in the different binding conformations, which

in turn might affect the stereochemistry of the product and

the overall catalytic rate.

Few confirmed hits were obtained when screening against

poor or nonsubstrates of PETN reductase, such as 3-methyl-cy-

clohexen-1-one (5a), and a,b-unsaturated carboxylic acids 6a–

7a yielded few confirmed hits (Table 2). Activity was improved

against alkene 5a to a moderate extent with the mutation

T26A, while weak activity was detected with carboxylic acid-

containing substrates 6a–7a, which were not reduced by wild-

type PETN reductase, with mutants H184F and H181I/H181S,

respectively. Libraries screened against other substrates (3-

methyl-cyclopenten-1-one (8a), cinnamonitrile (9a), cyclohex-

ene carbonitrile (10a), and methyl cinnamate (11 a); Table S4 in

the Supporting Information) yielded no confirmed hits. Some

false positive hits with substrates 8a and 11 a in the Q241X

library included changes to Asp, Val and His (Table S4 in the

Supporting Information). Overall, out of 56 possible hits, there

were six confirmed minor improvements in catalytic activity

(>twofold; five mutations), three of which came from the

W102X library (Table S4 in the Supporting Information).

When PETN reductase library Q241X was screened against

cinnamaldehyde derivatives 3a–4a, a hit was obtained with

mutant Q241L with both substrates (Table 3). Surprisingly, this

mutant showed similar conversions and ee values to the wild

type, but the yield of product 4b was dramatically reduced.

This was due to the reduction of the aldehyde to form signifi-

cant quantities of the equivalent phenylethanol derivatives as

side products. Similarly, reductions of 4a with the mutant

W102Q yielded almost entirely 2-phenyl-propan-1-ol, while

wild-type and other mutant reactions generated only small

quantities of side products (up to 6%). While PETN reductase

enzyme preparations were typically ~95% pure, some minor

contaminants were seen in SDS-PAGE analysis of the Q241L

mutant. The contaminating proteins were analysed by in-gel

trypsin digestion followed by mass spectrometry,^[31]and re-

vealed the presence of an E. coli putative zinc-type alcohol de-

hydrogenase-like protein, YahK, similar to biotransformations

with extracts of YqjM mutants^[16](results not shown). It is pre-

sumed that such contaminating enzymes can be responsible

for the aldehyde to alcohol conversion. The absence or low

quantities of alcohol by-products in reactions with other mu-

tants against aldehyde substrates suggest that protein purifica-

tion had been more successful in removing the contaminating

alcohol dehydrogenase activity.

Biotransformations with confirmed mutants

A number of mutants had been identified as potentially useful

by kinetic screening, however, these assays monitor NADPH

oxidation, and so, are an indirect measure of alkene reduction.

To see what impact these mutations had on alkane product

yields and/or enantiopurity, biotransformation reactions with

the mutants were performed. Wild-type PETN reductase has

been shown previously to reduce a-methylcinnamaldehyde

(4a) with reasonable conversions, but poor ee values,^[28]in

agreement with our results (5–13% ee; 99–100% yield;

Table 3). To our surprise, mutant W102F reduced alkene 4a

with a significant increase in product ee (62% vs. 13%) and a

switch in product enantiopreference to the R enantiomer. In

addition, mutant Y351F showed a moderate increase in enan-

tiopurity of (S)-4b. In all reactions with alkene 4a, the product

ee was significantly higher after 4 h than after 24 h reactions,

probably due to product racemisation in aqueous media as

seen with a,b-disubstituted nitroalkanes.^[20,22,29]

Reactions of PETN reductase mutants T26A and Y68F with

the cyclopentenone substrate 2a showed a reduction in con-

version and yield by about a third compared to the wild-type

enzyme. This compares to reactions with mutant W102F, which

A switch in product enantiopreference caused by a single

amino acid change is remarkable, but not unique. Mutations of

the equivalent residue W116 of OYE1 to phenylalanine resulted

ChemBioChem 2010, 11, 2433 – 2447

ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chembiochem.org

2437

N. S. Scrutton et al.

Table 3. Reduction of activated alkenes by PETNR wild-type and library mutants using NADPH.^[a]

Substrate

Product

Enzyme^[b]

Conv.^[c]

[%]

Yield^[c]

[%]

By-product(s)^[d]

[%]

ee^[d]

[%]

wild type

100 (100)

93 (100)

0

n.a.

T26S

Y68L

W102Y

Q241T

Y351F

100 (100)

100 (96)

100 (82)

100 (90)

100 (88)

100 (85)

0

n.a.

wild type

89

0

69

T26A

Y68F

59

93

72

89

84

94

59

53

82

81

72

89

84

94

0

56

49

72

46

62

66

70

W102I

W102F

Q241V

Q241W

Y351M

Y351P

wild type

93 (72)

0 (<1)

n.a.

T26S

Y68S

Q241L

Y351F

100 (82)

92 (92)

100

100 (69)

91 (68)

57

0 (<1)

43

n.a.

95 (100)

95 (32)

0 (<1)

wild type

100 (99)

2 (<1)

5 (13) (S)

T26A

Y68S

W102F

W102Q

Q241L

Y351F

100 (10)

99 (81)

99 (100)

100

99

99 (99)

98 (0)

90 (77)

94 (97)

1

77

94 (99)

6 (0)

8 (6)

5 (3)

99

22

2 (1)

7 (0) (S)

1 (6) (R)

18 (62) (R)

n.d.

3 (5) (S)

14 (38) (S)

wild type

29

0

98

T26A

Y68L

W102D

W102I

Q241W

Y351E

52

8

0

10

8

52

8

0

10

8

0

66

n.d.

91

n.d.

wild type

9

0

>99

T26L

1

0

1

0

1

0

1

0

n.d.

W102Y

H181D

H181Y

Q241D

Q241V

Y351M

4

11

4

11

n.d.

>99

[a] Conditions: standard reactions (1.0 mL) were performed in buffer (50 mm KH₂PO₄/K₂HPO₄, pH 7.0) containing alkene (5 mm; added as a concentrated

stock in ethanol), NADPH (6 mm) and PETN reductase (2 mm). The reactions were shaken at 378C at 130 rpm for 4 or 24 h. Reactions in parentheses were

performed for 4 h. [b] Wild-type data were determined previously.^[20,22][c] Determined by GC by using DB-Wax column. [d] Determined by GC or HPLC;

n.a.: not applicable; n.d.: not determined.

showed similar product yields as wild type, but a significant

drop in ee. Biotransformations of almost all mutants tested

with the related poor substrate 3-methyl cyclopenten-2-one

(8a) showed essentially no activity compared to an 8% yield

for the wild-type enzyme. The only mutant with significant ac-

tivity was Y351M, which yielded 11% product with >99% ee.

Mutations at equivalent residues in YqjM and/or OYE1 gave

similar types of changes to those observed for PETN reductase

2438

www.chembiochem.org

ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ChemBioChem 2010, 11, 2433 – 2447

Site-Saturated Library Screening of PETN Reductase

mutants. YqjM mutants C26A and A102F (equivalent to T26A

and W102F in PETN reductase) gave slightly lower %ee values

and minor changes in conversion rates compared to wild-type

enzyme with related substrates cyclohex-2-en-1-one-3-methyl

carboxylate and cyclopent-2-en-1-one-3-methyl carboxylate.^[16]

However, the YqjM mutant A102W, which is equivalent to wild-

type PETN reductase, showed no significant change in conver-

sions and %ee to wild-type enzyme.

The most dramatic change was seen in reactions of T26S

with the nitro-olefin substrate (E)-19a. The mutant preferential-

ly catalysed the formation of R enantiomer of product 19b

(37–44% ee) whereas the wild-type enzyme yielded the S

enantiomer (48–37% ee). Moreover, less by-product forma-

tion—namely the corresponding oxime—was observed for this

mutant (Table 4). Interestingly, the switch of enantiopreference

was not observed for the equivalent p-chlorinated substrate

(E)-21a, which was reduced to (S)-21b by both enzymes. PETN

reductase is known to reduce the E isomers of 2-aryl-1-nitro-

propenes with variable enantioselectivity due to a nonenzy-

matic E/Z isomerisation of the substrates,^[26]in addition to

other factors, such as the nature of the cofactor.^[9]

Mutant Y351F reduced substrates 1a and 3a with a dramat-

ic decrease in product yield after 4 h, but not 24 h compared

to wild-type enzyme. These results are consistent with the ob-

served differences in steady state reaction rates with these mu-

tants (Table 3). This is an example of how mutations causing

moderate decreases in steady state rates can still be useful bio-

catalysts as the reduction in rate can be compensated for by

running biotransformations for extended time periods (24 h vs.

2 min). Interestingly, in some cases specific mutations would

only show significant changes in reactions against some of the

substrates tested. For example, mutants W102I and Q241W

showed dramatic decreases in conversion with substrate 5a,

but not 2a. This highlights the importance of screening libra-

ries of mutants against multiple substrates as specific muta-

tions are unlikely to improve the PETN reductase-catalysed

reduction of all target alkenes.

In addition, the poor ee values are thought to be due to the

possibility of multiple binding conformations of the E isomers

of the substrates in PETN reductase (Scheme 2). The removal

of the CG2 methyl group of T26 appears to have altered the

shape of the substrate binding site to enable an increase in

the binding of the substrate in the “flipped” binding

mode,^[22,23]thereby, improving the yield of the (R)-19b product.

Interestingly, the reaction of T26S with the p-chlorinated sub-

strate (E)-21a resulted in a significant increase in ee compared

to wild type; this also suggests a change in the relative active

site occupancy of the substrate in the different binding con-

formations.

Previous kinetic studies of the wild-type enzyme showed

that the reduction of (Z)-19a is rapid and quantitative, and

yields almost enantiopure products.^[26]This contrasts with the

reduction of (E)-19a, which is considerably slower (~30-fold)

with poorer product enantiopurities. These results suggest that

(Z)-19a binds more optimally compared to (E)-19a, with the

possibility of multiple binding conformations with (E)-19a. A

model of the preferred substrate (Z)-1a bound to wild-type

PETN reductase shows both the nitro group and the phenyl

ring pointing down towards the ribityl chain of FMN, to yield

(S)-19b (Scheme 2A). In this position, the nitro group interacts

with H181 and/or H184. In contrast, (E)-19a would need to

bind in the “flipped” binding conformation to produce the (S)-

19b product. To produce (R)-19b, the substrate (E)-19a would

need to bind in a more conventional conformation, as seen for

(Z)-19a binding, with the nitro group pointing downwards

(Scheme 2B). Such binding conformations could reduce any

potential clash between the substrate methyl substituent and

T26 CG2 atom. This could explain why (S)-19b is the preferred

enantiomeric product of (E)-19a, as the binding mode to pro-

duce (R)-19b is likely to cause a clash between the substrate

phenyl ring and T26 CG2 atom (Scheme 2B).

Biotransformations of PETN reductase T26S mutant

Although the mutant T26S did not show significant improve-

ments in the screening reactions, it was selected for further

investigation as previous studies with the wild-type enzyme

suggested the loss of the side chain methyl group (CG2 of

T26) might potentially remove clashes with some sub-

strates.^{[15,20,21,23]}In addition, the mutant is likely to retain both

hydrogen bonds between residue 26 and the FMN seen in the

wild-type structures^[23]due to the retention of the OG atom in

S26. Therefore, reactions with wild type and T26S mutant were

carried out with 2- or 3-methyl substituted cyclopenten-1-one

and cyclohexen-1-one derivatives 2a, 5a, 8a and 12a, alde-

hydes 13a–15a, (R)-carvone ((5R)-16a), ketoisophorone (17a),

N-phenylmethyl maleimide (18a), and nitro-olefins (E)-19a to

(E)-22a over 24 h by using NADPH and/or a NADP⁺/G6PDH co-

factor regenerating system (Table 4). In spite of a near sixfold

reduction in steady state rate of T26S with alkene 1a com-

pared to wild-type enzyme (Table 2), this did not significantly

effect the yield and enantioselectivity when biotransformations

were performed for 24 h.

The reactions with the cyclic ketone substrates 12a and

17a–18a with the wild type and T26S generated the R prod-

ucts with high yields and enantiopurities, while reactions with

substrates 5a–6a and 15a yielded only low levels of highly

enantiopure products. In contrast, the reaction of T26S with 2-

methylcyclopent-2-en-1-one (2a) showed a considerable de-

crease in conversion and yield compared to wild-type enzyme,

although similar product enantiopurities were seen. Both the

wild-type enzyme and T26S mutant generated high yields of

the (rac)-20b product, which should be attributed to product

racemisation rather than a lack of enzyme stereoselectivity.^[20]

Nitroalkene (E)-19a has been shown to undergo a slow, re-

versible isomerisation to (Z)-19a under the reaction conditions

(Scheme 2C).^[20]This suggests that the moderate enantiopuri-

ties of alkane 19b obtained during the reduction of (E)-19a

with both wild-type and T26S PETN reductase might be at

least partially due to a form of dynamic kinetic resolution.^[29]In

this case, the binding of (E)-19a in a (Z)-19a-like manner

would lead to a slow reduction (k_i) to (R)-19b, while (Z)-19a

formed by isomerisation is likely to bind to PETN reductase in

the conventional manner to rapidly form (S)-19b (~30 k_i;

ChemBioChem 2010, 11, 2433 – 2447

ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chembiochem.org

2439

N. S. Scrutton et al.

Table 4. Reduction of various activated alkenes by PETNR wild-type and T26S mutant by using NADP⁺/G6PDH cofactor regeneration system or NADPH.^[a]

Substrate

Product

Enzyme^[b]

t [h]

Conv.^[c][%]

Yield^[c][%]

ee^[d][%]

wt-NADPH

48

>99

99

57

wild type

T26S

24

62

23

62

16

70

69

wt-NADPH

48

39

37

>99

wild type

T26S

24

25

28

18

13

>99

wt-NADPH

48

23

22

>99

wild type

T26S

24

14

9

14

7

>99

wt-NADPH

48

>99

95

wild type

T26S

24

>99

92

93

80

97

96

wt-NADPH

48

>99

78

n.a.

wild type

T26S

24

>99

85

84

n.a.

wt-NADPH

48

>99

78

27

wild type

T26S

24

>99

85

84

66

70

wt-NADPH

48

>99

18

>95

wild type

T26S

24

>99

23

33

>95

wt-NADPH

24

>99

88

93 de

wild type

T26S

24

>99

76

95

94 de

wild type

T26S

1.5

>99

80

89

95

1.5

wild type

48

>99

T26S

48

>99

wild type

48 (48)

90 (>99)

72^[e](88)

48 (37)

T26S

48 (48)

48

>99 (>99)

>99

37^[e](>95)

37 (44) (R)

wild type

T26S

93

92

0

48

>99

wild type

T26S

48 (48)

78 (77)

65 (58)

63^[e](50)

55^[e](44)

67 (55)

82 (87)

2440

www.chembiochem.org

ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ChemBioChem 2010, 11, 2433 – 2447

Site-Saturated Library Screening of PETN Reductase

Table 4. (Continued)

Substrate

Product

Enzyme^[b]

t [h]

Conv.^[c][%]

Yield^[c][%]

ee^[d][%]

wild type

T26S^[f]

7 d

64

60

40

38

64

60

wild type

48

92

88

20

T26S

48

7 d

84

>99

86

84

97

86

13

73

60

wild type^[f]

T26S^[f]

[a] Conditions: reactions (1 mL) were performed in buffer (50 mm KH₂PO₄/K₂HPO4, pH 7.0), alkene (5 mm; added as a DMF solution with 2% (v/v) final con-

centration), PETNR (2 mm), glucose-6-phosphate dehydrogenase (8 U), glucose-6-phosphate (20 mm) and NADP⁺(6 mm). Reactions were agitated at 308C

at 130 rpm. Values in parentheses were repeated reactions. [b] Wild-type data were determined previously.^{[20, 22]}[c] Determined by GC by using DB-Wax

column. [d] Determined by GC or HPLC. [e] Oxime by-product detected. [f] Biphasic reactions: reactions were carried out anaerobically in 30 mL screw top

vials sealed with PTFE-silicon septa. Each reaction mixture (12 mL) contained buffer (50 mm KH₂PO₄/K₂HPO₄, pH 7.0) containing nitroalkene (1.7 mm) dis-

solved in 4.8 mL anaerobic iso-octane, PETN reductase (~3 mm), NADP (20 mm) glucose-6-phosphate (14–20 mm), glucose-6-phosphate dehydrogenase

(10 U) and sec-butyl benzene (25 mL). Reactions were agitated at 200 rpm for 1–7 days at 308C. [g] Absolute configuration uncertain; n.a.: not applicable;

d: days. wt-NADPH: wild-type PETN reductase reactions performed in the presence of NADPH.

Scheme 2. Predicted binding modes of the PETN reductase-catalysed reduction of the nitro-olefin substrate (Z)- and (E)-19a to form: A) (S)-19b, and B) (R)-

19b. C) Model of the proposed dynamic kinetic resolution of (E)-19a by a combination of (E)- to (Z)-19a isomerisation combined with both forms of the sub-

strate binding to the enzyme in the conventional manner.

Scheme 2C). The switch in product stereochemistry by the

T26S mutant is suggestive of an increase in the relative popu-

lation of the E substrate in the conventional binding conforma-

tion, which is likely due to the removal of a potential clash be-

tween the substrate phenyl ring and the T26 CG2 atom. Sever-

al studies^[9,20]have demonstrated variability in enantiopurity

and stereochemistry of OYE-catalysed reductions depending

on factors such as the cofactor used and reaction time, which

could impact on substrate binding, enzyme kinetics and sub-

strate isomerisation. Therefore, in the absence of a co-crystal

structure of (E)- or (Z)-19a-bound PETN reductase, the exact

cause(s) of the moderate product ee values are not known.

In addition to the changes seen in product ee and/or enan-

tioselectivity, reactions with (E)-19a and (E)-21a with both the

wild-type and T26S enzyme gave lower than expected product

yields. This was due to the formation of significant quantities

of the equivalent oxime, aldehyde, alcohol and other unidenti-

fied by-products, as seen previously with wild-type PETN re-

ductase (Scheme 3).^[26]Oxime formation from aliphatic and aro-

matic nitro compounds has also been seen with other OYEs.^[32]

The mechanism of OYE-catalysed oxime formation is not clear;

however, others have suggested that it proceeds after alkane

product formation by the reduction of the nitro group to yield

the nitroso derivative, which spontaneously rearranges to form

ChemBioChem 2010, 11, 2433 – 2447

ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chembiochem.org

2441

N. S. Scrutton et al.

Scheme 3. Products obtained from the reactions of nitro-olefins (E)-19a and (E)-21a with wild-type and T26S mutant PETN reductase. Further unidentified

minor side products were also produced. The formation of the alcohol product might be due to the reduction of the equivalent aldehyde, possibly by the

presence of a minor contaminating keto-reductase.

the oxime.^[32]However, our studies suggest that PETN reduc-

tase-catalysed oxime formation proceeds by nitro reduction of

the alkene substrate rather than the equivalent alkane product

(results not shown). Further discussion and possible mecha-

nisms of oxime formation will be presented elsewhere.

bone and side chain interactions,^[14,16,34]is thought to increase

the enzyme redox potential by stabilising the negative charge

of the reduced flavin.^[34]The loss of the hydrogen bond be-

tween the OG1 atom of the equivalent residue T37A of OYE1

(SG atom of C26 in YqjM) decreases the redox potential of the

enzyme by 33 mV; this results in both a decrease and an in-

crease in the reductive and oxidative half reactions, respective-

ly. It is apparent that residue T26 is a key target in fine-tuning

the enantiopreference of these enzymes against a number of

substrates. Interestingly, the YqjM mutants C26T or C26S

(equivalent to wild-type and T26S of PETN reductase, respec-

tively) were not apparent “hits” with this substrate.^[16]

The general trend seen with reactions containing NADPH

versus the NADP⁺/G6PDH recycling system is an increase in %

conversion (Table 4) due to the longer time scale of the reac-

tions (48 vs. 24 h). However, the wild-type enzyme gave lower

yields of products with aldehyde substrates 14a and 15a in

the presence of NADPH due to an increase in the levels of side

products. In addition, the enantiopurity of 14b was dramatical-

ly reduced in the longer reactions presumably due to water-

mediated racemisation of the product.^[9,20,33]This was also seen

in the reactions with substrate 2a, which has previously been

shown to undergo a time-dependent loss of product enantio-

purity under aqueous conditions.^[20]

Structure of PETN reductase mutant T26S

The X-ray crystal structure of the wild-type His₈-tagged PETN

reductase was determined to 1.4 ꢁ to see if the presence of

the non-native C-terminal tag had any impact on the overall

structure of the enzyme. In addition, the crystal structure of

the T26S mutant (1.6 ꢁ) was determined to investigate the

cause of the switch in enantiopurity in reactions with substrate

(E)-1a. The data collection and refinement statistics can be

found in Table S6 in the Supporting Information. Both struc-

tures were virtually identical to previously described monomer-

ic wild-type PETN reductase structures (e.g., PDB ID: 1H50), in-

cluding the presence of an acetate molecule bound in a similar

conformation in the active site (Figure 2A).^[21]The structures

Recent site-saturated mutagenic screening of YqjM at the

equivalent residue C26 by others^[16]against the non-wild-type

substrate 3-ethylcyclohex-2-en-1-one yielded six mutants capa-

ble of reducing this compound. Mutants C26V/G/D were capa-

ble of forming the S enantiomer of product 3-methyl-2-cyclo-

hexan-1-one with moderate enantiopurities. This is similar to

reductions of the poor substrate 3-methyl-2-cyclohexen-1-one

(5a) with PETN reductase, which yielded the S enantiomeric

product with low yields, but high enantiopurities. This residue

in OYEs, which is known to interact with the FMN via back-

Figure 2. A) Stereo diagram of the X-ray crystal structure of the active site of PETNR mutant T26S superimposed with wild-type enzyme. The T26S structure

residues, FMN cofactor and bound acetate ion are shown as atom coloured sticks with green, yellow and magenta carbons, respectively. Residues L25 to R27

are also shown as a green ribbon. The omit jF_ojꢀjF_cj map of the T26S side chain is contoured at 1s (green mesh). B) X-ray crystal structure of the ribityl

chain of the FMN of PETNR–His₈superimposed with wild-type enzyme. The enzyme residues and FMN are shown as atom coloured sticks with green and

yellow carbons, respectively. The omit jF_ojꢀjF_cj map of the FMN molecule is contoured at 1s (green mesh). The wild-type structure was obtained from the

PDB (ID: 1H50).^[21]The wild-type residues are shown as atom coloured lines with grey carbons while waters and interactions are shown as red spheres and

black dotted lines, respectively. Residue loops T239–D244 (left) and T273–K279 (right) are shown as green cartoons and grey ribbons, respectively. All figures

were generated in PyMOL.^[35]

2442

www.chembiochem.org

ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ChemBioChem 2010, 11, 2433 – 2447

Site-Saturated Library Screening of PETN Reductase

show minor changes in backbone positions of two surface

loops near the substrate binding site (T239–D244; T273–K279;

Figure 2A) as well as a shift in the orientation and position of

the backbone atoms of residue D360. Both structures show a

minor “butterfly-bend” in the isoalloxazine ring of the FMN

prosthetic group, as seen in the “thermostable-like” OYE

member TOYE.^[14]

the subtle changes in the active site of PETN reductase caused

by the T26S mutation (and W102F) do not effect the binding

of all substrates; this highlights the importance of screening

multiple libraries of mutants to find suitable biocatalysts with

improved functionality against selected substrates.

Conclusions

A significant change in the active site of both structures was

the orientation of the ribityl chain of FMN starting from atom

C3* to the terminal phosphate atoms (Figure 2B and Figure S4

in the Supporting Information). This is similar to the second

position of the ribityl chain in the reduced progesterone-

bound PETN reductase crystal structure.^[23]In spite of these

changes, the only difference in the interactions between the

ribityl chain and the protein/solvent is the presence of an addi-

tional interaction between FMN O1P and R324 N. Residues

R324 and A302 are slightly altered in position to enable the

wild-type interactions to be maintained in the His₈structure.

Further discussion on the minor differences between the wild-

type and His₈-tagged PETN reductase structures can be found

in the Supporting Information.

Our screening procedures were more successful in detecting

PETN reductase mutants with significantly improved steady

state reactions rather than improvements in enantiopreference

and/or ee against selected substrates (i.e., “you get what you

screen for”). Not all libraries generated improved mutants, sug-

gesting these amino acid positions are not important (at least

in isolation) for influencing the reaction rate with these sub-

strates. As expected, our selection criteria were more accurate

in finding “hits” with substrates exhibiting large improvement

in catalytic rate. To complicate matters, the determination of a

true “hit” for improved reaction of poor substrates was prob-

lematic because only small changes in an already low activity

were detected. Thus, the robotic screen should be used as a

first stage in a two-stage screening process and complement-

ed by larger scale tests to determine product yield and enan-

tiopurity. Setting the threshold of criteria higher for determin-

ing a “hit” will reduce the overall number of mutants needed

to be further characterised, and increase our chance of finding

useful mutants.

The structure of the active site of PETN reductase T26S

mutant revealed only minor changes in the position of resi-

dues surrounding the FMN compared to wild type (Figure 2C).

The most significant change, albeit minor, was seen in the po-

sition of residue Y351, which is known to be involved in sub-

strate/inhibitor binding in OYEs.^[22]The omit map jF₀jꢀjF_cj of

residue 26 showed a clear absence of density for the CG2

atom of T26, confirming the presence of the T26S mutation.

The mutant retained both its wild-type-like hydrogen bonding

interactions between side chain and backbone atoms of S26

and the FMN (S26 OG–FMN O4 and S26 N–FMN N5). Therefore,

changes in the activity and stereoselectivity of this mutant are

due to more subtle structural changes within the active site.

A sequence alignment of OYE family members shows a high

sequence conservation of residue 27 as either threonine or

cysteine within the classical and thermophilic-like subclasses,

respectively.^[14,36]The extensive structural information available

for many OYEs shows a highly conserved interaction between

the threonine or serine hydroxyl group and the FMN isoalloxa-

zine O4 atom.^[22,30]This has been shown to influence the FMN

redox potential by stabilising the negative charge of the re-

duced flavin.^[25,37]In addition, the structure of the picric acid-

bound wild-type PETN reductase shows that the T26 OG1

atom interacts with the O61 atom (nitro group) of the ligand.

Previous modelling studies of PETN reductase with nitro-olefin

substrates (E)-19a to (E)-22a suggested that multiple binding

conformations could be possible, with one position likely to be

less favoured due to potential clashes between the substrate

Cb atom and T26 side chain.^[26]Therefore, the change in prod-

uct enantiopreference of product 19b by the T26S mutant is

likely to be due to an increase in binding of the substrate in

an alternate conformation made possible by the removal of

the Cb atom of T26. No switch in enantiopreference was seen

with the related substrate (E)-21a, which contains a p-chloro

substitutent on the aromatic ring; this might restrict the bind-

ing of the substrate in the alternate conformation. Therefore,

The advantages of using a steady-state reaction based

screening method over biotransformation-based yield and ee

monitoring include considerably shorter reaction times (10 min

vs. 24 h), no downstream product extractions, and the screen-

ing of up to three substrates per library extraction. In addition,

the purification step not only normalises the enzyme concen-

tration in the reactions, but also dramatically reduces levels of

contaminating E. coli enzymes that either compete for the sub-

strate or catalyse side reactions on the products.^[16]The use of

a novel screening facility housed within an anaerobic chamber

was essential to eliminate the significant reoxidation rate of

NADPH by molecular O₂, especially when poorly reduced oxi-

dising substrates were being tested.

This method enabled us to rapidly screen large numbers of

mutants and eliminate significant numbers of library members

with compromised activity. This allowed us to dramatically

decrease the number of clones required to undergo the more

laborious biotransformation reactions. Thus, a hybrid approach

has been used incorporating both high-throughput screening

by steady state catalytic rates followed by biotransformation

reactions with positive screening hits.

Screening methods employing only the detection of reac-

tion rates do not enable us to screen for improvements in ee

or the maintenance of existing high ee. In spite of this, our

screen enabled us to detect two mutants (T26S and W102F)

with both a switch in product enantiopreference and an in-

crease in ee. In addition, the detection of mutants with weak

activity against a,b-unsaturated carboxylic acid substrates

shows progress in the expansion of substrate scope in PETN

reductase. These mutants have given us insight into the impor-

ChemBioChem 2010, 11, 2433 – 2447

ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chembiochem.org

2443

N. S. Scrutton et al.

Generation of site-saturated single position libraries of PETN re-

ductase-His₈: Amino acid randomisation was performed by using

protocols based on the CASTing method of mutagenesis.^[38]Single

site-saturated libraries were generated at selected amino acid posi-

tions (T26, Y68, W102, H181, H184, Y186, Q241 and Y351) by PCR

with NNK degenerate oligonucleotides (T26X: GTT TAT GGC CCC

ACT TNN KCG TCT GCG CAG CAT CG, CGA TGC TGC GCA GAC GMN

NAA GTG GGG CCA TAA AC; Y68X: GCT CAG GCA AAA GGC NNK

GCC GGT GCA CCG GG, CCC GGT GCA CCG GCM NNG CCT TTT

GCC TGA GC; W102X: CGT ATT GCG GTT CAG CTG NNK CAC ACC

GGT CGT ATC, GAT ACG ACC GGT GTG MNN CAG CTG AAC CGC

AAT ACG; H181X: CGA CCT GGT TGA GCT TNN KTC TGC GCA CGG

TTA CC, GGT AAC CGT GCG CAG AMN NAA GCT CAA CCA GGT CG;

H184X: GTT GAG CTT CAC TCT GCG NNK GGT TAC CTG CTG CAT

CAG, CTG ATG CAG CAG GTA ACC MNN CGC AGA GTG AAG CTC

AAC; Y186X: CAC TCT GCG CAC GGT NNK CTG CTG CAT CAG TTC

CTG, CAG GAA CTG ATG CAG CAG MNN ACC GTG CGC AGA GTG;

Q241W: CCC CGA TCG GTA CTT TCN NKA ACG TCG ACA ACG GTC,

GAC CGT TGT CGA CGT TMN NGA AAG TAC CGA TCG GGG; Y351X:

GTC CTG AAA GCT TCN NKG GCG GCG GCG CGG, CCG CGC CGC

CGC CMN NGA AGC TTT CAG GAC). Following template removal

by selective restriction digestion (DpnI), PCR products (50 ng) were

transformed into the E. coli strain JM109 (Promega) according to

the manufacturer’s protocol. Each library of transformants were

grown on three plates of LB agar containing ampicillin

(100 mgmL^ꢀ1) for 24 h at 378C to a density of 150–300 colonies per

plate. To generate a non-mutation control library, the wild-type

enzyme–His₈construct was transformed into the same E. coli strain

and processed as above.

tance of these residues in influencing the relative populations

of substrate binding conformations. Recent work by others^[16]

has showed that more dramatic improvements in biotransfor-

mation yields and/or enantioselectivities of mutants tend to be

obtained after at least two rounds of iterative saturation muta-

genesis; this suggests the importance of cooperative effects of

mutants. Given the low numbers of useful mutants detected in

our single-site saturation libraries, mutants generated from

multiple site-saturated mutagenesis techniques could be rapid-

ly screened in our facility to potentially improve the activity

and enantioselectivity of PETN reductase against poorly re-

duced substrates.

Experimental Section

General: All reagents were of analytical grade. All kinetic measure-

ments were performed within an anaerobic glove box (Belle Tech-

nology, Ltd.) under a nitrogen atmosphere (<5 ppm oxygen). Puri-

fied PETN reductase and substrate extinction coefficients used

were as described previously.^[20,22,29]All medium components were

obtained from Formedium. Full gene sequences of all mutants

were confirmed by DNA sequencing (Eurofins MWG Operon).

His₈-tag incorporation: A C-terminal His₈-tag was incorporated

into an existing high expression PETN reductase construct (pONR1;

pBluescript SK⁺)^[17]by PCR with the overlapping complementary

tails method (Figure S1A in the Supporting Information) based on

the QuikChange whole plasmid synthesis protocol (Stratagene).

Both the forward and reverse primers contained an 18-base pair

(bp) His₆5’ overhang followed by the native sequence beginning

with the stop codon and last PETN reductase residue codon, re-

spectively (CAT CAT CAC CAT CAC CAT TAA TCC CGC TTT GTA CAT

TG and ATG GTG ATG GTG ATG ATG CAG TGA AGG GTA GTC GG).

The 5’ overhang regions were the only complementary sequences

within these primers, with incomplete primer overlap resulting in

the production of C-terminal His₈tags. Constructs were trans-

formed into the E. coli strain XL-10 Gold (Stratagene) according to

the manufacturer’s protocol and incubated on LB medium agar

containing ampicillin (100 mgmL^ꢀ1) for 24 h at 378C. Insertion

clones were identified by colony pick PCR^[19]in which a 302–

326 bp C-terminal PCR product was amplified corresponding to

the absence and presence of a His₈-tag (CGG GTG CGT ATA CGG

CAG AAA AAG C and GCG CAT AAT TTC CTC ATT AAC CAG TCG

AAT G for the forward and reverse primers, respectively). Gel elec-

trophoresis in agarose (1.3%) in TAE buffer (40 mm Tris, pH 8.5,

containing 1.14 mL glacial acetic acid, 4 mm EDTA) for 1–1.5 h at

100 V was sufficient to visually distinguish between the two small

PCR products (Figure S1B in the Supporting Information). Positive

clones were fully sequenced to confirm the presence of the His₈-

tag. Sequencing revealed the presence of a silent mutation A921G

(DNA sequence) in the original clone, which was propagated

throughout the mutants.

Anaerobic robotic library screening protocols: All robotic screen-

ing procedures were carried out by using a custom-designed Mi-

crolab Star platform (Hamilton Robotics; Figure S5 in the Support-

ing Information) located within an anaerobic glove box under a ni-

trogen atmosphere (<5 ppm oxygen; Belle Technology) to avoid

false positive enzyme activity detection due to unproductive flavin

reoxidation by oxygen.

1) Colony picking: Automated colony picking was used to pick 96

colonies per agar plate. These were used to inoculate terrific broth

medium (TB; 96ꢂ1 mL) containing ampicillin (200 mgmL^ꢀ1) in

2.2 mL 96-deep-well blocks. The cultures were sealed with a sterile

gas-permeable membrane (StarLab) and incubated aerobically for

16 h at 378C in a plate incubator (Titramax 1000, Heidolph) at

1050 rpm. Glycerol stocks of each library were produced by com-

bining the culture (50 mL) with a cryoprotectant buffer (80%, v/v)

phosphate buffered saline^[39]containing glycerol (20%, v/v; 50 mL)

within 96-well PCR plates and stored at ꢀ808C.

2) PETN reductase library culture generation: Each library of clones

was grown in TB autoinduction medium (TBAIM; 1.0 mL) contain-

ing ampicillin (200 mgmL^ꢀ1) in 96-deep-well blocks as before for

24 h, by using the glycerol stocks as the inoculums. The cultures

were harvested by centrifugation for 20 min at 1109g in an Allegra

X-22R bench top centrifuge (Beckman Coulter) after replacement

of the gas-permeable seal for an aluminium foil lid (Beckman

Coulter). The pellets were retained within the 96-well block and

stored at ꢀ208C.

Site-directed mutagenesis: Prior to library generation, the mutant

T26S (no His₈-tag) was generated by site-directed mutagenesis

based on the QuikChange whole plasmid synthesis protocol. PCR

reactions were performed by using native pONR1^[17]as the tem-

plate and the oligonucleotides GGC CCC ACT TAG CCG TCT GCG

CA and TGC GCA GAC GGC TAA GTG GGG CC. Constructs were

transformed into the E. coli strain JM109 (Promega) according to

the manufacturer’s protocol and incubated on LB agar containing

ampicillin (100 mgmL^ꢀ1) for 24 h at 378C. Positive clones were fully

sequenced to confirm the presence of the T26S mutation.

3) Robotic protein purification: High-throughput PETN reductase

protein extraction and microscale purification was achieved roboti-

cally in a 96-well format, by using Ni-magnetic agarose beads (No-

vagen) as an affinity resin specific for the His-tag. Cell pellets were

completely lysed by the addition of buffer A (50 mm KH₂PO₄/

K₂HPO₄, pH 7.9, 200 mgmL^ꢀ1lysozyme, 15 UmL^ꢀ1benzonase (Nova-

2444

www.chembiochem.org

ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ChemBioChem 2010, 11, 2433 – 2447

Site-Saturated Library Screening of PETN Reductase

gen); 200 mL), followed by ten rapid mixing cycles. The lysates

were agitated for 5 min followed by the addition of buffer B (1.5m

NaCl, 0.2m imidazole, pH 7.9, 1 mm FMN; 50 mL) to reduce subse-

quent unspecific protein binding to the affinity resin and to ensure

maximal flavination of PETN reductase. A slurry of Ni-magnetic

agarose beads (1% beads in 50 mm KH₂PO₄/K₂HPO₄, pH 7.9, 0.3m

NaCl, 5 mm imidazole and 40% (w/v) sorbitol; 50 mL) was added to

the lysates followed by agitation for 3 min to maximise PETN re-

ductase binding. Sorbitol was added to the magnetic beads resus-

pension buffer to increase the viscosity and ensure more even dis-

pension of beads across the plate. Automated lysate removal (and

subsequent wash/elution buffer removal) was achieved after affini-

ty resin immobilisation by using a 96-well magnet (Qiagen). The

resin was washed three times by resuspension in buffer C (50 mm

KH₂PO₄/K₂HPO₄, pH 7.9, 0.3m NaCl, 250 mm imidazole; 700 mL then

2ꢂ200 mL). PETN reductase was eluted in buffer D (50 mm KH₂PO₄/

K₂HPO₄, pH 7.9, 0.3m NaCl, 250 mm imidazole; 120 mL) into 96-well

microtiter plates (Greiner) containing buffer E (50 mm KH₂PO₄/

K₂HPO₄, pH 7.0; 180 mL).

crystals were flash frozen in liquid nitrogen in the absence of addi-

tional cryoprotectant. A full PETN reductase–His₈(1.4 ꢁ) X-ray dif-

fraction data set were collected from a single crystal at the Europe-

an Synchrotron Radiation Facility (Grenoble, France) on Station ID

14.4 (wavelength 1.07 ꢁ; 100 K) by using an ADSC CCD detector. A

full T26S PETN reductase (1.5 ꢁ) X-ray diffraction data set was col-

lected in-house from a single crystal by using a Rigaku Micro-

max007 generator (wavelength 0.97 ꢁ; 100 K) with an R-Axis IV⁺⁺

detector.

Structure determination and refinement: Both data sets were

processed and scaled by using the programs MOSFLM^[40]and

Scala.^[41]The structures were solved by molecular replacement by

using the coordinates for the acetate-bound PETN reductase struc-

ture (PDB ID: 1H50).^[21]Model rebuilding and water addition was

performed automatically by using REFMAC combined with ARP/

warp.^[42]Positional and anisotropic B-factor refinement was per-

formed by using REFMAC5,^[41](hydrogen atoms included in the re-

finement) with alternate rounds of manual rebuilding of the model

in COOT.^[43]The final models were refined to 1.40, 1.50 and 1.6 ꢁ

resolution giving final R_factor/R_freeof 12.6/15.6 and 14.7/17.0 for

PETN reductase–His₈and T26S mutant, respectively. The atomic co-

ordinates and structure factors for PETN reductase-His8 (ID: 3P62)

and for the T26S mutant (ID: 3P67) have been deposited in the

Protein Data Bank, Research Collaboratory for Structural Bioinfor-

matics, Rutgers University, New Brunswick, NJ (http://

www.rcsb.org/)

4) Robotic library activity screening: Automated anaerobic steady-

state kinetics assays of library eluants were performed in 96-well

UV transparent plates (Greiner) by monitoring the consumption of

NADPH in the presence of an oxidising alkene substrate.^[20]Reac-

tions (0.3 mL) were performed in buffer (50 mm KH₂PO₄/K₂HPO₄,

pH 7.0) containing NADPH (100 mm) and alkene (1 mm added as

concentrated stocks in ethanol), and started by the addition of

PETN reductase eluant (50–150 mL). The reactions were continuous-

ly monitored at 340 nm for 10 min at 258C in a Biotek Synergy HT

microtiter plate reader. Rates of each library member with each

substrate were compared to the average activity from 96 wild-type

clones. In addition, potential “hits” were identified by calculating

the relative Z scores for each mutant [Eq. (1)]:^[26]

Chemistry: NMR spectra were recorded on 400 or 500 MHz spec-

trometers and referenced to the solvent, unless stated otherwise.

The chemical shifts are reported in ppm and coupling constants (J)

are given in hertz (Hz). Melting points were determined by using

electrothermal capillary apparatus and are uncorrected. IR spectra

were recorded neat on NaCl plates. HPLC analysis was performed

by using an instrument equipped with a UV detector. UV/Vis data

were recorded with a diode array spectrophotometer. The reaction

progress was monitored by GC or TLC on standard silica gel plates.

All chemicals were obtained from commercial sources and the sol-

vents were of analytical grade. Substrates 1a–15a, (5R)-16a and

17a and dihydro products 1–9b, 12b–15b and 17b were ob-

tained from commercial suppliers (Aldrich and Alfa Aesar). (+)-Di-

hydrocarvone was obtained from a commercial supplier as a mix-

ture of two isomers (2R,5R)-16b and (2R,5S)-16b in a 77:20 ratio

(>99% ee). Nitroalkenes (E)-19a and (E)-21a were synthesised as

described previously.^[22,29]Racemic nitroalkane 21b were obtained

in 80–85% yields by silica gel-assisted reduction of their respective

nitroalkenes with NaBH₄, as described in the literature.^[44]Model

optically active nitroalkane (R)-21b was synthesized as described

before.^[45]The substrates 2a, 5a, 18a and products 2b, 5b,

(2R,5R)-16b and (R)-17b–(R)-18b, were synthesised as described

previously.^[20,22,29]The nitroalkenes (E)-19a to (E)-22a and their re-

spective alkene products (R/S)-19b to (R/S)-22b were synthesised

as described previously.^[20,45]The remaining compounds were syn-

thesized as follows:

Z_i¼ ðx_iꢀXÞ=S_x

ð1Þ

where X is the mean of all the sample values x_i(no controls) of the

library, and S is the standard deviation of these values.

Medium-scale protein production and purification: PETN reduc-

tase–His₈wild-type and mutant cultures (1–6 L) were grown in

TBAIM containing ampicillin (100 mgmL^ꢀ1) for 24 h at 378C. Cells

were harvested by centrifugation at 6000g for 30 min at 58C

(Avanti J-26 XP; Beckman Coulter). Cell pellets were resuspended

in lysis buffer (50 mm KH₂PO₄/K₂HPO₄, pH 8.0, containing the EDTA-

free complete protease inhibitor cocktail (Roche), 0.1 mgmL^ꢀ1

DNase I, 1 mgmL^ꢀ1lysozyme and excess free FMN) and stirred for

30 min at 48C. Cells were disrupted by sonication (Sonics Vibra

Cell) followed by extract clarification by centrifugation for 20 min

at 26600g.

PETN reductase–His₈was purified by binding to a 50 mL Ni-NTA

column (Qiagen), pre-equilibrated in buffer A (50 mm KH₂PO₄/

K₂HPO₄, pH 8.0, 0.3m NaCl, 20 mm imidazole). The column was

washed with 5ꢂcolumn volumes of buffer B (50 mm KH₂PO₄/

K₂HPO₄, pH 8.0, 0.3m NaCl, 40 mm imidazole) followed by elution

in a step to buffer C (50 mm KH₂PO₄/K₂HPO₄, pH 8.0, 0.3m NaCl,

250 mm imidazole). Purity was assessed by SDS-PAGE and the con-

centration determined by using the extinction coefficient

method.^[20,22]

1) 2-Methyl-3-phenyl-propan-1-ol (rac) (4c): This was synthesised

from a mixture of (E)-2-methyl-3-phenyl-prop-2-en-1-ol (10.0 mmol)

and Pd-C (10%, 0.10 g) in ethyl acetate (150 mL) stirred under H₂

(g) for 24 h. The catalyst was removed by filtration through a pad

of Celite and the solvent was removed in vacuo. The crude product

was purified by flash column chromatography (hexane/ethyl ace-

tate, 9:1, 4:1) to give 2-methyl-3-phenylpropanol (1.43 g, 95.3%).

¹H NMR (400 MHz; CDCl₃): d=0.92 (d, J=6.8 Hz, 3H; CH₃). 1.28

(brs, 1H; OH), 1.90–2.00 (m, C1H; H), 2.43 (dd, J=13.4, 8.0 Hz, 1H;

CH_AH_B), 2.76 (dd, J=13.4, 6.3 Hz, 1H; CH_AH_B), 3.48 (dd, J=10.5,

Crystallogenesis and data collection: Crystals of oxidised wild-

type PETN reductase–His₈and the T26S (no tag) were grown by

using the sitting-drop method in sodium cacodylate, pH 6.2,

(100 mm) containing sodium acetate (0.1m), isopropanol (16–18%)

and poly(ethylene glycol) 3000 (18–22%) for 3 days at 208C. The

ChemBioChem 2010, 11, 2433 – 2447

ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chembiochem.org

2445

N. S. Scrutton et al.

6.2 Hz, 1H; CH_AH_B-OH), 3.54 (dd, J=10.7, 5.9 Hz, 1H; CH_AH_B-OH),

7.15–7.32 (m, 5H; Ph); ¹³C NMR (101 MHz; CDCl₃): d=140.7 (s, Cq),

129.27 (s, 2CH), 128.4 (s, 2CH), 126.0 (s, CH), 67.8 (s, CH₂), 39.8 (s,

CH₂), 37.92 (s, CH), 16.59 (s, CH₃).

stopped-flow spectrophotometer. The transients were analysed by

using nonlinear least squares regression analysis on an Acorn Risc

PC microcomputer with Spectrakinetics software (Applied Photo-

physics). The reductive half reactions (1.0 mL) were performed in

buffer (50 mm KH₂PO₄/K₂HPO₄, pH 7.0) containing NADPH (0.05–

1 mm) and PETN reductase (10 mm). The oxidative half reactions

(1.0 mL) were performed in buffer (50 mm KH₂PO₄/K₂HPO₄, pH 7.0)

containing dithionite-reduced PETN reductase (10 mm) and 2-cyclo-

hexan-1-one (2.5–50 mm). Absorption change of flavin reduction/

reoxidation was monitored continuously at 464 nm at 258C for 1 s.

The concentrations of substrates were always at least tenfold

greater than the enzyme concentration, ensuring pseudo-first

order conditions. The data points are the average of at least three

transients. The transients were fitted to a single exponential plot

from which observed rates were determined by using the rapid

equilibrium formalism of Strickland et al^[46][Eq. (2)] for the follow-

ing kinetic scheme [Eq. (3)].

2) (rac)-2-Phenylpropan-1-al oxime (19c): This was synthesised from

its respective aldehyde (10.0 mmol) by refluxing for 4 h in ethanol

(60 mL), hydroxylamine hydrochloride (19.0 mmol) and sodium ace-

tate (dry). The reaction was poured into water and extracted with

DCM, dried (MgSO₄powder) and evaporated, in vacuo. The residue

was purified by chromatography to yield 19c as an oil, which was

1

a mixture of two isomers E/Z 2.54:1 for both compounds. H NMR

(400 MHz, CDCl₃): d=1.44 (d, J=6.8 Hz, CH_3min), 1.46 (d, J=6.8 Hz,

CH_3maj), 3.68 (pentet, J=6.8 Hz, CH_maj), 4.45 (pentet, J=7.2 Hz,

CH_min), 6.83 (d, J=7.6 Hz, N=CH_min), 7.21–7.37 (m, 5H), 7.53 (d, J=

6.0 Hz, N=CH_maj); ¹³C NMR (100 MHz, CDCl₃): d=18.3, 18.8, 34.9,

40.4, 126.7, 126.9, 127.2, 127.4, 128.7, 128.8, 141.9, 154.9.

Analytical procedures: GC analysis of conversion and yields was

performed on a 30m DB-Wax (0.32 mm, 0.25 mm) column as de-

scribed previously.^[20,22,29]Yields of compounds (rac)-4c and oximes

(rac)-19c and (rac)-21c were determined by GC analysis as de-

scribed previously for their respective alkene substrate deriva-

tives.^[20,22,29]HPLC analysis of conversion, yields and enantiomeric

excess of 4a and (R/S)-4b was performed on a Chiracel OJ column

(hexane/iPrOH, 99:1 at 188C, flow 1 mLmin^ꢀ1, t_R: (R)- and (S)-dihy-

dro-methylcinnamaldehyde 11.5 and 14.6 min, respectively). The

detector wavelength was 195 nm. Determination of the enantio-

meric excess and absolute configuration of all other compounds

were performed as described previously.^{[20,22,29,31]}

k_ox¼ k₃½Sꢂ=ðK_dþ ½SꢂÞ

ð2Þ

ð3Þ

k₁

k₃

A þ B $ C

D

!

Acknowledgements

We thank Victoria Hare for assistance in the generation and crys-

tallography of PETN reductase T26S mutant. M.H. was supported

by a UK Biotechnology and Biological Sciences Research Council

(BBSRC) studentship. This work was funded by the BBSRC. N.S.S.

is a BBSRC Professorial Research Fellow and a Royal Society Wolf-

son Merit Award holder.

Biotransformations of alkenes by PETN reductase: Standard re-

actions for screening library mutants (1.0 mL) were performed in

buffer (50 mm KH₂PO₄/K₂HPO₄, pH 7.0) containing alkene (5 mm;

added as a concentrated stock in ethanol to a 5% final concentra-

tion), NADPH (6 mm) and PETN reductase (2 mm). The reactions

were shaken at 378C at 130 rpm for 24 h followed by reaction ter-

mination by extraction with ethyl acetate (0.9 mL) containing an in-

ternal standard (0.5% limonene), and dried by using MgSO₄. The

extracts were analysed by GC or HPLC to determine the %yield,

%conversion, and enantiomeric excess as described above.^[20,22,29]

Comparative biotransformation reactions between wild-type and

T26S enzyme were carried out as above, except for the use of a

NADP⁺/glucose-6-phosphate dehydrogenase or NADP⁺/glucose

dehydrogenase cofactor regeneration system^[20,22]in the place of

NADPH with a reaction temperature of 308C. Biphasic reactions

were carried out as described previously.^[20,22]

Keywords: asymmetric bioreduction · biocatalysis · directed

evolution · enantioselectivity · robotic library screening

[1] N. J. Turner, Trends Biotechnol. 2003 , 21 , 47 4 –4 78.

[2] R. R. Chirumamilla, R. Muralidhar, R. Marchant, P. Nigam, Mol. Cell. Bio-

chem. 2001 , 224 , 15 9 –1 68.

[3] S. Panke, M. Held, M. Wubbolts, Curr. Opin. Biotechnol. 2004 , 15 , 27 2 –

279.

[4] M. T. Reetz, D. Kahakeaw, J. Sanchis, Mol. Biosyst. 2009 , 5 , 11 5 –1 22.

[5] S. Luetz, L. Giver, J. Lalonde, Biotechnol. Bioeng. 2008 , 101 , 64 7 –6 53.

[6] M. T. Reetz, Angew. Chem. 2010, 122, DOI: 10.1002/ange.201000826;

Angew. Chem. Int. Ed. 2010, 49, DOI: 10.1002/anie.201000826; M. T.

Reetz, M. Bocola, L. W. Wang, J. Sanchis, A. Cronin, M. Arand, J. Zou, A.

Archelas, A. L. Bottalla, A. Naworyta, S. L. Mowbray, J. Am. Chem. Soc.

2009 , 131 , 733 4 –7 343.

[7] N. M. Antikainen, S. F. Martin, Bioorg. Med. Chem. 2005 , 13 , 270 1 –2 716.

[8] R. Stuermer, B. Hauer, M. Hall, K. Faber, Curr. Opin. Chem. Biol. 2007 , 11 ,

20 3 –2 13.

[9] M. Hall, C. Stueckler, W. Kroutil, P. Macheroux, K. Faber, Angew. Chem.

2007 , 119 , 400 8 –4 011; Angew. Chem. Int. Ed. 2007 , 46 , 393 4 –3 937.

[10] J. F. Chaparro-Riggers, T. A. Rogers, E. Vazquez-Figueroa, K. M. Polizzi,

A. S. Bommarius, Adv. Synth. Catal. 2007 , 349 , 152 1 –1 531; M. Hall, C.

Stueckler, H. Ehammer, E. Pointner, G. Oberdorfer, K. Gruber, B. Hauer, R.

Stuermer, W. Kroutil, P. Macheroux, K. Faber, Adv. Synth. Catal. 2008 ,

350 , 41 1 –4 18; M. Hall, C. Stueckler, B. Hauer, R. Stuermer, T. Friedrich,

M. Breuer, W. Kroutil, K. Faber, Eur. J. Org. Chem. 2008 , 151 1 –1 516; M.

Hall, C. Stueckler, B. Hauer, R. Stuermer, W. Kroutil, P. Macheroux, K.

Faber in Biotrans—8th International Symposium on Biocatalysis and Bio-

transformations, Oviedo, Spain, 2007; C. Stueckler, M. Hall, H. Ehammer,

E. Pointner, W. Kroutil, P. Macheroux, K. Faber, Org. Lett. 2007 , 9 , 540 9 –

5411; M. A. Swiderska, J. D. Stewart, Org. Lett. 2006 , 8 , 613 1 –6 133; M. A.

Swiderska, J. D. Stewart, J. Mol. Catal. B 2006 , 42 , 5 2 –5 4.

Steady-state kinetic analysis: The enzymes were deoxygenated

by passage through a BioRad 10DG column equilibrated in anaero-

bic reaction buffer. Steady state analyses were performed on a

Cary UV-50 Bio UV/Vis scanning spectrophotometer by using a

quartz cuvette (1 mL; Hellma) with a 1 cm path length. Reactions

(1 mL) were performed in buffer (50 mm KH₂PO₄/K₂HPO₄, pH 7.0)

containing NADPH (0.1 mm), PETN reductase (0–5 mm) and oxida-

tive substrate (1 mm added as concentrated stocks in ethanol to a

final concentration of 5%, v/v). The reactions were followed contin-

uously by monitoring NADPH oxidation at 340 nm for 1 min at

258C. Reactions with oxidizing substrates that absorb at 340 nm

were monitored at 365 nm. The concentration of enzyme, NADPH

and substrates within the reactions were determined by the extinc-

tion coefficient method by using values determined previously.^[20]

Fast reaction kinetics: Reductive and oxidative half reaction kinet-

ic experiments with PETN wild-type and His₈reductase were per-

formed anaerobically by using an Applied Photophysics SX.18MV-R

2446

www.chembiochem.org

ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ChemBioChem 2010, 11, 2433 – 2447

Site-Saturated Library Screening of PETN Reductase

[11] O. Warburg, W. Christian, Naturwissenschaften 1932 , 20 , 688.

[12] P. R. Binks, C. E. French, S. Nicklin, N. C. Bruce, Appl. Environ. Microbiol.

1996, 62, 1214–1219.

[13] T. B. Fitzpatrick, N. Amrhein, P. Macheroux, J. Biol. Chem. 2003 , 278 ,

1989 1 –1 9897.

[14] B. V. Adalbjçrnsson, H. S. Toogood, A. Fryszkowska, C. R. Pudney, T. A.

Jowitt, D. Leys, N. S. Scrutton, ChemBioChem 2010 , 11 , 19 7 –2 07.

[15] S. K. Padhi, D. J. Bougioukou, J. D. Stewart, J. Am. Chem. Soc. 2009 , 131 ,

327 1 –3 280.

[16] D. J. Bougioukou, S. Kille, A. Taglieber, M. T. Reetz, Adv. Synth. Catal.

2009 , 351 , 328 7 –3 305.

[30] H. Khan, T. Barna, R. J. Harris, N. C. Bruce, I. Barsukov, A. W. Munro, P. C.

Moody, N. S. Scrutton, J. Biol. Chem. 2004 , 279 , 3056 3 –3 0572.

[31] N. K. Scheffler, S. W. Miller, A. K. Carroll, C. Anderson, R. E. Davis, S. S.

Ghosh, B. W. Gibson, Mitochondrion 2001 , 1 , 16 1 –1 79.

[32] K. Durchschein, B. Ferreira-da Silva, S. Wallner, P. Macheroux, W. Kroutil,

S. M. Glueck, K. Faber, Green Chem. 2010 , 12 , 61 6 –6 19.

[33] H. Ohta, N. Kobayashi, K. Ozaki, J. Org. Chem. 1989 , 54 , 180 2 –1 804.

[34] K. Kitzing, T. B. Fitzpatrick, C. Wilken, J. Sawa, G. P. Bourenkov, P. Macher-

oux, T. Clausen, J. Biol. Chem. 2005 , 280 , 2790 4 –2 7913.

[35] W. L. DeLano, The PyMOL User’s Manual, DeLano Scientific, Palo Alto,

CA, USA, 2002.

[36] H. S. Toogood, J. M. Gardiner, N. S. Scrutton, ChemCatChem 2010 , 2 ,

89 2 –9 14.

[37] D. Xu, R. M. Kohli, V. Massey, Proc. Natl. Acad. Sci. USA 1999 , 96 , 355 6 –

3561.

[17] C. E. French, S. Nicklin, N. C. Bruce, J. Bacteriol. 1996, 178, 6623–6627.

[18] M. K. Trower, Methods Mol. Biol. 1996, 58, 329–333.

[19] T. B. Fitzpatrick, S. Auweter, K. Kitzing, T. Clausen, N. Amrhein, P. Macher-

oux, Protein Expr. Purif. 2004 , 36 , 28 0 –2 91.

[38] C. M. Clouthier, M. M. Kayser, M. T. Reetz, J. Org. Chem. 2006 , 71 , 843 1 –

8437; M. T. Reetz, L. W. Wang, M. Bocola, Angew. Chem. 2006 , 118 ,

125 8 –1 263; Angew. Chem. Int. Ed. 2006 , 45 , 123 6 –1 241.

[39] J. Sambrook, E. F. Fritsch, T. Maniatis, Molecular Cloning: A Laboratory

Manual, Vol. 3, 2nd ed., Cold Spring Harbor Laboratory Press, New York,

1989, appendix B.12.

[40] A. G. W. Leslie in Joint CCP4 & ESRF-EACBM Newsletter on Protein Crystal-

lography, Vol. 26, Serc Laboratory, Daresbury, Warrington, UK, 1992.

[41] Collaborative Computational Project Number 4, Acta Crystallogr. Sect. D

Biol. Crystallogr. 1994, 50, 760–763.

[42] A. Perrakis, R. Morris, V. S. Lamzin, Nat. Struct. Biol. 1999 , 6 , 45 8 –4 63.

[43] P. Emsley, K. Cowtan, Acta Crystallogr. Sect. D Biol. Crystallogr. 2004 , 60 ,

212 6 –2 132.

[44] A. K. Sinhababu, R. T. Borchardt, Tetrahedron Lett. 1983 , 24 , 22 7 –2 30.

[45] A. Fryszkowska, K. Fisher, J. M. Gardiner, G. M. Stephens, J. Org. Chem.

2008 , 73 , 429 5 –4 298.

[20] A. Fryszkowska, H. S. Toogood, M. Sakuma, J. M. Gardiner, G. M. Ste-

phens, N. S. Scrutton, Adv. Synth. Catal. 2009 , 351 , 297 6 –2 990.

[21] T. M. Barna, H. Khan, N. C. Bruce, I. Barsukov, N. S. Scrutton, P. C. Moody,

J. Mol. Biol. 2001 , 310 , 43 3 –4 47.

[22] K. M. Fox, P. A. Karplus, Structure 1994 , 2 , 108 9 –1 105.

[23] H. Khan, T. Barna, N. C. Bruce, A. W. Munro, D. Leys, N. S. Scrutton, FEBS

J. 2005 , 272 , 466 0 –4 671.

[24] H. Khan, R. J. Harris, T. Barna, D. H. Craig, N. C. Bruce, A. W. Munro, P. C.

Moody, N. S. Scrutton, J. Biol. Chem. 2002 , 277 , 2190 6 –2 1912.

[25] H. L. Messiha, N. C. Bruce, B. M. Sattelle, M. J. Sutcliffe, A. W. Munro, N. S.

Scrutton, J. Biol. Chem. 2005 , 280 , 2710 3 –2 7110.

[26] H. S. Toogood, A. Fryszkowska, V. Hare, K. Fisher, A. Roujeinikova, D.

Leys, J. M. Gardiner, G. M. Stephens, N. S. Scrutton, Adv. Synth. Catal.

2008 , 350 , 278 9 –2 803.

[27] C. Brideau, B. Gunter, B. Pikounis, A. Liaw, J. Biomol. Screening 2003 , 8 ,

63 4 –6 47.

[28] C. Stueckler, N. J. Mueller, C. K. Winkler, S. M. Glueck, K. Gruber, G. Stein-

kellner, K. Faber, Dalton Trans. 2010 , 39 , 847 2 –8 476.

[46] S. Strickland, G. Palmer, V. Massey, J. Biol. Chem. 1975, 250, 4048–4052.

[29] P. Galletti, E. Emer, G. Gucciardo, A. Quintavalla, M. Pori, D. Giacomini,

Org. Biomol. Chem. 2010 , 8 , 411 7 –4 123.

Received: September 3, 2010

Published online on November 9, 2010

ChemBioChem 2010, 11, 2433 – 2447

ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chembiochem.org

2447

Article Doi

Focused Directed Evolution of Pentaerythritol Tetranitrate Reductase by Using Automated Anaerobic Kinetic Screening of Site-Saturated Libraries

DOI: 10.1002/cbic.201000527

Source and publish data:

Authors:

Article abstract of DOI:10.1002/cbic.201000527

Full text of DOI:10.1002/cbic.201000527

Products guided by the article

R&D Labs maybe for 70005-01-5

Relevant to this article

Hot Product