Altering the regioselectivity of a nitroreductase in the synthesis of arylhydroxylamines by structure-based engineering

Article abstract of DOI:10.1002/cbic.201500070

Nitroreductases have great potential for the highly efficient reduction of aryl nitro compounds to arylhydroxylamines. However, regioselective reduction of the desired nitro group in polynitroarenes is still a challenge. Here, we describe the structure-based engineering of Escherichia coli nitroreductase NfsB to alter its regioselectivity, in order to achieve reduction of a target nitro group. When 2,4-dinitrotoluene was used as the substrate, the wild-type enzyme regioselectively reduced the 4-NO₂ group, but the T41L/N71S/F124W mutant primarily reduced the 2-NO₂ group, without loss of activity. The crystal structure of T41L/N71S/F124W and docking experiments indicated that the regioselectivity change (from 4-NO₂ to 2-NO₂) might result from the increased hydrophobicity of residues 41 and 124 (proximal to FMN) and conformational changes in residues 70 and 124. The regioselectivity of nitroreductase NfsB from E. coli toward 2,4-dinitrotoluene was shifted from the 4-NO₂ group to the 2-NO₂ group without loss of activity, by introducing three mutations: T41L, N71S, and F124W. This study provides an example of a tailored enzyme for regioselective synthesis of the target arylhydroxylamines.

Full text of DOI:10.1002/cbic.201500070

DOI: 10.1002/cbic.201500070
Full Papers
Altering the Regioselectivity of a Nitroreductase in the
Synthesis of Arylhydroxylamines by Structure-Based
Engineering
[
a]
[b]
[a]
[a]
[a]
Jing Bai, Yong Zhou, Qi Chen, Qing Yang, and Jun Yang*
Nitroreductases have great potential for the highly efficient re-
duction of aryl nitro compounds to arylhydroxylamines. How-
ever, regioselective reduction of the desired nitro group in pol-
ynitroarenes is still a challenge. Here, we describe the struc-
ture-based engineering of Escherichia coli nitroreductase NfsB
to alter its regioselectivity, in order to achieve reduction of
a target nitro group. When 2,4-dinitrotoluene was used as the
substrate, the wild-type enzyme regioselectively reduced the
4-NO group, but the T41L/N71S/F124W mutant primarily re-
2
duced the 2-NO group, without loss of activity. The crystal
2
structure of T41L/N71S/F124W and docking experiments indi-
cated that the regioselectivity change (from 4-NO to 2-NO )
2
2
might result from the increased hydrophobicity of residues 41
and 124 (proximal to FMN) and conformational changes in resi-
dues 70 and 124.
Introduction
Arylhydroxylamines are important intermediates in the synthe-
nitrobenzoate (another dinitroarene), PnbA reductase from
[
1]
[2]
sis of pharmaceuticals and fine chemicals. They are primarily
prepared through reduction of aryl nitro compounds, and vari-
ous methods have been developed to accomplish this transfor-
Lactobacillus plantarum strain WCFS1 selectively reduces the 4-
[14]
NO group. Because of their high regioselectivity toward pol-
2
ynitro and/or dinitro compounds, nitroreductases are potential
catalysts for regioselective synthesis of target hydroxylamines.
However, the native regioselectivity of a nitroreductase
toward a particular dinitroarene might not satisfy the require-
ments of industrial or clinical applications. Protein engineering
has been used to alter the regioselectivity of nitroreductases.
The preferences of two mutants of E. coli nitroreductase NfsB
(T41L and T41L/N71S, obtained by random mutagenesis) for
the more cytotoxic 4-hydroxylamine derivative of CB1954 were
[
3]
mation, including metal-mediated reduction, catalytic hydro-
[
4]
genation reduction using immobilized Pt catalysts or doped
[
5]
[6]
Pd catalysts, and biocatalytic reduction using baker’s yeast,
[
7]
[8]
plant cells, or bacterial nitroreductases. These processes can
chemoselectively reduce mononitro compounds bearing other
substituent groups, and achieve the direct conversion of aryl
nitro compounds into their corresponding arylhydroxylamines.
However, regioselective reduction of a nitro group at an arbi-
trarily selected site in the case of dinitro or polynitro com-
pounds is still a challenge.
[15,16]
shown to be greater than that of the wild-type enzyme.
Although the crystal structures of the single mutants T41L and
[16]
Nitroreductases catalyze the NAD(P)H-dependent reduction
of nitro groups to hydroxylamino groups using FMN as pros-
thetic group, in many cases with high regioselectivity. Bacterial
nitroreductases show different regioselectivities toward the
prodrug CB1954 (5-(aziridin-1-yl)-2,4-dinitrobenzamide). NfsA
N71S were determined, the effects of these two mutations
on regioselectivity were unclear, as were the effects of other
residues. Biochemical analysis of enzyme regioselectivity
toward various substrates might enable comparative studies to
be performed, thereby resulting in a better understanding of
the influences of different substituent groups on regioselectivi-
ty. However, current research is mainly focused on the regio-
selective activation of CB1954; nitroreductase regioselectivity
against other dinitroarenes have not been widely reported. It
would therefore be useful to investigate this further, in order
to achieve the rational design of mutant enzymes with ideal
regioselectivity.
from Escherichia coli selectively reduces the 2-NO group rather
2
[
9]
than the 4-NO group, whereas NfsB reduces the 2-NO and
2
2
[
10]
[11]
4
-NO groups equally; NemA and AzoR from E. coli, YwrO
2
[
12]
from Bacillus amyloliquefaciens, and Yfk0 from Bacillus lichen-
[
13]
iformis show a preference for the 4-NO group. For 2,4-di-
2
[
a] Dr. J. Bai, Dr. Q. Chen, Prof. Dr. Q. Yang, Dr. J. Yang
School of Life Science and Biotechnology
Dalian University of Technology
The aim of this study was to investigate the effects of key ni-
troreductase residues and substrate substituents on enzyme
regioselectivity. E. coli NfsB was chosen as the enzyme for
No. 2 Linggong Road, Dalian-116023 (China)
E-mail: junyang@dlut.edu.cn
mutagenic alteration, because it has been well characterized
[b] Dr. Y. Zhou
[10,17]
and crystal structures in both the ligand-bound
and un-
School of Software Technology, Dalian University of Technology
[18]
bound
forms have been determined. 2,4-Dinitrotoluene
3
21 Tuqiang Street, Development Zone, Dalian (China)
(
24DNT) and 2,4-dinitroanisole (24DNAN) were selected so as
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cbic.201500070.
to examine the possible effects of substituent groups on regio-
ChemBioChem 2015, 16, 1219 – 1225
1219
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
selectivity. The best mutant successfully altered the native re-
gioselectivity of NfsB toward these two substrates, and did not
exhibit any activity loss. Based on site-directed mutagenesis,
structural analysis, and molecular-docking experiments, the
molecular basis of regioselective reduction of the nitroreduc-
tase NfsB is proposed. This might provide guidance for the
design of desired regioselective catalysts in the future.
small peak corresponding to 2-amino-4-nitrotoluene (2ANT),
thus indicating that purified NfsB regioselectively reduced the
4-NO group of 24DNT.
2
Alteration of regioselectivity by site-directed mutagenesis
It is of interest to explore nitroreductases with different regio-
selectivity, capable of producing different hydroxylamine iso-
mers, some of which could potentially be used as precursors
of pharmaceuticals, pesticides, and dyes. Therefore, we tried to
evolve mutant enzyme(s) for regioselective reduction of the 2-
Results and Discussion
Analysis of regioselectivity of nitroreductase NfsB toward
NO group.
2
2,4-dinitrotoluene
[10,17,18]
Based on the crystal structures of NfsB,
four residues
E. coli NfsB has been shown to be active toward a variety of
were chosen for mutagenesis: T41, N71, F70, and F124. Resi-
dues 41 and 71 were selected because they are inside the
active pocket and their mutations were shown to alter the re-
[
19]
nitro substrates. The reductive products of 24DNT generated
by purified NfsB were identified by analyzing the metabolites
by HPLC and mass spectrometry.
[15,16]
gioselectivity of NfsB against CB1954.
Eight diverse mu-
After anaerobic incubation for 30 min with nitroreductase
tants were constructed (T41V, T41I, T41L, T41F, N71T, N71S,
N71Q, and N71L), and their effects on 24DNT reduction were
investigated. F70 and F124 were previously thought to be re-
NfsB, the substrate 24DNT (t =25.4 min) disappeared, and
R
a major product was detected (t =13.9 min; Figure 1A). The
R
[20]
molecular mass of the product was 168 Da, and it was identi-
fied as the monohydroxylamine, thus suggesting that NfsB re-
duced only one nitro group of 24DNT to generate the hydrox-
ylamine; no other amine products were observed (Figure 1B).
To further investigate the nitroreduction regioselectivity, we
identified the position of the nitro group that was reduced, by
converting the product to the corresponding amine. The prod-
ucts of 24DNT reduction were converted almost exclusively to
lated to the catalytic activity; they are located at the en-
trance to the active site, and thus might mediate regioselectivi-
ty by controlling substrate access to the pocket. They were
therefore mutated to Ala, Asn, and Trp (small, polar, and aro-
matic residues, respectively). Activity and regioselectivity
against 24DNT were determined by HPLC.
Substitutions at residue 41 resulted in clear changes to the
preference for the 2- and 4-NO groups of 24DNT (Figure 2).
2
4-amino-2-nitrotoluene (4ANT; Figure 1C), with only a very
T41V and T41I mutants catalyzed reduction of 2-NO to a lesser
2
extent, whereas T41L and T41F
shifted the selectivity from exclu-
sively 4-NO2 to both groups
equally. The N71 and F124 mu-
tants showed only slightly modi-
fied regioselectivity; no change
was observed for F70 mutants
(
Figure S1 in the Supporting In-
formation). These results suggest
that hydrophobicity of the side
chain at position 41 is a determi-
nant of NfsB regioselectivity
(
change increasing with increas-
ing hydrophobicity).
In order to investigate addi-
tive and/or synergistic effects of
[21]
multiple residues, double mu-
tants were construct based on
mutations that yielded improve-
ments in the desired regioselec-
tivity (T41L, T41F, N71S, F124N,
and F124W). Apart from T41L/
F124N, all mutants showed im-
proved regioselectivity for 2-NO2
over the corresponding single
mutants (Figure S1). T41L/N71S
was highly active and regioselec-
Figure 1. HPLC and MS analyses of products of 24DNT reduction by purified NfsB. A) HPLC profile after incuba-
tion for 30 min (lower trace) compared with control (no enzyme, upper trace). B) MS spectra of the reduced prod-
+
uct (P) of 24DNT. The ions at m/z 191.1 and m/z 213.1 (positive mode) are the adducts [M+Na] and [MÀ
+
À
H+2Na] , respectively. The ion at m/z 202.9 (negative mode) is the adduct [M<M+ >Cl] . Therefore, the molec-
ular mass of the product was 168 Da. C) HPLC profile of the conversion product of peak P (lower trace) compared
with that of authentic standard of amine derivative (upper trace). D) Structure of reduced product (P) of 24DNT.
[15]
tive toward CB1954,
but was
ChemBioChem 2015, 16, 1219 – 1225
www.chembiochem.org
1220
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
possible synergistic effects with residues 41 and 71. The triple
mutant T41L/N71S/F124W reduced the 2-NO group to obtain
2
2-hydroxylamine as the major product (Figure 3C), thus dem-
onstrating that the combination of the single-site mutation
with the double mutation T41L/N71S further improved the de-
sired regioselectivity.
Effects of substituent groups on regioselectivity of nitro-
reductase NfsB
Wild-type NfsB displays different regioselectivity toward 24DNT
[10]
and CB1954, thus implying that substrate substituents also
affect regioselectivity. Mutagenesis studies showed that in-
creasing the hydrophobicity of residue 41 changes enzyme
substrate preference, thereby enabling it to reduce the 2-NO2
group of 24DNT, probably by anchoring and stabilizing the 1-
methyl group within the active pocket so that its adjacent
nitro group is settled above N5 of FMN. The difference in hy-
drophobicity between the methyl and aziridinyl groups might
therefore lead to different regioselectivity.
Figure 2. Effects of side-chain variation at residue 41 on regioselectivity.
A) Side-chains at residue 41 of NfsB. B) Amounts of 2- or 4-hydroxylamine
products of 24DNT reduction catalyzed by wild-type NfsB and the T41 mu-
tants. The hydroxylamines were analyzed by HPLC (UV detection at 254 nm).
To test this hypothesis, the substrate 24DNAN (containing
the larger methoxy group adjacent to the 2-NO group) was
2
unable to produce 2-hydroxylamine from 24DNT regioselec-
tively (Figure 3B).
reduced by NfsB and selected mutants. Wild-type NfsB reduced
either 2-NO or 4-NO , similarly to CB1954 (Figures 3D and S2).
2
2
Notably, although F124W (single mutant) was less beneficial
for altering regioselectivity, F124W/T41L and F124W/N71S dis-
T41L and T41F preferentially reduced 2-NO , and the double
2
and triple mutants showed further improved regioselectivity.
T41L/N71S/F124W displayed 100% regioselectivity in favor of
the 2-hydroxylamine product (Figure 3F); that is, as the hydro-
phobicity of the substituent group increases (CH <CH O<
played a preference for the 2-NO group of 24DNT, similarly to
2
T41L/N71S. Substitution to a larger aromatic residue at posi-
tion 124 (T41L/F124N and T41L/F124W) favored 2-hydroxyl-
amine production, probably because of stronger hydrophobic
and/or p-stacking interactions. Residue 124 was therefore a
candidate for altering the enzyme regioselectivity, based on
3
3
C H N), the preference for the adjacent nitro group increases.
2
4
The changes in regioselectivity of the mutants were not ac-
companied by activity loss: T41L/N71S/F124W (the most active
mutant) exhibited similar activity against 24DNT
and a 4.4-fold increase in k /K against 24DNAN
cat
m
(
Table 1).
Molecular basis of regioselectivity of NfsB
Nitroreductase-catalyzed reduction occurs by a “ping-
pong” mechanism: the prosthetic group FMN trans-
fers a hydride from NAD(P)H to the adjacent nitro
[10,22]
group of the bound substrate.
The regioselectivi-
ty might therefore be determined by the binding ori-
entation of the substrate: either a 2-NO or 4-NO
2
2
group is positioned above N5 of reduced FMN, there-
by leading to the corresponding 2-hydroxylamine or
4-hydroxylamine product.
The crystal structure of the T41L/N71S/F124W
mutant was resolved to gain insight into how its
active pocket is remodeled to accommodate sub-
strates in different orientations for regioselective re-
duction of the 2-NO group (Table 2). A comparison
2
of the structures of T41L/N71S/F124W and wild-type
NfsB showed no significant difference in overall struc-
ture (root mean square deviation 0.57 Š) but some
conformational changes were observed in the active
pocket (Figure 4A and B). The backbone conforma-
Figure 3. HPLC analysis of products of left: 24DNT and right: 24DNAN reduction cata-
lyzed by wild-type NfsB, T41L/N71S, and T41L/N71S/F124W. Peaks corresponding to 2-hy-
droxylamine and 4-hydroxylamine products and substrates (24DNT and 24DNAN) are in-
dicated.
ChemBioChem 2015, 16, 1219 – 1225
www.chembiochem.org
1221
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
sults. The wild-type NfsB allowed
only one orientation of 24DNT,
with the 4-NO2 group pointing
to N5 of FMN and the methyl
group exposed to the solution
[
a,b]
Table 1. Apparent steady-state kinetic parameters
NfsB and selected mutants.
for the reduction of 24DNT and 24DNAN by wild-type
24DNAN
NfsB/mutant
24DNT
K
m(app)
k
[s
cat(app)
k
cat(app)/Km(app)
À1 À1
K
[mm]
m(app)
k
[s
cat(app)
k
cat(app)/Km(app)
À1 À1
À1
À1
[mm]
]
[s mm
]
]
[s mm ]
(
Figure 5). 24DNT assumed a dif-
wild-type
T41L
T41L/N71S
T41L/N71S/F124W
171.7Æ1.4
297.3Æ5.0
63.9Æ5.6
23.15Æ0.79
31.40Æ0.89
5.32Æ0.21
2.95Æ0.04
0.135
0.106
0.083
0.118
514.2Æ40.1
9.33Æ0.13
0.018
0.021
0.052
0.079
ferent orientation with T41L/
N71S/F124W, with 2-NO2 close
to the catalytic site and the
methyl group bound in a hydro-
phobic region formed by the
side chains of L41 and W124.
851.9Æ75.1 18.01Æ0.91
160.6Æ13.3
58.4Æ6.7
8.35Æ1.14
4.61Æ0.35
25.04Æ0.36
[
a] Kinetic parameters were obtained by continuous monitoring of NADH consumption at 340 nm. All reactions
used NADH (60 mm) as a co-substrate in Tris·HCl buffer (20 mm, pH 7.0) at 378C. The data were fitted to the Mi-
chaelis–Menten equation by nonlinear regression (Origin 8.5), and the errors were calculated from triplicates
from the same enzyme preparation. [b] Calculated assuming two moles of NADH per mole of nitro group.
[22]
Analysis of the docking struc-
tures showed that 24DNT con-
Table 2. X-ray data collection and model refinement statistics for T41L/
N71S/F124W.
PDB ID
3X21
Refinement
no. reflections
(F>0)
48630
Data collection
space group
unit cell [Š] a
P2
95.410,
1
R
R
work (95% data)
free (5% data)
0.210
0.256
b
59.924,
r.m.s.d. bond
c
220.063
99.723
50.00–3.00
distance [Š]
r.m.s.d. bond
angle [8]
0.007
1.127
b [8]
resolution [Š]
(
3.05–3.00)
Ramachandran plot
[% residues]
no. observations
no. unique
completeness [%]
average I/s(I)
2496629
48758(2405)
99.4(99.9)
9.869(6.206)
9.9(16.6)
most-favored regions
additional allowed
regions
94.6
5.4
Rmerge [%]
generously allowed
0
regions
2
average B [Š ] no. non-H atoms
protein
water
FMN
52.3/16860
49.1/6
50.6/310
tion of L41 was identical to that of T41 (wild-type), but the leu-
cine side chain was longer and in contact with W124, thus pos-
sibly leading to a more hydrophobic region above FMN. More-
over, the W124 ring was further from the isoalloxazine ring of
FMN (relative to wild-type NfsB), thus allowing space for FMN
to expand slightly. Although S71 was in a similar position to
that of N71 (wild-type), the side chain of the adjacent residue
Figure 4. Structural comparison of wild-type NfsB and T41L/N71S/F124W
mutant. Overlay of A) overall structures and B) active sites of wild-type NfsB
(
blue) and T41L/N71S/F124W mutant (pink). Bound FMN and residues at po-
sitions 41, 70, 71, and 124 are shown as sticks. Images were prepared with
[23]
PyMOL version 1.5.0.4.
(
F70) had apparently rotated to be opposite the pocket, thus
tacts the active site residues mainly by hydrophobic and p-
stacking interactions (no hydrogen bonding). In both binding
modes, 24DNT stacked above the FMN plane and was bound
to the flavin by p–p stacking and nitro–p-stacking interactions.
Although the stacking interactions were important for 24DNT
binding, they were observed in two binding modes and might
not explain the shift in regioselectivity. In addition to the stack-
ing interactions with FMN, various other interactions with sur-
rounding residues occurred. For the wild-type enzyme, the 4-
generating a wider entrance to the active site. These confor-
mational changes (F70 and W124) might therefore reduce pos-
sible steric effects on substrate binding. The results suggest
that the alteration in regioselectivity of T41L/N71S/F124W
might be partly related to a hydrophobicity change and re-
duced steric effects, thereby enabling the 2-NO group to be
2
positioned close to the catalytic site.
To further understand the effects of these mutations on re-
gioselectivity at the molecular level, we performed molecular
docking simulations of 24DNT with T41L/N71S/F124W and the
wild-type enzyme in AutoDock. This showed that the binding
modes of 24DNT were different in the active pockets of the
two enzymes, in good agreement with the experimental re-
NO group was bound in a pocket containing T41, E165, G166,
2
and F124, and was involved in nitro–p stacking interactions
with F124, whereas the 2-NO group and the methyl group
2
were exposed to the solution (Figure 5A). For T41L/N71S/
F124W, the methyl group was in hydrophobic contact with
ChemBioChem 2015, 16, 1219 – 1225
www.chembiochem.org
1222
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Conclusion
The regioselectivity of NfsB toward 24DNT was successfully
shifted from the 4-NO group to the 2-NO group without loss
2
2
of activity, by introducing three mutations: T41L, N71S, and
F124W. The triple mutant displayed 100% regioselectivity for
the 2-NO2 group of 24DNAN. It is proposed that the side
chains of L41, S71, and W124 enhance the hydrophobicity of
the region above FMN and slightly change the shape of the
catalytic pocket, thereby resulting in more favorable binding of
2
-NO than of 4-NO . This study investigated alterations in ni-
2
2
troreductase regioselectivity by structure-based engineering,
and provides an example of a tailored enzyme for the reduc-
tion of a specific nitro group in a polynitroarene.
Experimental Section
Chemicals and reagents: NADH (Purity >98%) was obtained from
Sigma–Aldrich. 24DNT, 24DNAN, 2ANT, 4ANT and FMN were pur-
chased from J&K (Beijing, China). Unless otherwise stated, other
chemicals and reagents were of analytical grade. Chromatograph-
ic-grade methanol and acetonitrile were supplied by J&K (Beijing,
China). Oligonucleotide primers were obtained from Takara (Dalian,
China).
Cloning, expression, and purification of NfsB and its mutants:
The nfsB gene (Gene ID: 945778) was amplified from E. coli K12 ge-
nomic DNA with primers nfsBWTF and nfsBWTR (Table S1) and
cloned into pET28a (Novagen/Merck Millipore). Site-directed NfsB
mutant were generated by overlap extension PCR and sequenced
to confirm the amplification products. In brief, for each nfsB muta-
genesis, two partially overlapping PCR products were generated by
using the upstream primer nfsBWTF or downstream primer
nfsBWTR with the corresponding internal primer (Table S1), and
then annealed to create the full-length product, which was digest-
ed with Nde I and EcoR I restriction enzymes, ligated into pET28a,
and then transformed into E. coli DH5a for amplification. All mu-
tants were confirmed by sequencing and then transformed into
E. coli BL21(DE3) cells (Novagen) for expression as N-terminal His6
enzymes.
Figure 5. Molecular docking of 24DNT in the active sites of A) wild-type NfsB
residues involved in substrate binding shown in blue) and B) T41L/N71S/
F124W mutant (pink). The substrate 24DNT (yellow) assumes different orien-
tations in the two active pockets, thus leading to reduction of a 4-NO
group or a 2-NO group. Bound FMN is shown as a stick model. Images
were prepared with PyMOL version 1.5.0.4.
(
2
2
[
23]
L41, G166, and W124, and the 2-NO group was held by a
2
nitro–indo interaction with W124 and positioned close to N5
of FMN to be reduced (Figure 5B). The hydrophobic interac-
tions of L41 and W124 and/or the nitro–indo interaction of
W124 might favor the binding mode of the 2-NO group. In
2
the crystal structure of the T41L/N71S/F124W mutant, the side
chain of F70 (adjacent to S71) and the side chain of W124 had
rotated away from the active pocket to create an open en-
trance; this might reduce steric hindrance of the 1-methyl
group and favor its insertion. Residues 41, 71, and 124 exerted
synergistic effects on the regioselectivity of NfsB.
A single colony was grown overnight and inoculated into LB
medium supplemented with kanamycin. Nitroreductase expression
was induced with isopropyl-b-thiogalactopyranoside (IPTG, 0.3 mm)
at the early exponential phase (OD600 =0.5–0.6). The harvested cell
pellet was suspended in sodium phosphate buffer (pH 7.4) contain-
ing NaCl (500 mm) and imidazole (20 mm), and lysed by high pres-
sure homogenization at 48C. Recombinant nitroreductase was pu-
rified on a HiTrap chelating HP column (GE Healthcare) and buffer
exchanged into Tris·HCl buffer (20 mm, pH 7.0) containing NaCl
The substrate substituents also affected regioselectivity. In
contrast to the preference for the 4-NO group of 24DNT, the
2
wild-type NfsB reduced either of the two nitro groups of
(
50 mm) by using a PD-10 desalting column (GE Healthcare). For ki-
netic studies, the active fractions were incubated with pure FMN at
8C for at least 30 min before buffer exchange. The protein sam-
24DNAN and CB1954. In the NfsB–CB1954 complex, the hydro-
phobic aziridinyl group was bound within the pocket, so its
4
[
10]
adjacent nitro group was close to the active site. Similarly,
the docking results for 24DNAN show that the methoxy group
ples were more than 95% pure as estimated by SDS-PAGE, and
their concentrations were determined by Bradford assays (BSA as
the standard).
was inserted into the pocket when the 2-NO group was re-
2
duced (Figure S3). However, the methyl group of 24DNT was
Products and regioselectivity of 24DNT and 24DNAN reduction:
The products generated from the reduction of 24DNT and 24DNAN
catalyzed by the nitroreductases were identified by reverse phase
HPLC. The reaction mixture contained substrate (100 mm), NADH
(200 mm), and enzyme (300 nm) in Tris·HCl buffer (20 mm, pH 7.0),
and was incubated at 378C for 30 min in an anaerobic atmosphere
exposed to the solution and the 2-NO group could not be re-
2
duced (Figure 5A). The binding modes of the three substrates
suggest that the hydrophobic substituent groups might act as
anchors to position the adjacent nitro group above N5 of
FMN.
ChemBioChem 2015, 16, 1219 – 1225
www.chembiochem.org
1223
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
(
Mitsubishi Anaeropack; Thermo Scientific) 30 min prior to and fol-
server (http://lorentz.immstr.pasteur.fr/docking/index.php). Sub-
strate orientation and enzyme–substrate interactions were ana-
lyzed with PyMOL and LigPlot.
lowing the addition of nitroreductase. The reaction products were
extracted with ethyl acetate and injected into a Hypersil C18
column (5 mm, 4.6”250 mm; Thermo Scientific). For 24DNT, the re-
actants were injected into a column equilibrated with methanol/
[23,30]
water (20:80), and eluted in isocratic mixture of methanol/water Acknowledgements
(
50:50) over 15 min, followed by a 50–90% (v/v) methanol linear
À1
gradient over 10 min (flow rate 0.8 mLmin ). For 24DNAN, the
products were injected into a column equilibrated with acetoni-
We would like to thank Dr. Tian Liu for valuable discussions and
suggestions. This work was supported by China National Funds
for Distinguished Young Scientists (31425021), the National High
Technology Research and Development Program of China
trile/water (20:80), and eluted with a linear gradient of acetonitrile
À1
(
40–80%, flow-rate 0.8 mLmin , monitored at 254 nm). The prod-
ucts of 24DNT and 24DNAN reduction were identified by HPLC-MS
in an 1100 Series HPLC system (HewlettPackard) and a Finnigan
TSQ 7000 triple quadrupole mass spectrometer. The mass spec-
trometer was equipped with an atmospheric pressure ionization
(
(
2011AA10A204), National Natural Science Foundation of China
61202252), Specialized Research Fund for the Doctoral Program
of Higher Education of China (P20120041120052), China Postdoc-
toral Science Foundation (2013M541214), Dalian Foundation of
Science and Technology (2013E13SF114), The Fundamental Re-
search Funds for the Central Universities (DUT13JB08), and
Shanghai Key Lab of Chemical Biology (China).
(
API) interface and an electrospray ion (ESI) source in positive or
negative ion mode.
To distinguish the position of the hydroxylamino group, the reduc-
tive products of 24DNT and 24DNAN were chemically reduced to
their corresponding amines, which were identified by comparison
of their retention times. The hydroxylamino products were incubat-
ed for 15 min at 408C in freshly prepared potassium phosphate
Keywords: arylhydroxylamines
·
enzyme
catalysis
·
nitroreductases · protein engineering · regioselectivity
buffer (20 mm, pH 9.0) containing Na S (5 mm) and sodium thiogly-
2
[24]
colate (0.05%, w/v).
[
1] a) R. J. Knox, F. Friedlos, T. Marchbank, J. J. Roberts, Biochem. Pharmacol.
1991, 42, 1691–1697; b) D. G. Smith, A. D. Gribble, D. Haigh, R. J. Ife, P.
Lavery, P. Skett, B. P. Slingsby, R. Stacey, R. W. Ward, A. West, Bioorg.
Med. Chem. Lett. 1999, 9, 3137–3142; c) P. M. Vyas, S. Roychowdhury,
P. M. Woster, C. K. Svensson, Biochem. Pharmacol. 2005, 70, 275–286;
d) J. S. Yadav, B. V. S. Reddy, P. Sreedhar, Adv. Synth. Catal. 2003, 345,
Steady-state kinetic studies: Nitroreductase activity toward
2
4DNT and 24DNAN for purified NfsB and its mutants was deter-
mined by monitoring the initial rate of oxidation of NADH at
40 nm. All assays were performed in Tris·HCl buffer (20 mm,
3
pH 7.0) with NADH (60 mm) and substrate (varying concentrations),
initiated by addition of purified enzymes (50 nm). The reaction
temperature was 37Æ18C, and changes in absorbance were de-
tected every 15 s for 5 min (during linearity). All reactions were re-
plicated at least three times, and the data were fitted by non-linear
regression (Origin 8.5 software; OriginLab, Northampton, MA).
564–567.
[
2] a) C.-M. Ho, T.-C. Lau, New J. Chem. 2000, 24, 859–863; b) J. D. Spence,
A. E. Raymond, D. E. Norton, Tetrahedron Lett. 2003, 44, 849–851; c) H.
Takeuchi, J.-i. Tateiwa, S. Hata, K. Tsutsumi, Y. Osaki, Eur. J. Org. Chem.
2003, 3920–3922; d) V. Sridharan, K. Karthikeyan, S. Muthusubramanian,
Tetrahedron Lett. 2006, 47, 4221–4223; e) A. A. Lamar, K. M. Nicholas,
Tetrahedron 2009, 65, 3829–3833; f) Y. Wang, L. Ye, L. Zhang, Chem.
Commun. 2011, 47, 7815–7817; g) R. S. Srivastava, K. M. Nicholas, J. Am.
Chem. Soc. 1997, 119, 3302–3310.
Crystallization, data collection, and processing: Crystals of the
À1
T41L/N71S/F124W mutant (10 mgmL ) were grown in sodium
acetate (100 mm, pH 4.6) containing poly(ethylene glycol) 3350
[3] a) S. Ung, A. Falgui›res, A. Guy, C. Ferroud, Tetrahedron Lett. 2005, 46,
5913–5917; b) Q. X. Shi, R. W. Lu, K. Jin, Z. X. Zhang, D. F. Zhao, Chem.
Lett. 2006, 35, 226–227; c) A. B. Gamble, J. Garner, C. P. Gordon, S. M. J.
O’Conner, P. A. Keller, Synth. Commun. 2007, 37, 2777–2786; d) S. Liu, Y.
Wang, J. Jiang, Z. Jin, Green Chem. 2009, 11, 1397–1400; e) S. Liu, Y.
Wang, X. Yang, J. Jiang, Res. Chem. Intermed. 2012, 38, 2471–2478; f) M.
Bartra, P. Romea, F. Urpí, J. Vilarrasa, Tetrahedron 1990, 46, 587–594.
(
16%, w/v; Sigma—Aldrich) and Tacsimate (pH 4.0, 2%, v/v; Hamp-
ton Research Corp.) at 48C by the hanging-drop vapor-diffusion
method. Crystals, which appeared within two weeks, were rinsed
in reservoir solution containing glycerol (25%, v/v) as a cryoprotec-
tant for several seconds, and flash-frozen in liquid nitrogen. Diffrac-
tion data were collected at beamline BL-17U (Shanghai Synchro-
tron Radiation Facility, Shanghai, China) and processed using
[
4] a) Y. Takenaka, T. Kiyosu, J.-C. Choi, T. Sakakuraa, H. Yasuda, Green Chem.
009, 11, 1385–1390; b) Z. Rong, W. Du, Y. Wang, L. Lu, Chem.
Commun. 2010, 46, 1559–1561; c) A. K. Shil, P. Das, Green Chem. 2013,
5, 3421–3428.
2
[25]
HKL2000. Full data collection statistics are listed in Table 2.
1
Structure determination and refinement: The structure of the
[
5] a) G. Neri, G. Rizzo, C. Milone, S. Galvagno, M. G. Musolino, G. Capannel-
li, Appl. Catal. A 2003, 249, 303–311; b) Y. Takenaka, T. Kiyosu, G. Mori,
J.-C. Choi, T. Sakakura, H. Yasuda, Catal. Today 2011, 164, 580–584; c) D.
Beaudoin, J. D. Wuest, Tetrahedron Lett. 2011, 52, 2221–2223; d) Y. Tak-
enaka, T. Kiyosu, J.-C. Choi, T. Sakakura, H. Yasuda, ChemSusChem 2010,
T41L/N71S/F124W mutant was solved by molecular replacement
[
26]
with Phaser by using the structure of wild-type NfsB (PDB ID:
[
18]
1
DS7) as the starting structure. Further refinement and model
[27]
[28]
adjustment were performed with PHENIX and Coot programs,
respectively (structure-refinement statistics in Table 2). The coordi-
nates and structure factors of the T41L/N71S/F124W mutant have
been deposited in the Protein Data Bank (PDB ID: 3X21).
3, 1166–1168.
[
[
6] F. Li, J. Cui, X. Qian, R. Zhang, Chem. Commun. 2004, 2338–2339.
7] F. Li, J. Cui, X. Qian, R. Zhang, Y. Xiao, Chem. Commun. 2005, 1901–
1
903.
8] H.-H. Nguyen-Tran, G.-W. Zheng, X.-H. Qian, J.-H. Xu, Chem. Commun.
014, 50, 2861–2864.
9] S. O. Vass, D. Jarrom, W. R. Wilson, E. I. Hyde, P. F. Searle, Br. J. Cancer
009, 100, 1903–1911.
10] E. Johansson, G. N. Parkinson, W. A. Denny, S. Neidle, J. Med. Chem.
003, 46, 4009–4020.
[
[
Molecular docking: To elucidate the molecular basis of the regio-
selectivity of the wild-type and T41L/N71S/F124W mutant, we
docked 24DNT and 24DNAN to the active pockets of wild-type
NfsB (PDB ID: 1DS7) and the T41L/N71S/F124W mutant (PDB ID:
2
2
[
[
[29]
3
X21) in AutoDock 4.2. The GridBox parameters for docking the
2
substrates to enzymes were determined (grid center coordinates:
x=2.103, y=À14.394, z=0.138; size coordinates: x=28, y=28,
z=28). The docked structures were minimized by using a web
11] G. A. Prosser, J. N. Copp, S. P. Syddall, E. M. Williams, J. B. Smaill, W. R.
Wilson, A. V. Patterson, D. F. Ackerley, Biochem. Pharmacol. 2010, 79,
678–687.
ChemBioChem 2015, 16, 1219 – 1225
www.chembiochem.org
1224
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
[
12] G. M. Anlezark, T. Vaughan, E. Fashola-Stone, N. P. Michael, H. Murdoch,
[21] M. T. Reetz, J. D. Carballeira, Nat. Protoc. 2007, 2, 891–903.
[22] P. R. Race, A. L. Lovering, R. M. Green, A. Ossor, S. A. White, P. F. Searle,
C. J. Wrighton, E. I. Hyde, J. Biol. Chem. 2005, 280, 13256–13264.
[23] http://www.pymol.org.
[24] P. D. Fiorella, J. C. Spain, Appl. Environ. Microbiol. 1997, 63, 2007–2015.
[25] Z. Otwinowski, W. Minor, Methods Enzymol. 1997, 276, 307–326.
[26] A. J. McCoy, R. W. Grosse-Kunstleve, P. D. Adams, M. D. Winn, L. C. Storo-
ni, R. J. Read, J. Appl. Crystallogr. 2007, 40, 658–674.
[27] P. D. Adams, P. V. Afonine, G. Bunkóczi, V. B. Chen, I. W. Davis, N. Echols,
J. J. Headd, L.-W. Hung, G. J. Kapral, R. W. Grosse-Kunstleve, A. J. McCoy,
N. W. Moriarty, R. Oeffner, R. J. Read, D. C. Richardson, J. S. Richardson,
T. C. Terwilliger, P. H. Zwart, Acta Crystallogr. Sect. D 2010, 66, 213–221.
[28] P. Emsley, K. Cowtan, Acta Crystallogr. Sect. D 2004, 60, 2126–2132.
[29] G. M. Morris, R. Huey, W. Lindstrom, M. F. Sanner, R. K. Belew, D. S. Good-
sell, A. J. Olson, J. Comput. Chem. 2009, 30, 2785–2791.
M. A. Sims, S. Stubbs, S. Wigley, N. P. Minton, Microbiology 2002, 148,
297–306.
[
[
[
[
[
[
[
[
13] C. D. Emptage, R. J. Knox, M. J. Danson, D. W. Hough, Biochem. Pharma-
col. 2009, 77, 21–29.
14] H. GuillØn, J. A. Curiel, J. M. Landete, R. MuÇoz, T. Herraiz, J. Agric. Food
Chem. 2009, 57, 10457–10465.
15] D. Jarrom, M. Jaberipour, C. P. Guise, S. Daff II, S. A. White, P. F. Searle,
E. I. Hyde, Biochemistry 2009, 48, 7665–7672.
16] P. R. Race, A. L. Lovering, S. A. White, J. I. Grove, P. F. Searle, C. W. Wright-
on, E. I. Hyde, J. Mol. Biol. 2007, 368, 481–492.
17] A. L. Lovering, E. I. Hyde, P. F. Searle, S. A. White, J. Mol. Biol. 2001, 309,
2
03–213.
18] G. N. Parkinson, J. V. Skelly, S. Neidle, J. Med. Chem. 2000, 43, 3624–
631.
19] S. Zenno, H. Koike, M. Tanokura, K. Saigo, J. Biochem. 1996, 120, 736–
44.
20] a) J. I. Grove, A. L. Lovering, C. Guise, P. R. Race, C. J. Wrighton, S. A.
White, E. I. Hyde, P. F. Searle, Cancer Res. 2003, 63, 5532–5537; b) S.-W.
LinWu, C.-A. Wu, F.-C. Peng, A. H.-J. Wang, Biochem. Pharmacol. 2012,
3
[30] A. C. Wallace, R. A. Laskowski, J. M. Thornton, Protein Eng. 1995, 8, 127–
134.
7
Manuscript received: February 9, 2015
Final article published: April 27, 2015
83, 1690–1699.
ChemBioChem 2015, 16, 1219 – 1225
www.chembiochem.org
1225
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim