Structure, activity and stereoselectivity of NADPH-dependent oxidoreductases catalysing the S-selective reduction of the imine substrate 2-methylpyrroline

Article abstract of DOI:10.1002/cbic.201402625

Oxidoreductases from Streptomyces sp. GF3546 [3546-IRED], Bacillus cereus BAG3X2 (BcIRED) and Nocardiopsis halophila (NhIRED) each reduce prochiral 2-methylpyrroline (2MPN) to (S)-2-methylpyrrolidine with >95 % ee and also a number of other imine substrates with good selectivity. Structures of BcIRED and NhIRED have helped to identify conserved active site residues within this subgroup of imine reductases that have S selectivity towards 2MPN, including a tyrosine residue that has a possible role in catalysis and superimposes with an aspartate in related enzymes that display R selectivity towards the same substrate. Mutation of this tyrosine residue - Tyr169 - in 3546-IRED to Phe resulted in a mutant of negligible activity. The data together provide structural evidence for the location and significance of the Tyr residue in this group of imine reductases, and permit a comparison of the active sites of enzymes that reduce 2MPN with either R or S selectivity.

Full text of DOI:10.1002/cbic.201402625

DOI: 10.1002/cbic.201402625
Full Papers
Structure, Activity and Stereoselectivity of NADPH-
Dependent Oxidoreductases Catalysing the S-Selective
Reduction of the Imine Substrate 2-Methylpyrroline
[
a]
[a]
[b]
[b]
[a]
Henry Man, Elizabeth Wells, Shahed Hussain, Friedemann Leipold, Sam Hart,
[a]
[b]
[a]
Johan P. Turkenburg, Nicholas J. Turner, and Gideon Grogan*
Oxidoreductases from Streptomyces sp. GF3546 [3546-IRED],
Bacillus cereus BAG3X2 (BcIRED) and Nocardiopsis halophila
that has a possible role in catalysis and superimposes with an
aspartate in related enzymes that display R selectivity towards
the same substrate. Mutation of this tyrosine residue—
Tyr169—in 3546-IRED to Phe resulted in a mutant of negligible
activity. The data together provide structural evidence for the
location and significance of the Tyr residue in this group of
imine reductases, and permit a comparison of the active sites
of enzymes that reduce 2MPN with either R or S selectivity.
(
NhIRED) each reduce prochiral 2-methylpyrroline (2MPN) to
S)-2-methylpyrrolidine with >95% ee and also a number of
(
other imine substrates with good selectivity. Structures of
BcIRED and NhIRED have helped to identify conserved active
site residues within this subgroup of imine reductases that
have S selectivity towards 2MPN, including a tyrosine residue
Introduction
[
7,8]
[9,10]
Chiral amines are important entities in pharmaceutical chemis-
try, both as synthetic intermediates and as drug compounds
themselves, and hence there has been a significant amount of
research directed towards their asymmetric synthesis. Many of
(DHFR)
and pteridine reductase (PTR),
which reduces the
C=N bond in pterin systems with the aid of NADPH as the
hydride donor. More recently, a number of naturally occurring
IREDs have also been applied in preparative biocatalytic syn-
[
1]
[11–19]
the approaches have used biocatalytic methods, owing to
the inherent asymmetric selectivity of enzymes, but also be-
cause of increasing pressure to adopt sustainable chemical
technology in industrial pharmaceutical synthesis. Biocatalytic
approaches to the asymmetric synthesis of amines have thus
thesis.
The most interesting of these were identified by
[
11–14]
Mitsukura, Nagasawa and co-workers,
who screened a
range of organisms for their ability to selectively reduce the
commercially available cyclic imine 2-methylpyrroline (2MPN,
1, Scheme 1).
[
2]
far included kinetic resolutions with hydrolases and asymmet-
ric transformations of a broad range of prochiral and racemic
starting materials by use of a range of enzyme classes includ-
[
3]
[4]
ing transaminases, amine oxidases and amine dehydrogen-
[
5]
ases.
A less well-explored route to asymmetric biocatalytic amine
synthesis is through the reduction of prochiral imines, in
a mode similar to that achieved for chiral alcohols through the
[6]
reduction of prochiral ketones by ketoreductases (KREDs).
This would be a convenient synthetic route due to the use of
prochiral imine substrates, which may be converted into the
corresponding amines with a theoretical yield of 100% and
with 100% ee. Some “imine reductase” (IRED) activities have
been studied extensively in the past, if not for biocatalytic
application; they include the enzymes dihydrofolate reductase
Scheme 1. Structures of substrates and products used in this study.
[
a] H. Man, E. Wells, S. Hart, Dr. J. P. Turkenburg, Prof. Dr. G. Grogan
Department of Chemistry, University of York
Heslington, York, YO10 5DD (UK)
They discovered two strains, Streptomyces GF3587 and Strep-
tomyces GF3546, which catalysed the reduction of 1 to the 2-
methylpyrrolidine (amine) products (R)-2 and (S)-2, respectively,
with high enantiomeric excess. The Mitsukura group purified
the enzymes responsible for these stereocomplementary re-
ductions—“3587-IRED” (Uniprot code M4ZRJ3) and “3546-IRED”
E-mail: gideon.grogan@york.ac.uk
[b] S. Hussain, Dr. F. Leipold, Prof. Dr. N. J. Turner
Manchester Institute of Biotechnology, University of Manchester
1
31 Princess Street, Manchester, M1 7DN (UK)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cbic.201402625.
ChemBioChem 2015, 16, 1052 – 1059
1052
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
(
M4ZS15), respectively—and in subsequent papers also report-
Results and Discussion
ed the cloning and heterologous expression in Escherichia coli
[
12,14]
of the genes that encode them.
These IREDs were shown
Activity of homologues of 3546-IRED from Streptomyces sp.
to be dependent for activity exclusively on the nicotinamide
cofactor NADPH, and to exist in solution as homodimers of
monomers, each of approximately 30 kDa. Subsequently,
Turner and co-workers have shown that cells of E. coli express-
ing 3546-IRED are able to catalyse the asymmetric reduction of
a wide range of monocyclic and bicyclic imines, providing
a biocatalyst with the potential for the preparation of chiral
GF 3546
Three IREDs with S selectivity towards 1 were chosen for activi-
ty and structural studies: 3546-IRED from Streptomyces sp.
GF 3546, BcIRED from B. cereus BAG3X2-2 and NhIRED from
N. halophila. The genes encoding each IRED were synthesised
with codon-optimisation for expression in E. coli and subcloned
[
15]
[21]
amine products.
In previous work,
into the pETYSBLIC-3C vector. The proteins, which were en-
[
16]
we determined the first X-ray crystal
coded with an N-terminal hexahistidine tag, were expressed in
the soluble fraction of E. coli BL21(DE3) cells and purified by
nickel affinity chromatography, followed by gel filtration.
The activity and stereoselectivity of 3546-IRED has been
structure of an NADPH-dependent IRED demonstrated to have
asymmetric imine reductase activity towards 1. The structure
of Q1EQE0 from Streptomyces kanamyceticus, which shares
[15]
50% sequence identity with 3587-IRED, was shown to be com-
documented by Turner and co-workers, and it has continued
to serve as a model enzyme for mutational studies reported
below. The first new enzyme was a putative IRED from
B. cereus BAG3X2-2 (J7M26, BcIRED). This enzyme shared 57%
sequence identity with 3546-IRED and 78% similarity (Figure S1
in the Supporting Information). The purified enzyme (Fig-
ure S2) was challenged with six substrates in the presence of
NADPH, and the results of the biotransformations are shown in
Table 1.
posed of an intimate dimer of monomers each displaying a
two-domain structure: an N-terminal Rossman fold largely
responsible for NADPH binding, and a C-terminal bundle.
NADPH was observed bound at the interface between the N-
terminal domain of the first subunit and the C-terminal
domain of the second. The two domains were connected by
a long interdomain helix, from which an aspartate residue—
Asp187—projected into the active site towards the nicotin-
amide ring. Superposition of the
Q1EQE0 structure with structur-
ally related oxidoreductases of
the hydroxyisobutyrate dehydro-
Table 1. Results of biotransformations of imine substrates in Scheme 1 with IRED enzymes. Each reaction mix-
ture (12 mL) contained 0.25 mgmL of pure IRED and 5 mm substrate from a 0.25m stock solution in DMF,
with NADPH and cofactor recycling as detailed in the Experimental Section. Chiral chromatography traces for
product analysis can be found in Figures S3–S8.
À1
[
20]
genase family suggested that
Asp187 might be involved in the
catalytic mechanism of imine re-
duction, and substitution of this
residue for either an Asn or an
Ala residue indeed removed the
activity of Q1EQE0 towards 1 as
measured by spectrophotomet-
ric assays of NADPH oxidation in
the presence of 1.
Absolute configuration, conversion [%], enantiomeric excess [%]
Enzyme
1
3
5
7
9
11
[
a]
3546-IRED
BcIRED
NhIRED
S, 57, 95
S, 98, >99
S, 80, >99
S, >98, 98
S, 90, >99
S, 86, >99
n.d.
S, 57, 35
R, 90, 82
n.d.
R, 76, 63
S, 100, 85
n.d.
S, 8, 18
S, 53, 98
S, 98, 98
S, 94, 81
S, 97, 68
[a] Data taken from ref. [15], in which reactions were performed with whole cells of E. coli, expressing 3546-
IRED. n.d.=not determined.
Now that the structure of an
IRED that is R-selective for the reduction of 2MPN has been
solved, it would be of interest to compare this to structures of
homologues that are S-selective for reduction of this substrate.
BcIRED catalysed the reduction of 1 to give the S-amine 2 as
predicted, and also reduced 2-methylpiperidine (3) to amine
(S)-4, each with >99% ee. Imines 5, 7 and 9, with increasingly
larger groups in the 2-position, were reduced with lower levels
of conversion and stereoselectivity. Conversion was very poor
for substrate 9. In the case of 7, the stereoselectivity was for-
mally reversed as a result of a switch according to the Cahn–
Ingold–Prelog priority rules, but the product was still the result
of hydride attack at the equivalent face of the imine substrate
as for 1. 1-Methyl-3,4-dihydroisoquinoline (11) was reduced to
the S-amine with 94% ee.
[17]
The structure of 3546-IRED has recently been reported, but
neither a detailed analysis of the active site, nor mutational
studies on this enzyme as a representative member of en-
zymes with S selectivity towards 2MPN were presented. In this
report we describe further studies on IREDs that catalyse the S-
selective reduction of 2MPN, including the structures and activ-
ities of enzymes from Bacillus cereus BAG3X2-2 (BcIRED) and
Nocardiopsis halophila (NhIRED). These data permit a compari-
son of IREDs that display stereocomplementary activity toward
Purified NhIRED (Figure S2), from N. halophila, which shared
45% sequence identity with 3546-IRED, and 65% similarity,
was also shown to reduce the panel of imine substrates, again
with predominantly S selectivity. In contrast with BcIRED, how-
ever, the activity and stereoselectivity with the larger sub-
strates 5, 7 and 9 were inverted (Table 1), with products now
resulting from hydride attack from the opposite face to that
2MPN. We also provide structural evidence for an active tyro-
sine residue, in place of an aspartate, as a possible residue in-
volved in catalysis in this group of enzymes having S selectivity
towards 1, its location having previously only been suggested
[
18]
on the basis of modelling studies.
ChemBioChem 2015, 16, 1052 – 1059
www.chembiochem.org
1053
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
seen for 1 and 3. Although unexpected, this result has many
precedents in the history of alcohol dehydrogenase (ADH) bio-
transformations, with the selectivity of ADHs such as those
were very few differences observed in secondary elements or
tertiary structure throughout. The superimposition of BcIRED
with the Q1EQE0–NADPH complex (PDB ID: 3ZHB) also re-
vealed that in place of Asp187 in Q1EQE0, which had been
suggested to have a catalytic role in IRED catalysis in that
enzyme, there was a tyrosine residue—Tyr188—at the “ceiling”
of the NADPH-binding cleft and projecting into the proximal
section of the NADPH binding site as presented in Figure 1
(right). Superimposition of BcIRED with 3546-IRED (4OQY) also
reveals a tyrosine residue (Tyr169) in an equivalent position.
The structure of BcIRED complexed with NADPH in one
active site displayed an rmsd with the apo structure of 1.3 Š,
reflective of a perturbation on NADPH binding leading to a
relative closure of the active site between the domains. The
distance between, for example, the Ca atoms of Val138 and
Met192, which are on opposite faces of the NADPH binding
cleft on either side of the plane of the side chain Tyr188,
closes from 12.6 to 11.8 Š. Such a perturbation was also
[
22]
[23]
from Thermoanaerobium
and Sporobolomyces,
for exam-
ple, inverted when challenged with larger ketone substrates.
A comparison of the selectivity and activity of BcIRED and
[
15]
NhIRED with those of 3546-IRED
illustrates that the last
enzyme displays better conservation of stereoselectivity to-
wards 1, 3 and 11 (Table 1). BcIRED and NhIRED are, however,
slightly more active than 3546-IRED towards 1, as determined
by comparison of k /K values, by factors of 0.5 and 6.0, re-
cat
m
spectively (Table 2).
Table 2. Kinetic constants determined for IRED enzymes and substrate
MPN (1).
2
Enzyme
k
cat
À1
K
[mm
m
k
cat/K
À1
m
[17]
À1
À1
À3
[
s
]
]
[s mm ”10
]
observed for (S)-IRED by Müller and co-workers and explains
in part the larger rmsd for 3546-IRED in relation to BcIRED,
against the value in relation to Q1EQE0, in which no such
movement was observed. On binding of NADPH, the side
chain of Ser113 disengages from H-bonding contact with the
side chain of Gln189 and His262 and rotates 1808 to form an
[a]
3
546-IRED
0.024
0.032
0.013
11.82
8.32
1.10
2
3
11
BcIRED
NhIRED
[a] Data taken from ref. [15].
Structures of IRED homologues
from Bacillus and Nocardiopsis
The structures of each of the
IRED homologues were then de-
termined by X-ray crystallogra-
phy. The selection of multiple al-
ternative enzymes for structural
studies proved fruitful in respect
of the acquisition of structures
of IREDs that were S-selective for
the reduction of 1. A structure of
Figure 1. Structure of BcIRED in complex with NADPH. Subunits A and B are shown in ribbon format in light blue
and green, respectively. NADPH is shown in cylinder format with carbon atoms in grey at the dimer interface.
Left: Structure of dimer. Right: Active site showing Tyr188 projecting towards the nicotinamide ring of NADPH.
Residues are labelled with one-letter codes for clarity, and distances are given in Š. Electron density corresponds
3546-IRED to a resolution of 3 Š
was obtained, but during the to the F
o
ÀF
c
(omit) map calculated in the absence of NADPH ligand and contoured at a level of 3s. NADPH
atoms have been added for clarity.
preparation of this manuscript
a structure at a superior resolu-
tion of 1.9 Š was published by
[17]
Müller and co-workers (PDB ID: 4OQY), so our data are not
included here. The structure of BcIRED was determined in two
forms: an apo form to a resolution of 1.74 Š and a complex
with NADPH that was determined to a resolution of 1.81 Š.
Each BcIRED structure was determined from crystals belonging
to the P2 2 2 space group and featuring one dimer in the
H-bond with the O2D oxygen of the NADPH ribose (at a
distance of 3.0 Š), and also a weak H-bond (3.7 Š) with the
phenolic hydroxy group of Tyr188. The side chain phenolic
hydroxy group of Tyr188 itself is 5.0 Š from the C-4 atom of
the nicotinamide ring of NADPH that donates hydride in the
reductive reaction, and this compares well with the 5.2 Š ob-
served in NADPH-dependent PTR, which is thought to employ
Tyr194 as a proton donor to the nascent amine in the relevant
1
1 1
asymmetric unit. Details of the model can be found in Fig-
ure S9.
[10]
The structure of the apo form of BcIRED, again a dimer dis-
playing the domain-swapping structure with an interdomain
helix, superimposed with Q1EQE0—which has R selectivity to-
wards 1— with an overall rmsd of 1.8 Š. The BcIRED dimer also
superimposed with 3546-IRED (4OQY) with an overall rmsd of
reaction. Other interactions with the cofactor are summar-
ised in Figure S10. A discussion of further structural differences
between the IREDs follows a description of the structure of
NhIRED.
The structure of NhIRED was determined and refined to a res-
olution of 2.39 Š. Details of the model can be found in the
Supporting Information (Figure S11).The rmsds between the
2.6 Š. Each of these values was illustrative of the structural sim-
ilarity within the wider IRED family of enzymes. Indeed, there
ChemBioChem 2015, 16, 1052 – 1059
www.chembiochem.org
1054
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
dimer structure of this enzyme and those of BcIRED, 3546-IRED
and Q1EQE0 were 1.3, 2.3 and 1.9 Š, respectively. As in the
cases of BcIRED and 3546-IRED, a tyrosine residue—Tyr174—
was observed to project into the putative NADPH binding cleft
between domains. Although no complex of this NhIRED with
NADPH was forthcoming, the structure was determined in
complex with the crystallisation detergent additive n-octyl b-d-
glucoside (OBDG) within the active site (Figure 2, right). The
detergent molecule participated in many active site interac-
3546-IRED as Tyr174 and Tyr169, respectively, and also that it
superimposed with Asp187 in Q1EQE0, Asp187 having been
[16]
suggested to have a role in catalysis in that enzyme. Further
conserved residues adjacent to this Tyr, and within 6 Š of the
side-chain phenol, include Gln189 (BcIRED numbering), Ser113
as described previously, along with His262, with which it inter-
acts in the apo form of BcIRED, and Met192, which corre-
sponds to Met173 in 3546-IRED, but to Leu178 in NhIRED.
Additionally, the phenolic hydroxy group of Tyr188 in BcIRED is
approximately 4.5 Š from the
backbone carbonyl of a Val resi-
due—Val138—conserved in both
NhIRED and 3546-IRED. The hy-
drophobic residues detailed as
interacting with the tail of OBDG
in NhIRED are also conserved in
both BcIRED and 3546-IRED. In
addition to Leu178, however,
NhSIRED displays other differen-
ces from 3546-IRED and BcSIRED,
such as Asp241, which corre-
Figure 2. Structure of NhIRED in complex with OBDG. Subunits A and B are shown in ribbon format in lilac and
coral, respectively. OBDG is shown in cylinder format with carbon atoms in grey at the dimer interface. Left: Struc- sponds to Gly in the other ho-
ture of dimer. Right: Active site showing Tyr174 projecting into the active site, together with H-bonding contacts
of the ligand with residues at the active site. Residues are labelled with one-letter codes for clarity, and distances
mologues. These substitutions
might contribute to differences
are given in Š. Electron density corresponds to the F
ligand and contoured at a level of 3s. OBDG atoms have been added for clarity.
o
ÀF
c
(omit) map calculated in the absence of the OBDG
in selectivity observed between
NhSIRED and the other enzymes
for substrates with larger side
chains.
tions with residues including Tyr174, which was observed to
form a H-bond at a distance of 2.7 Š from the 4-hydroxy
group, whereas the 6-hydroxy group of the exocyclic hydroxy-
methyl system formed a H-bond of 3.1 Š in length with the
side chain of Asn213. The n-octyl tail was bound in a hydropho-
bic region of the distal portion of the binding cleft, interacting
with Phe181, Phe218, Trp182 and Phe282. These interactions
are summarised in Figure S12. Despite the contact with Tyr174,
it is unlikely that this represents a substrate-binding mode for
this IRED, however, because the molecule is bound in the
distal portion of the binding cleft as observed in Figure 2. In
this location, the ligand projects away from the NADPH bind-
ing site, where the necessary interaction between the C-4 hy-
dride of NADPH and the electrophilic carbon would take place.
Indeed NhIRED did not display any activity towards OBDG or
its parent sugar (glucose) in spectrophotometric assays. The
possibility that this complex is illustrative of a mode of entry
for imine substrates cannot be excluded, however.
Structural evidence for tyrosine as putative catalytic residue
in IREDs that are S-selective for the reduction of 2MPN
Mutation of Asp187 to either Asn or Ala in Q1EQE0 resulted in
mutants that displayed negligible activity towards 1 as deter-
mined by the standard UV assay of NADPH oxidation measured
at 340 nm. The observed substitution of Asp for a Tyr residue
in BcIRED, NhIRED and 3546-IRED prompted us to mutate the
Tyr169 to Phe in 3546-IRED. This resulted in a mutant that also
displayed negligible activity towards 1 in the standard UV
assay. This is in agreement with a recent report by Hauer and
co-workers, in which an equivalent Tyr residue in a homologue
from Paenibacillus elgii was mutated to Ala to yield a largely
[18]
inactive mutant.
These results suggest, by analogy with
Q1EQE0, that this Tyr residue might have an active role in cat-
alysis in 3546-IRED, BcIRED and NhIRED, perhaps as proton
donor to the nascent amine in the reductive reaction. Certainly
Tyr has been shown to fulfil a similar role in other imine reduc-
Conservation of active sites in structures of IREDs that are
S-selective for the reduction of 2MPN
[10]
tases such as NADPH-dependent PTR and also in sugar de-
[24,25]
hydrogenases such as xylose–xylitol reductase.
The side-
A sequence alignment of the IREDs discussed in this study, as
well as of 3546-IRED and the enzymes Q1EQE0 and 3587-IRED,
which are R-selective for 2MPN reduction, is given in the Sup-
porting Information (Figure S1). Between BcIRED, NhIRED and
chain phenolic hydroxy group of Tyr188 in BcIRED itself is 5.0 Š
from the C-4 atom of the nicotinamide ring of NADPH that do-
nates hydride in the reductive reaction, and this compares well
with the distances of 5.2 and 4.6 Š observed for relevant inter-
actions in PTR and xylose–xylitol reductase, respectively. In
PTR, however, the active site environment is somewhat differ-
ent, because the Tyr side chain is part of a triad in which it in-
teracts with a lysine residue that might have a role in lowering
3546-IRED enzymes, an examination of the alignment and
available structures confirmed that Tyr188 (BcIRED), at the “ceil-
ing” of the NADPH-binding cleft and projecting towards the
nicotinamide ring of NADPH, is conserved in NhSIRED and
ChemBioChem 2015, 16, 1052 – 1059
www.chembiochem.org
1055
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
the pK of the Tyr and stabilising the transition state of the re-
Comparison of active sites of IREDs that are stereo-
complementary for the reduction of 2MPN
a
action, as well as an aspartate residue that might be a proton
[
10]
donor to Tyr. In fact, PTR exhibits a pH optimum for activity
[
9]
of 4.7–6.0 depending on the substrate, whereas the pH opti-
mum for 3546-IRED activity with 1, for example, was deter-
mined to be approximately 7.5 (Figure 3), with the midpoint of
the declining alkaline slope at approximately 8.0, indicative of
A superimposition of the active sites of BcIRED and Q1EQE0,
which are specific for the S- and R-selective reductions of
2MPN, respectively, is shown in Figure 4. Truncated forms of
a higher pK for the Tyr in IRED catalysis by that enzyme.
a
Figure 4. Superimposition of Q1EQE0 (green ribbons) with BcIRED (coral).
Carbon atoms of side chains and of NADPH are shown in green and coral,
respectively. Residues are labelled with one-letter codes for clarity and with
the prefix designations “R” (e.g., R-D187) for Q1EQE0 and “S” (e.g., S-Y188)
for BcIRED indicative of their selectivity for 2MPN.
Figure 3. pH optima of IREDs illustrated by the activity of “tyrosine” IRED
3
546-IRED (^) and “aspartate” IREDs Q1EQE0 (
&
) and 3587-IRED (~), assayed
against 2MPN (1).
the structures that featured only 85 residues of the secondary
elements completing the active site cleft superimposed with
an rmsd of 1.3 Š, lower than that for whole dimer. Coupled
with the sequence alignment, a comparison of the BcIRED
structure with that of Q1EQE0 shows that, apart from the re-
placement of Tyr188 with Asp187, Met255 (BcIRED) is replaced
by Thr254 (Q1EQE0), Trp196 is conserved as Trp195, and
Ser259 is replaced by Thr258 (Figure 4).
Interestingly, the pH activity profile of the tyrosine-contain-
ing 3546-IRED is different from those of the aspartate-contain-
ing IREDs (Figure 3), which exhibit a broader range of activity
overall, together with pK values suggested by the midpoint of
a
the declining alkaline slope of 8.5 (Figure 3), high for an aspar-
tate, but similar to that reported for the aspartate-dependent
[
26]
IRED E. coli dihydrofolate reductase.
The high pK values for the aspartate residues in Q1EQE0
Gly256 in BcSIRED (not shown) is a His residue both in
Q1EQE0 and also in 3587-IRED. From the sequences and struc-
tures, all of the determinants of NADPH binding in BcIRED,
NhIRED, 3546-IRED and Q1EQE0 appear to be conserved, and
the cofactor-binding loops are very similar, with the re-face of
the nicotinamide ring stacking against a Met residue (Met33 in
BcIRED) in all cases, the O2 atom of ribose bound by Ser113,
and the 2’-ribose hydroxy phosphate on NADPH secured by
a phosphate-binding loop with interactions to Arg53, Thr54
and Lys57 as part of a conserved R53TXXK motif in all enzyme
structures.
a
and 3587-IRED are not unexpected because the aspartate is in
each case held within a hydrophobic pocket formed by
Ala135, Leu137 and Leu191 (Q1EQE0 numbering), the first two
of which are not conserved as hydrophobic residues in 3546-
IRED and homologues. Also, in Q1EQE0 the aspartate is 7.9 Š
from the NADPH C4 atom, similar to the equivalent distance in
DHFR, in which a network of hydrogen bonds through the
substrate, rather than direct proton transfer, is implicated in
the protonation of the nascent amine in the reductive reaction.
Overall, an interesting analogy can therefore be made between
the tyrosine-containing IREDs and the aspartate-containing
IREDs, in which the distance between the putative “proton
donor” and the NADPH C4 atom might provide clues to possi-
ble contributions to mechanism; in the former case there is
crystallographic evidence from PTR to suggest that proton
donation is directly from the side chain, whereas in the latter
case the consensus on DHFR now appears to be that although
the aspartate residue indeed assists in catalysis (and is buried
in a hydrophobic hole as in Q1EQE0), its role is as part of an H-
bonding network that assists the role of a water molecule in
donation of the proton to the nitrogen atom. The clarification
of these mechanistic aspects awaits the structures of both
classes of IRED in complex with imine substrates.
The comparison detailed above shows that there is remarka-
bly little difference within the active sites of IREDs that display
stereocomplementary behaviour towards the model substrate
2MPN, even within the region at or near the site between C4
of the nicotinamide cofactor and the putative “catalytic” active
site residue, either Asp or Tyr in each case. One way in which
to examine the structural basis for stereocomplementary be-
haviour would be to compare the sequences of enzymes of S
and R selectivity towards 2-MPN, to perform reciprocal muta-
tions at sites in the NADPH binding cleft, and to assess their al-
tered stereoselective behaviour. However in the case of IREDs,
simple reciprocal mutation of residues might be uninformative,
owing to probable contributions to catalysis from neighbour-
ChemBioChem 2015, 16, 1052 – 1059
www.chembiochem.org
1056
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
ing residues. Such reciprocal mutations on enzymes 3587-IRED
and 3546-IRED have met with limited success: mutation of
Asp172 (Asp187 in Q1EQE0) to Tyr in 3587-IRED and of Tyr169
to Asp in 3546-IRED have led to poor expression and/or insolu-
ble expression of the mutants (data not shown). Among other
residues, the mutation of Met236 in 3546-IRED (conserved as
Met255 in BcIRED and shown at the opening of the active site
in Figure 4) to the Thr residue found in 3587-IRED also resulted
in insoluble protein. Conversely, mutation of Thr241 (Thr254 in
For gene expression, recombinant vector(s) containing IRED genes
and mutants were used to transform E. coli BL21(DE3), and resul-
tant colonies were grown overnight on lysogeny broth (LB) agar
À1
with kanamycin (30 mgmL ) as antibiotic marker. One colony of
the relevant recombinant strain was used to inoculate a starter cul-
À1
ture of LB medium (5 mL) containing kanamycin (30 mgmL ). This
was then grown for 18 h with shaking at 180 rpm at 378C. The
turbid starter culture was used to inoculate larger volumes of LB
À1
broth (500 mL) containing kanamycin (30 mgmL ) in a 2 L Erlen-
meyer flask, in which cells were grown until the OD600 had reached
a value of 0.6. The expression of IRED genes and mutants was in-
duced at this point by the addition of isopropyl b-d-1-thiogalacto-
pyranoside (IPTG, to a final concentration of 1 mm). Cultures were
then grown in an orbital shaker at 180 rpm at 188C for 18 h. After
this time, cell cultures were centrifuged at 4225g for 15 min in
a Sorvall GS3 rotor in a Sorvall RC5B Plus centrifuge. Cell pellets
were resuspended in Tris·HCl buffer (“buffer”, 50 mm, pH 7.5, 25 mL
per L of cell growth) that also contained sodium chloride
(300 mm). The cell suspensions were sonicated for 3”30 s bursts at
Q1EQE0) in 3587-IRED to
a Met residue gave mutant
Thr241Met, which converted the substrate 2MPN to the R
product 2 with 95% ee, much as was observed for the wild
type. The currently available structures of IREDs suggest that it
is probable that IRED stereoselectivity is governed by complex
interactions that might include not only the environment cre-
ated by the amino acid side chains in the active site, but also
domain motions that are hinted at in the relative movements
between domains observed in BcIRED on NADPH binding. It is
also the case that the known structures of IREDs represent
only a small fraction of IRED sequences now available, and that
the phenomena observed for known enzymes might not be
generally observed throughout the IRED family.
4
8C with 1 min intervals.
The insoluble and soluble fractions were separated by centrifuga-
tion at 26892g in a Sorvall SS34 rotor for 30 min. The supernatants
were filtered through a 2 mm Amicon filter and were then loaded
onto a 5 mL His-Trap Chelating HP column. The columns were
washed with five column volumes of buffer containing imidazole
(30 mm), and IRED proteins and mutants were eluted with an imi-
Conclusions
dazole gradient (30–500 mm) over twenty further column volumes.
Fractions containing IREDs were identified by analysis by SDS-PAGE
and pooled. These were then concentrated by use of a 10 kDa cut-
off Centricon filter membrane. Concentrated IREDs were then
loaded onto an S75 Superdex 16/60 gel filtration column that had
been pre-equilibrated with buffer. The columns were then eluted
with buffer at a flow rate of 1 mLmin . Fractions containing pure
IREDs or mutants as determined by SDS-PAGE analysis were
pooled and stored at 48C.
Imine reductases (IREDs) provide promising new biocatalysts
for the asymmetric production of chiral amines. The future ex-
ploitation of these enzymes should benefit greatly from struc-
tural investigations that help to reveal the molecular bases of
mechanism and selectivity. They should also provide a rational
basis for the engineering of IREDs for altered or improved ac-
tivities in the future.
À1
Protein crystallisation: Pure BcIRED and NhIRED were subjected to
crystallisation trials with use of a range of commercially available
screens in 96-well sitting-drop format in which each drop consisted
of protein (150 nL) and precipitant reservoir solution (150 nL). For
BcIRED, initial crystals were obtained in PEG 3350 (25%, w/v),
Experimental Section
Chemicals: Chemicals, including commercially available substrates,
cofactors and buffer and media components, were purchased in
general from Sigma–Aldrich. Imine substrates and products were
synthesised as described in ref. [15]. Substrates 3, 5, 7, 9 and 11
and amine products 2, 4, 6, 8, 10 and 12 were also synthesised as
reported in ref. [15].
MgCl (0.2m), and HEPES (0.1m) at pH 7.5, with protein at a concen-
2
À1
tration of 50 mgmL . For NhIRED, initial crystals were obtained in
PEG 3350 (18%, w/v), MgCl (0.25m) and Bis–Tris (0.1m) at pH 5.5,
2
À1
with protein at a concentration of 60 mgmL . Larger crystals for
diffraction analysis under optimised conditions were prepared by
the hanging-drop method in 24-well plate Linbro dishes with 2 mL
drops consisting of a 1:1 ratio of mother liquor to protein. For
BcIRED the best crystals were obtained in PEG 3350 (30%, w/v),
Gene synthesis, cloning, expression and protein purification:
The genes encoding 3546-IRED, 3587-IRED, BcIRED and NhIRED
were synthesised by GeneArt (Invitrogen), with the sequence opti-
mized by use of the GeneArt server programme for expression in
E. coli. For subcloning, the genes were amplified by PCR from the
received plasmids containing the synthetic genes with use of the
primers detailed in Table S1. After analysis of the PCR on agarose
gels, bands of the appropriate size were isolated from the gel with
the aid of a PCR Cleanup kit (Qiagen). Genes were then sub-cloned
into the pET-YSBL-LIC-3C vector by a previously published proce-
dure. The resultant recombinant vectors were used to transform
E. coli XL1-Blue cells (Novagen), yielding colonies that in turn gave
plasmids through the use of standard miniprep procedures that
were sequenced to confirm the identities and sequences of the
genes. Mutant genes 3546-IRED Tyr169Phe, 3546-IRED Met236Thr
and 3587-IRED Thr241Met were generated by use of a Quikchange
kit from Agilent according to the manufacturer’s instructions. The
incorporation of mutations was then verified by DNA sequencing.
MgCl (0.2m) and HEPES (0.1m) at pH 7.5 after 3 months at 188C.
2
For NhIRED, the best crystals were obtained in OBDG (1%, w/v),
PEG 3350 (18%, w/v), MgCl (0.25m) and Bis–Tris (0.1m) at pH 5.5
2
after 2 months at 188C.
For both enzymes, co-crystallisation with NADPH (10 mm) did not
yield crystals under the same buffer conditions. In order to obtain
the NADP(H) complex of BcIRED, native crystals were transferred
from the growth drop into a cryogenic solution consisting of the
mother liquor containing ethylene glycol (20%, v/v) and NADPH
(20 mm) and incubated for 1 h, after which they were immediately
flash-cooled in liquid nitrogen. Native crystals of both BcIRED and
NhIRED were flash-cooled in a cryogenic solution containing only
the mother liquor and ethylene glycol (10%, v/v). All crystals were
tested for diffraction in-house with a Rigaku Micromax-007HF in-
[
21]
ChemBioChem 2015, 16, 1052 – 1059
www.chembiochem.org
1057
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
strument fitted with Osmic multilayer optics and a Marresearch
MAR345 imaging plate detector. Those crystals that diffracted to
a resolution of equal to, or better than, 3 Š resolution were re-
tained for dataset collection at the Diamond synchrotron.
Biotransformations: Biotransformations of substrates 1, 3, 5, 7, 9
and 11 were performed with the purified enzyme(s) or mutants
[16]
and NADPH recycling by the method described previously. Reac-
tions were each performed in a 12 mL total reaction volume con-
À1
taining pure IRED (0.25 mgmL ) and substrate (5 mm) from a
Data collection, structure solution, model building and refine-
ment of BcIRED and NhIRED: Complete datasets described in this
report were collected at Diamond Light Source, Didcot, Oxford-
shire, UK. The apo-BcIRED and the NhIRED complex with OBDG
were collected on beamline I04-1, and the BcIRED NADPH complex
on beamline I04. Data were processed and integrated by use of
stock solution in DMF (0.25m). The final concentration of dimeth-
ylformamide in the reaction was 2% (v/v). For cofactor recycling,
NADPH (5 mg, 0.006 mmol), glucose-6-phosphate (3.5 mg,
0
.013 mmol) and a glucose-6-phosphate dehydrogenase suspen-
sion (G8404 from Sigma–Aldrich, 12 mL per mL of total reaction
volume) were added. After the overnight reaction at 308C, 500 mL
of the mixture was removed, quenched with sodium hydroxide
[
27]
[28]
XDS and scaled by using SCALA included in the Xia2 process-
[29]
ing system. Data collection statistics are given in Table S2.
(10m, 30 mL) and extracted with dichloromethane (2”500 mL).
Samples (100 mL) of biotransformations of 1, 3, 5 and 7 were deri-
vatised by N-acetylation with acetic anhydride and triethylamine
and analysed by GC as described in the Supporting Information.
Samples of biotransformations of 9 and 11 were analysed directly
by HPLC as detailed in ref. [15]. Chiral analysis of products was per-
formed as detailed in Figures S3–S8, which contain illustrative
chromatograms for the separation of enantiomers of products 2, 4,
In each case the crystals of BcIRED were in space group P2 2 2 .
1
1 1
[30]
The structure of BcIRED was solved with MOLREP, by use of a mo-
nomer model of the IRED Q1EQE0, PDB ID: 3ZGY. The solution(s)
each contained two molecules in the asymmetric unit, represent-
ing one dimer. The solvent content in each case was 60%. The
structures were built and refined by use of iterative cycles with
[31]
[32]
Coot and REFMAC, the latter employing local NCS restraints.
For the NADPH complex of BcIRED, after building and refinement
of the protein and water molecules, clear residual density was ob-
served in the omit maps at one of the subunit dimer interfaces.
This was modelled and refined as NADPH. The final structures ex-
hibited Rcryst and Rfree values of 17.6 and 21.7 (apo) and 23.2 and
6, 8, 10 and 12.
Acknowledgements
We thank the industrial affiliates of the Centre of Excellence for
Biocatalysis, Biotransformations and Biomanufacture (CoEBio3)
for awarding studentships to H.M. and E.W. S.H. was supported
by a CASE studentship with AstraZeneca. F.L. received funding
from the European Union’s Seventh Framework Programme FP7/
3
0.0% (NADPH complex), respectively. For NhIRED, the crystals
were in space group P2 2 2 and had eight molecules in the asym-
1
1 ,
metric unit, representing four dimers. The solvent content in this
case was 50%. The structure was again solved with MOLREP by
use of a monomer of Q1EQE0 as a search model, and was built
and refined as for BcIRED. Four dimers featured in the asymmetric
unit. After building and refinement, clear omit map density was
again observed at the NADPH-biding site at the dimer interface,
but was clearly not NADPH. It was successfully modelled as the
crystallisation additive OBDG (n-octyl b-d-glucoside). The final
structure exhibited Rcryst and Rfree values of 23.0 and 26.6. All struc-
2007–2013 under grant agreement n8 266025 (BIONEXGEN).
Keywords: amines · asymmetric catalysis · imines · IREDs ·
NADPH · oxidoreductases
[33]
tures were finally validated with PROCHECK. Refinement statistics
for all structures are presented in Table S2. The Ramachandran plot
for the apo-BcIRED showed 97.5% of residues to be situated in the
most favoured regions, 1.8% in additional allowed and 0.7% resi-
dues in outlier regions. For the NADPH complex, the correspond-
ing values were 97.5, 2.1 and 0.4% respectively. For NhIRED the
values were 96.3, 2.6 and 1.1%. The coordinates and structure fac-
tors for the BcIRED, the BcIRED NADPH complex and NhIRED have
been deposited in the Protein Data Bank with the accession codes
[
[
1] M. Hçhne, U. T. Bornscheuer, ChemCatChem 2009, 1, 42–51.
2] S. Buchholz, H. Grçger, Biocatalysis in the Pharmaceutical and Biotech-
nology Industries (Ed.: R. N. Patel), CRC, London, 2006, pp. 829–847.
[3] S. Mathew, H. Yun, ACS Catal. 2012, 2, 993–1001.
[4] D. Ghislieri, A. P. Green, M. Pontini, S. C. Willies, I. Rowles, A. Frank, G.
Grogan, N. J. Turner, J. Am. Chem. Soc. 2013, 135, 10863–10869.
[
5] M. J. Abrahamson, E. Vµzquez-Figueroa, N. B. Woodall, J. C. Moore, A. S.
Bommarius, Angew. Chem. Int. Ed. 2012, 51, 3969–3972; Angew. Chem.
2
012, 124, 4036–4040.
[
[
[
[
6] M. M. Musa, R. S. Phillips, Catal. Sci. Technol. 2011, 1, 1311–1323.
7] D. R. Smith, J. M. Calvo, Nucleic Acids Res. 1980, 8, 2255–2274.
8] M. R. Sawaya, J. Kraut, Biochemistry 1997, 36, 586–603.
4
D3D, 4D3F and 4D3S, respectively.
Enzyme assays: The NADPH-dependent imine reductase activity of
9] B. Nare, L. W. Hardy, S. M. Beverley, J. Biol. Chem. 1997, 272, 13883–
recombinant IREDs and mutants in this study was assessed as pre-
13891.
[16]
viously for Q1EQE0 by means of UV spectrophotometry. Activity
was determined with a Spectramax M2 spectrophotometer/plate
reader (Molecular Devices) by monitoring the decrease in NADPH
[
10] D. G. Gourley, A. W. Schuttelkopf, G. A. Leonard, J. Luba, L. W. Hardy,
S. M. Beverley, W. N. Hunter, Nat. Struct. Mol. Biol. 2001, 8, 521–525.
[11] K. Mitsukura, M. Suzuki, K. Tada, T. Yoshida, T. Nagasawa, Org. Biomol.
Chem. 2010, 8, 4533–4535.
12] K. Mitsukura, M. Suzuki, S. Shinoda, T. Kuramoto, T. Yoshida, T. Naga-
sawa, Biosci. Biotechnol. Biochem. 2011, 75, 1778–1782.
13] T. Nagasawa, T. Yoshida, H. Yamamoto, N. Kimoto (Daicel Chemical
Industries, Ltd.), US Patent Application 20110287494, 2009.
14] K. Mitsukura, T. Kuramoto, T. Yoshida, N. Kimoto, H. Yamamoto, T. Naga-
sawa, Appl. Microbiol. Biotechnol. 2013, 97, 8079–8086.
[15] F. Leipold, S. Hussain, D. Ghislieri, N. J. Turner, ChemCatChem 2013, 5,
3505–3508.
[16] M. Rodríguez-Mata, A. Frank, E. Wells, F. Leipold, N. J. Turner, S. Hart,
J. P. Turkenburg, G. Grogan, ChemBioChem 2013, 14, 1372–1379.
17] T. Huber, L. Schneider, A. Präg, S. Gerhardt, O. Einsle, M. Müller, Chem-
CatChem 2014, 6, 2248–2252.
À1
À1
at 340 nm (e=6.22 mm cm ) for substrate 2MPN (1). Activity was
determined with microtest plates (Sterilin, Newport, UK). Reaction
mixtures (0.2 mL) contained pure enzyme (0.3 mg), NADPH
[
[
[
(
750 mm) and the appropriate amount of substrate (2 mL) from
a 100-fold concentrated stock in DMSO. The reaction was started
by adding the enzyme to the mixture. One unit of imine reductase
is defined as the amount of protein that oxidized 1 mmol NADPH
per minute. Kinetic parameters were determined with purified
enzyme and analysed by nonlinear regression analysis based on
Michaelis–Menten kinetics with use of the program QtiPlot.
[
ChemBioChem 2015, 16, 1052 – 1059
www.chembiochem.org
1058
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
[
[
[
18] P. N. Scheller, S. Fademrecht, S. Hofelzer, J. Pleiss, F. Leipold, N. J. Turner,
B. M. Nestl, B. Hauer, ChemBioChem 2014, 15, 2201–2204.
19] M. Gand, H. Müller, R. Wardenga, M. Hçhne, J. Mol. Catal. B 2014, 110,
[26] E. Ohmae, Y. Miyashita, S.-I. Tate, K. Gekko, S. Kitazawa, R. Kitahara, K.
Kuwajima, Biochim. Biophys. Acta Proteins Proteomics 2013, 1834, 2782–
2794.
[27] W. Kabsch, Acta Crystallogr. Sect. D Biol. Crystallogr. 2010, 66, 125–132.
[28] P. Evans, Acta Crystallogr. Sect. D Biol. Crystallogr. 2006, 62, 72–82.
[29] G. Winter, J. Appl. Crystallogr. 2010, 43, 186–190.
126–132.
20] N. K. Lokanath, N. Ohshima, K. Takio, I. Shiromizu, C. Kuroishi, N. Oka-
zaki, S. Kuramitsu, S. Yokoyama, M. Miyano, N. Kunishima, J. Mol. Biol.
2
005, 352, 905–917.
[30] A. Vagin, A. Teplyakov, J. Appl. Crystallogr. 1997, 30, 1022–1025.
[31] P. Emsley, K. Cowtan, Acta Crystallogr. Sect D Biol. Crystallogr. 2004, 60,
[
[
21] K. E. Atkin, R. Reiss, N. J. Turner, A. M. Brzozowski, G. Grogan, Acta Crys-
tallogr. Sect. F Struct. Biol. Cryst. Commun. 2008, 64, 182–185.
22] E. Keinan, E. K. Hafeli, K. K. Seth, R. Lamed, J. Am. Chem. Soc. 1986, 108,
2126–2132.
[
32] G. N. Murshudov, A. A. Vagin, E. J. Dodson, Acta Crystallogr. Sect. D Biol.
Crystallogr. 1997, 53, 240–255.
162–169.
[
33] R. A. Laskowski, M. W. Macarthur, D. S. Moss, J. M. Thornton, J. Appl.
Crystallogr. 1993, 26, 283–291.
[
23] D. Zhu, L. Hua, J. Org. Chem. 2006, 71, 9484–9486.
24] B. Petschacher, S. Leitgeb, K. L. Kavanagh, D. K. Wilson, B. Nidetzky, Bio-
chem. J. 2005, 385, 75–83.
[
[25] A. H. Ehrensberger, R. A. Elling, D. K. Wilson, Structure 2006, 14, 567–
Received: October 27, 2014
575.
Published online on March 24, 2015
ChemBioChem 2015, 16, 1052 – 1059
www.chembiochem.org
1059
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim