Crystal structure of yeast xylose reductase in complex with a novel NADP-DTT adduct provides insights into substrate recognition and catalysis

Article abstract of DOI:10.1111/febs.14667

Aldose reductases (ARs) belonging to the aldo-keto reductase (AKR) superfamily catalyze the conversion of carbonyl substrates into their respective alcohols. Here we report the crystal structures of the yeast Debaryomyces nepalensis xylose reductase (DnXR, AKR2B10) in the apo form and as a ternary complex with a novel NADP-DTT adduct. Xylose reductase, a key enzyme in the conversion of xylose to xylitol, has several industrial applications. The enzyme displayed the highest catalytic efficiency for l-threose (138 ± 7 mm^?1·s^?1) followed by d-erythrose (30 ± 3 mm^?1·s^?1). The crystal structure of the complex reveals a covalent linkage between the C4N atom of the nicotinamide ring of the cosubstrate and the S1 sulfur atom of DTT and provides the first structural evidence for a protein mediated NADP–low-molecular-mass thiol adduct. We hypothesize that the formation of the adduct is facilitated by an in-crystallo Michael addition of the DTT thiolate to the specific conformation of bound NADPH in the active site of DnXR. The interactions between DTT, a four-carbon sugar alcohol analog, and the enzyme are representative of a near-cognate product ternary complex and provide significant insights into the structural basis of aldose binding and specificity and the catalytic mechanism of ARs. Database: Structural data are available in the PDB under the accession numbers 5ZCI and 5ZCM.

Full text of DOI:10.1111/febs.14667

Received Date : 11-Jun-2018
Revised Date : 13-Sep-2018
Accepted Date : 26-Sep-2018
Crystal structure of yeast xylose reductase in complex with a novel NADP-
DTT adduct provides insights into substrate recognition and catalysis
Bhaskar Paidimuddala1, Samar Ballabha Mohapatra2, Sathyanarayana N. Gummadi1*,
Narayanan Manoj2*
1Applied and Industrial Microbiology Laboratory, Department of Biotechnology, Bhupat and
Jyoti Mehta School of Biosciences, Indian Institute of Technology Madras, Chennai 600036;
2Department of Biotechnology, Bhupat and Jyoti Mehta School of Biosciences, Indian
Institute of Technology Madras, Chennai 600036
*Correspondence: Sathyanarayana N. Gummadi (gummadi@iitm.ac.in) or Narayanan
Manoj (nmanoj@iitm.ac.in)
Article type : Original Articles
Abbreviations
AR, Aldose reductase; AKR, Aldo-keto reductase; CtXR, Candida tenuis xylose reductase;
DnXR, Debaryomyces nepalensis xylose reductase; DTT, Dithiothreitol; hAR, Human aldose
reductase; XR, Xylose reductase
This article has been accepted for publication and undergone full peer review but has not
been through the copyediting, typesetting, pagination and proofreading process, which may
lead to differences between this version and the Version of Record. Please cite this article as
doi: 10.1111/febs.14667
This article is protected by copyright. All rights reserved.

Keywords: Xylose reductase, AKR superfamily, Debaryomyces nepalensis, crystal structure
complex, NADP-DTT adduct, product analog
Running title: Xylose reductase in complex with NADP-DTT adduct
Database: Structural data are available in the PDB under the accession numbers 5ZCI and
5ZCM.
Abstract
Aldose reductases (ARs) belonging to the aldo-keto reductase (AKR) superfamily catalyze
the conversion of carbonyl substrates into their respective alcohols. Here we report the
crystal structures of the yeast Debaryomyces nepalensis xylose reductase (DnXR,
AKR2B10) in the apo form and as a ternary complex with a novel NADP-DTT adduct.
Xylose reductase, a key enzyme in the conversion of xylose to xylitol has several industrial
applications. The enzyme displayed the highest catalytic efficiency for L-threose (138±7
mM-1s-1) followed by D-erythrose (30±3 mM-1s-1). The crystal structure of the complex
reveals a covalent linkage between the C4N atom of the nicotinamide ring of the cosubstrate
and the S1 sulfur atom of DTT and provides the first structural evidence for a protein
mediated NADP–low-molecular-mass thiol adduct. We hypothesize that the formation of the
adduct is facilitated by an in crystallo Michael addition of the DTT thiolate to the specific
conformation of bound NADPH in the active site of DnXR. The interactions between DTT, a
four carbon sugar alcohol analog, and the enzyme are representative of a near cognate
product ternary complex and provide significant insights into the structural basis of aldose
binding and specificity and the catalytic mechanism of ARs.
This article is protected by copyright. All rights reserved.

Introduction
Aldose reductases (ARs) (EC. 1.1.1.21) are members of the aldo-keto reductase (AKR)
superfamily that contains homologs present across all phyla. AKRs catalyze the conversion
of a wide variety of carbonyl substrates to their corresponding alcohols with concomitant
oxidation of NAD(P)H [1, 2]. The superfamily has been classified into sixteen families in the
AKR superfamily database [1]. In this scheme, the ARs are present in the AKR 1, 2, 3 and 4
families. Human ARs (hARs) belonging to the AKR1 family are implicated in diabetic
complications, fatty liver, chronic kidney diseases and the oxidative stress-induced activation
of NF-kB and AP1 signals [1, 3-6]. Fungal ARs belong to the AKR2 and AKR3 families and
are inducible enzymes with important roles in growth and metabolism [1, 7]. The AKR2
family xylose reductase (XR) (EC. 1.1.1.307) is the first enzyme in the xylose utilization
pathway and can be exploited for the biosynthetic production of polyols such as xylitol,
arabitol and erythritol for applications in food and pharmaceutical industries [7-9]. The ARs
of the AKR3 family are plant erythrose reductases involved in the regulation of plant abiotic
stress defense and share similar substrate preferences of the homologous AKR2 members [1,
10-12, 13, 14].
The superfamily members share high levels of sequence conservation with the human,
plant and fungal ARs displaying nearly 40% sequence identity. The mostly conserved
residues involved in catalysis and cosubstrate binding suggest a common mode of catalytic
mechanism [1]. Numerous crystal structures of ARs from the AKR1 family have been
determined owing to their therapeutic importance for inhibitor development. These include
the well characterized Homo sapiens (hARs; PDB: 1ADS, 3V36, 2IKI, 1US0) [2, 15-17], Sus
scrofa (SsAR; PDB: 1DLA, 1AH0) [18, 19], Rattus norvegicus (RnAR; PDB: 3O3R, 3QKZ)
[20, 21] and Giardia lamblia (GlAR; PDB: 3KRB) homologs [22]. The AKR fold is a
conserved (β/α)8 barrel with the active site located in a large, deep elliptical pocket at the C-
This article is protected by copyright. All rights reserved.

terminal end of the barrel and a bound NADPH in an extended conformation [2, 23]. Another
structural feature common to all AKRs is the large conformational change of a ‘safety belt’
element that acts to sequester the cosubstrate from solvent [23]. The enzyme follows an
ordered bi-bi catalytic mechanism in which NADPH binds first followed by the carbonyl
substrate and the conformational changes, primarily of the ‘safety belt’ loops, regulate the
uptake and release of the cosubstrate [24, 25]. The structural basis of substrate recognition
and specificity in this superfamily primarily is derived from studies of inhibitor bound
complexes of AKR1 members [19, 26].
The fungal ARs (AKR2 and 3 families) have been poorly characterized with respect
to their structure-function properties despite the potential for use in biotechnological
applications. To date, the XR (AKR2B5) from Candida tenuis (CtXR; PDB: 1JEZ, 1K8C)
represents the sole instance of a structurally and biochemically well characterized fungal AR
[27]. It displays dual cosubstrate specificity to utilize both NADPH and NADH to convert
xylose into xylitol [27-32]. The safety belt loop is disordered in the apo form of CtXR and
also in the wild type apo forms of most reported AR structures [18, 27, 33-34]. The specific
transfer of the 4-pro-R hydride to the carbonyl carbon atom of the substrate is known to be
highly conserved in this superfamily. However, despite extensive structural studies (69
inhibitor bound AKR1 structures in the PDB), it is intriguing that only two structures of a
substrate bound ternary complex of an AKR superfamily member have been reported,
namely, hAR bound to D-glyceraldehyde (PDB: 3V36) and to glucose-6-phosphate (PDB:
2ACQ) [15, 35].
Here we present the crystal structures of the Debaryomyces nepalensis XR (DnXR)
in the apo form and as a ternary complex with cosubstrate (NADPH) and a product analog,
dithiothreitol (DTT). D. nepalensis, a non-pathogenic saccharomycetes yeast can utilize both
hexose and pentose sugars to produce polyols [36]. DnXR, a key metabolic enzyme in the D-
This article is protected by copyright. All rights reserved.

xylose utilization pathway belongs to the AKR2 family (AKR2B10) and has previously been
shown to have strict dependence on NADPH and displays activity for a range of aldehydes
[37]. In this study, an ordered conformation of the safety belt loop was observed in the
absence of the cosubstrate and provides a bonafide depiction of the apo form of an AR. The
structure also showed a novel covalent adduct between the nicotinamide ring and DTT
(NADP-DTT), a feature hitherto unobserved in a protein mediated complex. The first
structure of an AKR as a ternary complex with a four carbon sugar alcohol analog provides
new insights into substrate binding and specificity and the catalytic mechanism of AKRs.
The results of this study may help design and develop engineered yeast ARs to fine tune both
substrate and cosubstrate selectivity for biotechnological applications.
Results
Enzyme activity, kinetics and inhibition
Recombinant DnXR was overexpressed in E. coli Rosetta and purified to homogeneity by Ni-
NTA affinity chromatography as described previously [37]. The protein is composed of 320
amino acid residues and exists as a dimer in solution as indicated by size-exclusion
chromatography (SEC) studies. DnXR was earlier shown to exhibit broad substrate
specificity, with the highest catalytic efficiency for C5 sugars like arabinose, xylose and
ribose and a strict preference for cosubstrate NADPH [37]. In this study, DnXR was tested
for activity and kinetic parameters using two C4 sugars, namely, D-erythrose and its
diastereoisomer, L-threose. Interestingly, DnXR showed higher activity and catalytic
efficiency (20 - 92-fold) for the four carbon sugars in comparison to the five carbon sugars.
The enzyme displayed the highest catalytic efficiency for L-threose (138±7 mM-1s-1)
followed by D-erythrose (30±3 mM-1s-1) and is similar to the pattern of activity displayed by
This article is protected by copyright. All rights reserved.

CtXR [38]. Furthermore, the effect of the product erythritol and its dithio analog DL-DTT, on
DnXR activity were examined. Both alcohols inhibited the catalytic activity in a
concentration dependent manner. In order to identify the type of inhibition mechanism, the
kinetics of DnXR were performed using different concentrations of erythritol/DTT using
erythrose as a substrate. Enzyme kinetics were performed using erythritol in the range 500-
1100 mM and showed an increase in KM from 6.5 ± 0.4 to 8.6 ± 0.5 mM, while the Vmax
decreased from 176 ± 3 to 124 ± 3 U/mg (Table 1). The data suggests that erythritol inhibits
DnXR via mixed inhibition mechanism (Fig. 1A, B). Interestingly, DTT displayed a similar
behavior in affecting the kinetic parameters (Table 1) (Fig. 1C, D). Although the kinetic
studies suggest a mixed inhibition mechanism for both erythritol and DTT, the latter is a far
stronger inhibitor with an KI value of 1.8 mM in comparison to erythritol (KI = 1577 mM)
(Table 1).
Overall structures of apo DnXR and the ternary complex with cosubstrate and DTT
The crystal structure of apo DnXR was determined at a resolution of 2.0 Å in the space group
P212121 and the structure solved by molecular replacement using one subunit of apo form of
CtXR. The tertiary structure contains the characteristic AKR superfamily (β/α)8 barrel fold
(Fig. 2A) [1, 23]. The asymmetric unit contains two subunits related by a non-
crystallographic 2-fold axis with the dimeric arrangement identical to that in CtXR (Fig. 2B).
The root mean square deviation (rmsd) between the subunits of DnXR and CtXR is 0.9 Å,
over 298 C atoms. The crystal structure of the complex of DnXR bound to the cosubstrate
and D-DTT was determined at a resolution of 1.7 Å in the space group C2221 (Fig. 3). The
asymmetric unit in the complex contains one subunit. However, the crystal packing
arrangement reveals an identical arrangement of the expected dimer, with the subunits related
This article is protected by copyright. All rights reserved.

by a crystallographic 2-fold axis in this case. The subunit model in the complex contains
ordered electron density for an additional 9 residues from the N-terminal His6-tag.
The tertiary structure with an eight stranded barrel (1 to 8) is flanked by eight
helices (1 to 8) (Fig. 3A). A -hairpin caps the N-terminal end of the barrel while the C-
terminal end contains the conserved active site groove. The extended C-terminal region
containing two helices reaches over the active site with the terminal loop that is stabilized by
interactions with residues from loop 4. The overall quaternary, tertiary and secondary
structures are largely identical between the two forms except for the cosubstrate binding
regions (see below). The independent subunits in the apo form are virtually identical with a
rmsd of 0.46 Å, over 315 C atoms, whereas the rmsd between the subunit in the apo form
and that of the complex is 2.1 Å. The rmsd between the apo enzyme dimer and the symmetry-
related dimer in the complex is 0.75 Å. A summary of the data collection, refinement and
validation statistics is given in Table 2.
Binding of NADP-DTT covalent adduct
A well conserved feature of AKRs is the presence of the cosubstrate binding groove that
spans the region from the center of the barrel and extends to the solvent exposed edge lined
by helix 8. (Figs. 3A and 3B) [2, 27]. Electron density (higher than 3σ) for the cosubstrate
and the bound ligand D-DTT can be clearly seen in the unbiased omit Fo - Fc map and in
simulated annealing omit maps of this region for the ternary complex. Examination of the
maps showed an unexpected feature suggestive of a carbon-sulfur covalent bond between the
C4N atom of the nicotinamide ring and the sulfur atom (S1) of DTT in the linear form
(distance of ~1.8 Å) (Fig. 4A). Hence, a nicotinamide-DTT adduct was inferred and modeled
and refinement carried out by introducing a covalent link between the C4N and S1 atoms.
There was no evidence for residual difference electron density after the adduct model was
This article is protected by copyright. All rights reserved.

refined with full occupancy. In the ternary complex, the average B-factors for the all non-
hydrogen protein atoms (3011 atoms) is 15.7 Å2, while that of all atoms in the NADP-DTT
adduct is 16.7 Å2, indicating a well ordered conformation of the bound adduct.
The nicotinamide ring is in the boat-like conformation (anti-antiperiplanar) and the
attached ribose ring in the C2'-exo conformation. The deviation of the positions of the N1 and
C4 atoms of the nicotinamide ring from planarity can be described by the angle (αN) between
the planes N1N-C2N-C6N and C2N-C3N-C6N and the angle (αC) between the planes C2N-
C3N-C6N and C3N-C4N-C5N, respectively [39]. The values of αN = 160° and αC = 156°
indicate significant departures of the nicotinamide ring from planarity (Figs. 4B–D). The
resultant configuration of the C4N atom is sp3 like, resembling the reduced NADPH form.
The carboxamide group is almost in plane with respect to the pyridine ring and maintains a
dihedral angle (C2N-C3N-C7N-N7N) of 3˚. The S1 atom of the adduct occupies the
structurally equivalent position of the HR atom on the A-side (4-pro-R) face of the cosubstrate
(Fig. 4D).
The cosubstrate adopts an extended anti conformation with only one intramolecular
hydrogen bond present between atoms O3D of ribose and O2N of pyrophosphate. A large
number of specific non-bonded interactions are involved in the binding. Among, these, the
side chain of Tyr212 stacks with the nicotinamide ring and hydrogen bond interactions
between the side chains of Ser164, Asn165 and Gln186 with the carboxamide group orient
the A-side of the nicotinamide group towards the substrate binding cavity. The nicotinamide-
ribose and the pyrophosphate are mostly buried by the residues of loop 7 and loop 1, whereas
the adenosine and the associated ribose 2'-phosphate are located at the distal solvent exposed
edge of the groove formed between helices 7 and 8 (Fig. 3).
This article is protected by copyright. All rights reserved.

Extensive attempts to obtain crystals of cognate substrate/product bound ternary
complexes with tetrose, pentose and hexose sugar/sugar alcohols using co-crystallization and
soaking techniques proved unsuccessful. Crystals containing the NADP-DTT adduct were
obtained in a co-crystallization experiment condition which contained 10 mM racemic DL-
DTT, in addition to the cosubstrate. It is noteworthy that the D-isomer of DTT is
preferentially bound at the putative substrate binding pocket. Examination of the crystal
packing interactions of the ternary complex showed that the substrate binding pocket is
occluded from bulk solvent by crystal packing interactions from a neighboring symmetry
related molecule. The S4 atom of DTT makes multiple van der Waals interactions (< 4.0 Å)
with two residues (Arg-5 and Pro-6) of the N-terminal His6-tag of a neighboring subunit,
locking the ligand in place (Fig. 5A). However, in the apo form crystal packing, the substrate
binding pocket is accessible to the bulk solvent in both subunits. Furthermore, differences in
the crystal packing interactions have not induced any conformational changes of residues
lining the substrate binding site. Apparently, a Michael addition of the DTT thiolate to the
double bond of the α,β-unsaturated nicotinamide is favored and resulted in a derivative under
these conditions (Fig. 5B).
Covalent NADP-thiol adducts have previously been reported in the structures of
oxido-reductases. However, all these instances are adducts between the pyridine and a Cys
residue from the protein (Fig. 5C–E) [40-42]. To the best of our knowledge, this is the first
structural evidence of a protein mediated NADP - low-molecular-mass thiol adduct bound to
a protein active site. Based on earlier reports of NADP-thiol adducts being identified in
solution, we attempted to detect the formation of the adduct in the presence and absence of
the enzyme using spectrophotometric measurements under conditions used for the
crystallization [41-43]. However, the expected increase in absorbance around 340 nm,
This article is protected by copyright. All rights reserved.

indicating the formation of an adduct was not observed in solution (data not shown). In our
case it is probably a result of insufficient thiol oxidation of DTT compared to that of cysteine,
at pH 7.5.
Comparison of apo form and ternary complex of DnXR
In the reported structures of apo forms of wild type ARs, the loop 7 is disordered and
consequently, description of conformational changes upon cosubstrate binding is incomplete
in the ARs [18, 27, 33]. DnXR undergoes significant conformational changes upon formation
of the complex with the adduct. The rearrangements and conformational changes results in
burying 1011 Å2 out of 1200 Å2 the total accessible surface area of the adduct. Loop 7
(residues 213-237), together with loop 1 constitutes the ‘safety belt’ that sequesters the
cosubstrate from the bulk solvent (Fig. 3). Cosubstrate binding causes the side chain of Trp19
of loop 1 to move out (by ~2.1 Å) of the binding pocket to accommodate the nicotinamide
ring and the pyrophosphate. The largest change is in the conformation of loop 7 that shows a
rmsd value of 3.96 Å for the C atoms of the region containing residues 218-231 (Fig. 6A).
The Oγ atom of Ser219 in loop 7 shifts as much as 8.4 Å towards the cosubstrate to interact
with the O1N atom of pyrophosphate (Fig. 6B). These changes place the nicotinamide ring
deep in the center of the barrel close to active site (Fig. 3B).
Another major conformational change associated with cosubstrate binding involves
the region containing loop 8 and helix 8. In the complex, the helix undergoes a rotation of
~6˚, towards the cosubstrate, clamping it and allowing for the formation of non-bonded
interactions between residues 269-279 (rmsd of ~1.0 Å for C atoms) and the ribose 2'-
phosphate and adenosine groups (Fig. 6A). This is accompanied by several side-chain
rearrangements as well, resulting in the key interactions that enable the binding of the ribose
2'-phosphate via two hydrogen bonds with Oγ atom of Ser270 and the main chain amide
This article is protected by copyright. All rights reserved.

group from residue Asn271 and two ionic interactions with the side chains of Lys269 and
Arg275. The side chain of the highly conserved Arg275 is however disordered in the apo
form and appears to be stabilized on binding of the cosubstrate (Fig. 6B). These interactions
appear to be primarily responsible for the specificity of the enzyme for NADPH over NADH.
Finally, the side chain of Asn279 shifts by ~2.2 Å to establish bidentate hydrogen bond
interactions with the N6A and N7A atoms of adenosine.
Substrate binding site and catalytic mechanism
The catalytic site architecture of AKRs includes a highly conserved tetrad of residues Asp42,
Tyr47, Lys76, and His109 (DnXR numbering) lining the bottom of a deep open cavity.
Superposition of DnXR on the hAR ternary complexes (PDBs: 3V36, 2ACQ) [15, 35]
reflects the high degree of structural conservation of the tetrad across the families (Fig. 7).
The shape of the open substrate binding pocket in AKRs is defined by loop 2, loop 4 and the
C-terminal loop. To the best of our knowledge, the structures of hAR•NADP+ with D-
glyceraldehyde and glucose-6-phosphate represent the only examples of substrate bound
ternary complexes of an AKR superfamily member [15, 35]. Although Zheng et al., have not
described the enzyme-glyceraldehyde interactions in detail, the structure shows that the
acceptor carbonyl oxygen atom is 3.1 Å from the donor C4N atom of NADP+ and 2.9 Å from
the OH group of Tyr48, a catalytic residue [15]. However, in the complex with glucose-6-
phosphate, the substrate is bound in a “backward” and unproductive cyclic sugar
conformation with the phosphate group occupying the position equivalent to the carbonyl
group of an aldehyde substrate [35]. The equivalent distances of the O1P atom are 3.3 Å and
2.9 Å from the C4N atom and the OH group of Tyr48, respectively (Fig. 7).
This article is protected by copyright. All rights reserved.

In the absence of relevant complexes, numerous studies of AKRs have utilized
molecular modeling of substrates into the putative active sites of holo forms to establish the
enzyme-substrate interactions and explain key features, including the enzyme mechanism,
substrate specificity and active site plasticity of the AKRs [27, 44-49]. In the DnXR
complex, DTT is bound in the substrate binding pocket in an orientation expected for a four-
carbon open chain sugar alcohol, reminiscent of a product bound form (Fig. 7). Apparently,
DTT, a dithio analog of erythritol, can occupy the substrate binding pocket since it is a
product inhibitor of the forward reaction of D-erythrose reduction. The configuration of the
DTT C1 atom (equivalent of the carbonyl carbon of the substrate) is tetrahedral sp3 like and is
present on the C4N-pro-R side of the cosubstrate. Superposition of the DnXR complex and
the two hAR complexes shows that the ligands occupy the same location in the substrate
binding site (Fig. 7). The positions of the glyceraldehyde carbonyl oxygen atom, the
phosphate O1P atom and the equivalent S1 atom of DTT are within 0.6 Å. The S1 atom of
DTT is within hydrogen bonding distance to the conserved catalytic residues, namely, OH
group of Tyr47 (3.2 Å) and Nε2 atom of His109 (3.5 Å). In the hAR- glyceraldehyde
complex, the corresponding distances from the carbonyl oxygen atom are 2.9 Å and 3.0 Å,
respectively, while the corresponding distances of the O1P atom in the hAR-glucose-6-
phosphate complex are 2.9 Å and 3.4 Å, respectively. Thus, it is reasonable to suppose that
the E•NADP-DTT complex obtained here represents the geometry of the enzyme - near-
cognate sugar alcohol product complex, allowing for a precise description of the substrate
binding site and the putative interactions in an AKR-four carbon linear sugar substrate
complex for the first time.
The DTT moiety maintains an extended conformation (2R, 3R configuration) with the
S4 thiol group exposed to the bulk solvent (Figs. 8A and 8B). The S1 atom occupies the
position equivalent to the carbonyl oxygen of the substrate at the bottom of the pocket. The
This article is protected by copyright. All rights reserved.

binding mode of DTT is primarily determined by two specific hydrogen bond interactions
anchoring the ligand hydroxyls on either side, namely, between Nδ2 atom of Asn305 and the
C2 hydroxyl group (3.0 Å) and between the Oδ2 atom of Asp46 (2.7 Å) and the C3 hydroxyl
group. Additionally, DTT makes multiple van der Waals contacts with the side chain groups
of non-polar residues that line the active site cavity. These involve the C1 atom with Trp19
(3.5 Å), the C4 atom with the Trp78 (3.7 Å) and the S4 thiol group with Phe127 (3.9 Å) and
Trp310 (3.8 Å). The substrate binding site remains mostly unchanged upon ligand binding
suggesting the existence of a rigid and preformed cavity for sugar substrates (Fig. 8C). An
induced rearrangement occurs in the side chain of Asp46 where the Oδ2 atom moves towards
the C3 hydroxyl group by ~0.8 Å to form a hydrogen bond. The slight rearrangement
observed in the substrate binding pocket involving the side chain of Trp19 of loop 1 is
induced by cosubstrate binding to the protein (Fig. 8C). This rearrangement results in
significant van der Waals contacts between the Trp residue and the C1 atom (equivalent of
the carbonyl carbon of the substrate) and has been observed in the cosubstrate bound holo
forms of hAR, pig AR, CtXR and plant AR [2, 18, 27, 33]. Furthermore, conformations of
several conserved residues lining this pocket are nearly identical between the yeast, human
and plant AR homologs (Fig. 8D).
The reaction mechanism and the roles of the catalytic tetrad in the AKRs (Asp42,
Tyr47, Lys76, His109) have been dissected in detail using crystal structures of numerous apo
and holo forms in combination with biochemical and kinetic data from substitution mutants
and computational studies involving docking and QM/MM calculations [1, 2, 23, 27, 31, 44-
53]. The observed geometry of non-bonded interactions between the enzyme and the adduct
is consistent with the general mechanism for erythrose reduction by DnXR (Fig. 9A). The
C4N group of the nicotinamide moiety in the reduced puckered geometry is poised for the
stereospecific hydride transfer to the re side of the C1 atom of the ligand, the distance being
This article is protected by copyright. All rights reserved.

2.7 Å. The typical donor-acceptor hydride transfer distances in similar enzymes, for example
in alcohol dehydrogenases, are in the range of 2.8 - 3.9 Å [51, 52]. The phenolic oxygen of
Tyr47 is expected to have a lowered pKa as a result of the hydrogen bond to Lys76 amino
group (3.1 Å), which in turn maintains a salt bridge with the Oδ2 of Asp42 (2.5 Å). The
reprotonation of Tyr47 is expected to be assisted by a proton relay involving His109 and to
the bulk solvent, in turn [1, 23, 50].
Discussion
Fungal XRs, in general, have been poorly studied with respect to their detailed structure-
function properties. In particular, the structural basis of the broad specificity for in vitro
substrates that include aldoses of varying lengths and a range of non-sugar substituent groups
at the non-reacting end of the aldehyde is not clear. Furthermore, the ability of the enzyme to
select the physiological substrate xylose in the promiscuous substrate binding pocket has not
been established primarily due to the lack of structural data from an appropriate ternary
complex. This report describes the structures of DnXR in the apo form and as a complex of
the holo form with DTT, a dithio C4 sugar alcohol analog, and shows how recognition and
selectivity for aldoses is achieved in the enzyme.
Formation of NADP-DTT adduct
Attempts to generate ternary complexes of AKRs with cognate substrates/products appear to
have been generally unsuccessful. A likely explanation could be the presence of an
electrostatic barrier created by bound NADP+, a substrate inhibitor that may form during the
crystal growth phase preventing the entry of neutral substrates into the active site [54].
Alternatively, the open substrate binding cavity is lined by multiple non-polar residues and
hence the binding energy provided by these secondary hydrophobic interactions under
This article is protected by copyright. All rights reserved.

crystallization conditions may not be favorable for the formation of complexes with polar
substrates like sugars. The structure presented here is the first structure of a holo AKR in its
active conformation with the substrate binding site occupied by a tetrose sugar alcohol
analog. This complex appears to be a result of a serendipitous in crystallo formation of a
covalent NADP-DTT adduct. Under alkaline conditions, NADP+ is known to participate in
nucleophilic addition reactions with sulfhydryl compounds like sulfides, cysteines and a
range of mercaptans, including dithiols to form 1,2- or 1, 4- adducts with the nicotinamide
moiety [43]. Structural evidence for the NADP-Cys adduct has been reported earlier in
Synechococcus succinic semialdehyde dehydrogenase (SySSADH), Rattus 10-
formyltetrahydrofolate dehydrogenase (RaFDH) and Pseudomonas aeruginosa betaine
aldehyde dehydrogenase (PaBADH) [40-42]. Although these complexes may result from
“charge-transfer” reactions, the ability to purify the adducts suggests an actual addition
reaction [40-42, 55].
The crystallographic evidence in DnXR for the NADP-DTT covalent linkage provides
the first demonstrated evidence for the formation of a protein mediated NADP - low-
molecular-mass thiol adduct. We propose a mechanism where the thiolate group of DTT
performs a nucleophilic attack on the electrophilic C4N atom of the oxidized nicotinamide,
producing the 1,4-nicotinamide adduct (Fig. 9B). Although NADPH and reduced DTT were
initially present in the crystallization cocktail, it is likely that coupled auto-oxidation has
occurred under the alkaline pH (HEPES pH 7.5) and aerobic nature of the crystallization
conditions, thus favoring the reaction [42, 56, 57]. We also propose a structural argument
where competitive inhibition by DTT, a thiol analog of the reaction product, helps retain an
appropriate orientation that favors the interaction of the C4N atom of the nicotinamide ring
and the thiolate. Finally, the crystal packing environment appears to be crucial to hamper the
cleavage of the adduct by an elimination reaction. While the adduct species may not be an
This article is protected by copyright. All rights reserved.

intermediate in the catalytic cycle of AKRs, given the widely prevalent practice of adding
DTT as a reducing agent for many biochemical experiments, the side effect of such adduct
formation may not always be negligible.
Cosubstrate binding and specificity in ARs
The mode of cosubstrate binding in the AKR superfamily is highly conserved with mostly
identical and structurally conserved residues and water molecules involved in the
interactions. Unlike in the apo forms of most other wild type ARs, loop 7 is present in an
ordered form in DnXR. Interestingly, in a hAR•NADP+ complex with citrate in the substrate
binding pocket, two discrete conformations corresponding to a fully closed and a ‘half-open’
state of loop 7 were observed. This transition induced by conformational changes in bound
citrate is representative of a situation where reorientation of the product could trigger the
release of the cosubstrate via its interactions with loop 7 [58]. Together, changes observed in
loop 7 corroborate the two-step kinetic mechanism of carbonyl group reduction in the AKR
superfamily where the release of NADP+ from the E•NADP+ complex is known to be the rate
limiting step [1, 23, 24]. Furthermore, it is likely that the divergence in the length and residue
composition of this loop across the superfamily is an evolutionary mechanism for the
modulation of the kinetic properties of its members [1, 23] (Fig. 10). For instance, in the plant
AKR4 family, specific interactions of a shortened loop 7 that stabilize the closed
conformation of the loop were held responsible for the lower cosubstrate affinity and activity
of ZmAR in comparison to hAR [33, 59].
Although the AKRs in general are specific for NADPH, the CtXR represents an
example of an AR that can utilize both NADPH and NADH [28, 30]. However, DnXR, the
closest homolog of CtXR, displays strict specificity for NADPH. The structural basis of the
dual specificity was previously explained for CtXR based on the structures of NADP+ and
This article is protected by copyright. All rights reserved.

NAD+ bound holo complexes [28]. The enzyme was shown to adopt different side chain
conformations of two residues, namely Glu227 and Asn276, (Glu222 and Asn271 in DnXR)
that present different hydrogen bonding interaction patterns in the presence and absence of
the ribose 2'-phosphate [28]. Comparison of the DnXR-NADP complex and the CtXR-
NADP+ complex shows that all residues and their interactions with the entire cosubstrate are
completely identical and structurally conserved between the two homologs. The only
difference is that of Asn272 (Leu in CtXR) lining the outer edge of the 2'-phosphate binding
cavity that does not make direct interactions with the phosphate (distances 4.2 - 4.9 Å) (Fig.
11) Together, these observations strongly suggest that the side chain interactions that were
earlier proposed to determine the dual specificity in the CtXR may not be the sole
determinants of cosubstrate specificity within the AKR2 family. Re-engineering of
cosubstrate specificity from NADPH to NADH is of particular importance for biocatalytic
applications due to the higher stability and lower cost of NADH [30, 60, 61]. Our analysis
here of the two almost identical cosubstrate binding sites reveals the limitations of
generalizing the structural basis of cosubstrate specificity based solely on comparisons of
crystal structures. Considering the pleotropic nature of residue changes, a systematic
mutagenesis strategy at multiple locations scanning different substitutions is imperative here.
For instance, in a similar effort in the 2,5-DKG aldo-keto reductase, only one out of 40
unique substitution mutants yielded a significant improvement in NADH utilization over
NADPH [62].
Structural basis of substrate recognition
The ARs accept open chain form of aldoses of varying lengths and different configurations
for the carbon atoms at the non-reacting end of the sugars [1, 23, 27, 31]. The outer edge of
the AR active site pocket is solvent exposed, wide and lined by side chains of non-polar
This article is protected by copyright. All rights reserved.

residues whereas the bottom constituting the catalytic site accommodating the carbonyl
carbon and the oxygen atoms, is far narrower (Fig. 8). It appears that access and activity for a
wide range of aldose substrates is facilitated by the open active site.
In the structure presented here, assuming a structural mimic of a bound prochiral
substrate, the significant van der Waals interactions that C1 atom makes with the side chain
of the Trp19 (6 interactions < 4.0 Å) suggests its important steric role in positioning the
carbonyl carbon. This residue is completely conserved across the AKR superfamily (Fig. 10).
As expected, the catalytic efficiency of a CtXR mutant where the equivalent Trp is
substituted by Phe/Tyr shows a decrease of 95% of the value for the wild type. This is
primarily due to a large increase in the KM (16-fold) while the kcat decreased by 1.5-fold [32].
The interaction between Trp and the substrate is a result of a conformational change
occurring in the holo form, subsequent to binding of the cosubstrate. This observation
supports the notion that it is indeed the cosubstrate binding that induces the formation of the
competent substrate binding pocket in the AKRs via the Trp residue.
The orientation of DTT places the C2(R) hydroxyl group within hydrogen bonding
distance with the Nδ2 of Asn305, strongly indicating that this interaction is the primary
determinant of the stereospecificity of the carbonyl reduction (Fig. 8). Asn305 is completely
conserved across diverse members of AKR family 2. The crucial role of Asn305 is also
supported by data from kinetic studies of CtXR (Asn309) which revealed that a substitution
with Ala caused serious disruption of activity and efficiency [32]. For instance, a 29-fold
reduction in catalytic efficiency is observed in the mutant for the reduction reaction of D-
galactose. Furthermore, the 30-fold preference in catalytic efficiency of CtXR for D-
galactose over 2-deoxy-D-galactose was completely lost in the mutant [32]. Thus, within the
AKR family 2, the hydrogen bond between the Asn residue and C2 hydroxyl group also
This article is protected by copyright. All rights reserved.

appears to be a determinant of the regioselectivity of the C2 hydroxyl group of aldose
substrates, irrespective of the number or the configuration of the carbon atoms at the non-
reacting part.
The C-3(R) hydroxyl group of DTT is hydrogen bonded to Oδ2 of Asp46 of DnXR
suggesting that this interaction is another key element for binding of polyhydroxylated
substrates (Fig. 8). Asp46 is the sole ionizable group in the DnXR directly interacting with
the non-reacting part of the substrate and is completely conserved within AKR family 2.
Mutation of the equivalent Asp in CtXR (Asp50) lowered the catalytic efficiency for the
reduction of D-xylose by ~1.2-fold [32]. Furthermore, the mutant also showed no change in
substrate specificity profile compared with the wild type when tested with other aldoses like
D-glucose and D-galactose suggesting that this residue may not be a strong structural
determinant for the regioselectivity of hydroxyls groups at carbon positions C3 and further.
Taken together, it is evident that the specific sequence features of residues at these two
positions may play crucial roles in determining the substrate specificity and selectivity
profiles of the AKR family 2.
Among the aldose substrates tested in this study, DnXR has the highest catalytic
efficiency for four carbon sugars compared to five or six carbon sugars. Examination of
residues that interact with the non-reacting end of the substrate indicate the presence of
hydrophobic residues (Trp19, Trp78, Phe110, Phe127 and Trp310 in DnXR) that primarily
contribute through van der Waals and hydrophobic interactions [63]. Given the funnel shaped
geometry of the active site and the non-specific nature of the interactions with the hydroxyl at
carbon position C4, it is reasonable to suppose that sugars longer than C4 may possess higher
flexibility at the non-reacting end, potentially resulting in fewer/less favorable interactions
and thereby affecting the optimal orientation of the substrate in the catalytic site (Fig. 8B).
This article is protected by copyright. All rights reserved.

Interestingly, comparison of the available structures of ARs from AKR families 1, 2
and 4 shows that the main chain positions of residues at the outer edge of the pocket are
structurally conserved whereas some side chains form divergent patterns (Fig. 8D). Trp19
and Trp78 are conserved in all ARs, whereas Phe110 is replaced by Trp in hAR and HvAR.
Phe127 in DnXR is conserved in hAR and is substituted by Met132 in HvAR. Trp310 in
DnXR is substituted by Leu301 and Phe301 in hAR and HvAR, respectively [15, 33] (Fig.
10). We propose that the precise surface geometry presented by variations localized to this
region of the substrate binding pocket provides a structural basis for the extended substrate
specificity in the ARs of the superfamily and a similar rationale may be extended for the
divergence across the entire superfamily as well. It is likely that protein engineering strategies
employing rational or semi-rational redesign of residues identified in our study can be an
appropriate starting point to generate variants with desirable biocatalytic properties like
enhanced activity, improved enantioselectivity, thermal stability, among others. Previous
studies on AKR superfamily members report successful efforts at engineering such variants
[64-66]. For instance, in CtXR, a redesign of loop 1 that contains residue Trp19 (in DnXR),
yielded a variant with enhanced turnover number and enantioselectivity for bulky ketones of
industrial relevance [67].
The structural basis of the broad substrate specificity of the AKR superfamily has
been hampered by the lack of the structure of an appropriate complex. The crystal structure of
a yeast XR resembling the near-cognate ternary complex presented here provides a structural
rationale for the catalytic mechanism, substrate binding and specificity in the AKR family 2,
in particular and in the superfamily, in general.
This article is protected by copyright. All rights reserved.

Experimental procedures
Protein expression, purification and oligomeric state
DnXR (GenBank: KT239024) in the vector pET28a(+) was over-expressed and purified as
described previously [37]. The plasmid construct contains a 21 residue long N-terminal
expression and purification tag including a His6 region. A brief description of DnXR
purification as follows: the protein was expressed in E. coli Rosetta under the IPTG induction
and the harvested cell pellet was lysed by sonication. The obtained supernatant containing
overexpressed protein was taken for purification by Ni-NTA affinity chromatography. The
100 mM and 200 mM imidazole elutions of purification were subjected to buffer exchange
with 20 mM Tris-HCl (pH 7.5) and the protein was concentrated using 30 kDa-cut off
centrifugal filters. The purity of the obtained protein sample was evaluated on SDS-PAGE
(data not shown). The protein concentration was measured by the BCA method and the
purified protein was stored at -80 ºC until use [68]. The oligomeric state of the purified
protein was determined by performing size exclusion chromatography using BIO-RAD FPLC
system at 4 ºC. 2 ml of 20 mM Tris-HCl (pH 7.5) buffer containing 6 mg protein was loaded
onto the Superdex 200 GE Healthcare column which was pre-equilibrated with 20 mM Tris-
HCl (pH 7.5). The elution was carried out with same buffer at the flow rate of 0.5 ml/min
while recording the chromatograms by measuring the absorbance at 280 nm.
Enzyme assays and kinetics
The enzymatic activity of purified protein was measured at optimal pH 7.0 and temperature
45 ºC as determined previously [37]. The change in A340 was monitored continuously at least
for 1 min at every 1 sec time point in a 500 µl reaction mixture containing 50 mM sodium
phosphate buffer (pH 7.0), 0.3 mM NADPH (cosubstrate) and 6 mM erythrose (substrate)
using Lamda 25 UV-VIS spectrophotometer, Perkin Elmer, USA. The linear curve
This article is protected by copyright. All rights reserved.

(absorbance vs. time) at least for 1 min was obtained by the addition of a specific amount of
enzyme (0.3 µg) into the reaction mixture. One enzyme unit (U) was defined as the amount
of enzyme that caused the oxidation of 1 µmol of NADPH per minute [37]. The heat
inactivated enzyme and sodium phosphate buffer (pH 7.0) were considered as a references
[37]. The C4 sugars D-erythrose and L-threose were assayed for enzyme activity. The kinetic
parameters were determined at constant NADPH (0.3 mM as determined previously [37] and
different D-erythrose/L-threose concentrations (0.1-60 mM). The obtained data were used to
calculate the kinetic constants by fitting for the Michaelis-Menten equation using the
software GraphPad Prism 5.
The effect of product erythritol and its analogue dithiothreitol (DTT) on enzyme
activity was studied at various concentrations of erythritol (0-2000 mM) and DTT (0-10 mM)
at the same assay conditions as mentioned above and the obtained data were used to calculate
half maximal inhibitory concentrations (IC50) and minimal inhibitory concentrations (MIC)
using linear regression function of Graphpad Prism 5 (data not shown). Using the obtained
information on IC50 and MIC values, the inhibition kinetics of DnXR was performed at a
particular concentration of erythritol/DTT by measuring the initial velocities of enzyme at
different erythrose concentrations (0.1 - 60 mM). The obtained data were used to calculate
the kinetic parameters by fitting to the standard noncompetitive/mixed inhibition model for
Michaelis-Menten and Lineweaver-Burk plots [69]. The Michaelis–Menten equation for
noncompetitive/mixed inhibition model used was
S
max
S
The plot of KM vs. [I] yielded a linear trend upon linear regression analysis and the
values with intercept KM and slope KM/KI and was used for calculation of the inhibitory
constant KI.
This article is protected by copyright. All rights reserved.

Protein crystallization
The protein DnXR was concentrated to 15 mg/ml in buffer (20 mM Tris-HCl, pH 7.5).
Crystallization experiments were performed in 24 well plates by hanging drop vapor
diffusion method at 20 ºC using commercial screens (Hampton Research) and each drop was
prepared by mixing 1 µl each of protein solution and reservoir solution and equilibrated over
1 ml of reservoir solution. Crystals of the apo form of DnXR (rod shaped) appeared in 2 days
in the crystallization condition containing 24 % PEG 4000, 0.1 M sodium citrate (pH 6.2) and
0.15 M ammonium acetate. Crystals of the NADP-DTT complex were obtained using a
protein solution containing 15 mg/ml protein, 10 mM NADPH and 10 mM DTT which was
pre-incubated at 4 ºC for 4 h. The plate shaped crystals were formed after 3 days with the
reservoir solution containing 22 % PEG 3350, 0.1 M HEPES (pH 7.5) and 0.2 M ammonium
sulfate.
Data collection, structure determination and refinement
Diffraction data for DnXR crystals were collected at an in-house X-ray source with a
MAR345 image plate detector mounted on a Bruker Microstar rotating anode X-ray generator
(Cu K = 1.5418 Å). Crystals were mounted on cryoloops using the higher concentrations of
the reservoir solutions as cryoprotectants and flash-frozen in a nitrogen gas stream at 100 K.
All X-ray data were processed and scaled using MOSFLM and SCALA programs as
implemented in the CCP4 software package [70-72]. The apo form data was processed to 2.0
Å resolution and the data were indexed in the P212121 space group. The structure was solved
by molecular replacement with the PHASER program using one subunit of apo form of C.
tenuis xylose reductase (PDB: 1JEZ, 71 % sequence identity) as the search model [27, 73].
The final model of a dimer in the asymmetric unit contains all 320 amino acids encoded in
the protein, while the N-terminal (His)6-tag is disordered. The complex data was processed to
This article is protected by copyright. All rights reserved.

1.7 Å resolution and indexed in the space group C2221 and structure was solved using the
refined DnXR apo form as a model. The software package PHENIX was used to carry out
iterative rounds of restrained refinement and the program COOT was used for model building
[74-76]. The refinement included simulated annealing protocols to remove model bias. R-free
calculations were performed to cross validate the refinement with a test set containing 5 % of
the reflections. The structure and restraints for the adduct were generated using the electronic
Ligand Builder and Optimization Workbench (eLBOW) program, as implemented in the
PHENIX suite [77]. The coordinates of NADPH and D-DTT available in the PDB Chemical
Components dictionary were used to generate the structure and determine the ligand
geometry. The resulting model was further optimized using the AM1 semi-empirical
quantum-chemical method and corresponding restraints were generated as a CIF format file.
The output coordinates and the restraints were examined manually using the Restraints Editor
Especially Ligand (REEL) tool for validity. The model was then manually fitted into the
electron density map and followed by refinement carried out in PHENIX incorporating the
generated ligand restraints. The MolProbity program was used for the evaluation of
stereochemical quality of the final models [78]. PyMoL was used for the graphical
representation of structures [79]. The structure based sequence alignment was carried out
using the online program ESPript 3.0 [80].
Protein Data Bank accession numbers
The atomic coordinates and the corresponding structure factors of the apo form and the
ternary complex of DnXR have been deposited in the Protein Data Bank under the accession
numbers 5ZCI and 5ZCM, respectively.
This article is protected by copyright. All rights reserved.

Acknowledgements
Infrastructure support from the Macromolecular X-ray Diffraction Facility, IIT Madras and
the DST-FIST Facility, IIT Madras, are acknowledged. BP acknowledges Indian Council of
Medical Research for his fellowship.
Conflict of interest
The authors have no conflict of interest to declare.
Funding Information
This work was supported by a research grant from the Department of Biotechnology,
Ministry
of
Science
and
Technology,
Government
of
India
(Grant
BT/PR6157/PID/6/701/2012).
Author contributions
NM and SNG conceived and coordinated the study and wrote the paper. BP carried out the
majority of the experiments, wrote the initial draft of the manuscript and prepared the figures.
SBM and BP carried out data analysis and data visualization. NM carried out the
crystallographic analysis and supervised the structural data analysis.
References
1.
2.
Penning TM (2015) The aldo-keto reductases (AKRs): overview. Chem Biol Interact
234, 236–246.
Wilson DK, Bohren KM, Gabbay KH & Quiocho FA (1992) An unlikely sugar
substrate site in the 1.65 Å structure of the human aldose reductase holo enzyme
implicated in diabetic complications. Science 257, 81–84.
Kador PF, Wyman M & Oates P. J (2016) Aldose reductase, ocular diabetic
complications and the development of topical Kinostat (®). Prog Retin Eye Res 54, 1–
29.
3.
This article is protected by copyright. All rights reserved.

4.
5.
Lanaspa MA, Ishimoto T, Li N, Cicerchi C, Orlicky D. J, Ruzicky P, Rivard C, Inaba
S, Roncal-Jimenez CA, Bales ES, Diggle CP, Asipu A, Petrash JM, Kosugi T,
Maruyama S, Sanchez-Lozada LG, McManaman JL, Bonthron DT, Sautin YY &
Johnson RJ (2013) Endogenous fructose production and metabolism in the liver
contributes to the development of metabolic syndrome. Nat Commun 4, 2434.
Andres-Hernando A, Li N, Cicerchi C, Inaba S, Chen W, Roncal-Jimenez C, Le MT,
Wempe MF, Milagres T, Ishimoto T, Fini M, Nakagawa T, Johnson RJ & Lanaspa
MA (2017) Protective role of fructokinase blockade in the pathogenesis of acute
kidney injury in mice. Nat Commun 8, 14181.
6.
7.
8.
Ramana KV (2011) Aldose reductase: new insights for an old enzyme. Biomol
Concepts 2, 103–114.
Lee H (1998) The structure and function of yeast xylose (aldose) reductases. Yeast
14, 977–984.
Su B, Zhang Z, Wu M, Lin J & Yang L (2016) Construction of plasmid-free
Escherichia coli for the production of arabitol-free xylitol from corncob
hemicellulosic hydrolysate. Sci Rep 6, 26567.
9.
Moon HJ, Jeya M, Kim IW & Lee JK (2010) Biotechnological production of
erythritol and its applications. Appl Microbiol Biotechnol 86, 1017–1025.
Lee JK, Kim SY, Ryu YW, Seo JH & Kim JH (2003) Purification and
characterization of a novel erythrose reductase from Candida magnoliae. Appl
Environ Microbiol 69, 3710–3718.
10.
11.
12.
Jovanović B, Mach RL & Mach-Aigner AR (2013) Characterization of erythrose
reductases from filamentous fungi. AMB Express 3, 43.
Janek T, Dobrowolski A, Biegalska A & Mirończuk AM (2017) Characterization of
erythrose reductase from Yarrowia lipolytica and its influence on erythritol
synthesis. Microb. Cell Fact 16, 118.
13.
Simpson PJ, Tantitadapitak C, Reed AM, Mather OC, Bunce CM, White SA & Ride
JP (2009) Characterization of two novel aldo-keto reductases from Arabidopsis:
expression patterns, broad substrate specificity, and an open active-site structure
suggest a role in toxicant metabolism following stress. J Mol Biol 392, 465–480.
Bartels D, Engelhardt K, Roncarati R, Schneider K, Rotter M & Salamini F (1991)
An ABA and GA modulated gene expressed in the barley embryo encodes an aldose
reductase related protein. EMBO J 10, 1037–1043.
Zheng X, Zhang L, Chen W, Chen Y, Xie W & Hu X (2012) Partial inhibition of
aldose reductase by nitazoxanide and its molecular basis. ChemMedChem 7, 1921–
1923.
Steuber H, Heine A & Klebe G. (2007) Structural and thermodynamic study on
aldose reductase: nitro-substituted inhibitors with strong enthalpic binding
contribution. J Mol Biol 368, 618–638.
Howard EI, Sanishivili R, Cachau RE, Mitschler A, Chevrier B, Barth P, Lamour V,
Van Zandt M, Sibley E, Bon C, Moras D, Schneider TR, Joachimiak A & Podjarny
A (2004) Ultrahigh resolution drug design I: details of interactions in human aldose
reductase-inhibitor complex at 0.66 A. Proteins 55, 792–804.
14.
15.
16.
17.
This article is protected by copyright. All rights reserved.

18.
19.
Rondeau JM, Tête-Favier F, Podjarny A, Reymann JM, Barth P, Biellmann JF &
Moras D (1992) Novel NADPH-binding domain revealed by the crystal structure of
aldose reductase. Nature 355, 469–472.
Urzhumtsev A, Tête-Favier F, Mitschler A, Barbanton J, Barth P, Urzhumtseva L,
Biellmann JF, Podjarny A & Moras D (1997) A “specificity” pocket inferred from
the crystal structures of the complexes of aldose reductase with the pharmaceutically
important inhibitors tolrestat and sorbinil. Structure 5, 601–612.
20.
21.
Sundaram K, Dhagat U, Endo S, Chung R, Matsunaga T, Hara A & El-Kabbani O
(2011) Structure of rat aldose reductase-like protein AKR1B14 holoenzyme: Probing
the role of His269 in coenzyme binding by site-directed mutagenesis. Bioorganic
Med Chem Lett 21, 801–804.
Sundaram K, Endo S, Matsunaga T, Tanaka N, Hara A & El-Kabbani O (2012)
Structure of the His269Arg mutant of the rat aldose reductase-like protein AKR1B14
complexed with NADPH. Acta Crystallogr Sect F Struct Biol Cryst Commun 68,
400–403.
22.
23.
24.
Ferrell M, Abendroth J, Zhang Y, Sankaran B, Edwards TE, Staker BL, Van Voorhis
W. C, Stewart LJ & Myler PJ (2011) Structure of aldose reductase from Giardia
lamblia. Acta Crystallogr Sect F Struct Biol Cryst Commun 67, 1113–1117.
Sanli G, Dudley JI & Blaber M (2003) Structural biology of the aldo-keto reductase
family of enzymes: catalysis and cofactor binding. Cell Biochem Biophys 38, 79–
101.
Kubiseski TJ, Hyndman DJ, Morjana NA & Flynn TG (1992) Studies on pig muscle
aldose reductase: kinetic mechanism and evidence for a slow conformational change
upon coenzyme binding. J Biol Chem 267, 6510–6517.
25.
26.
Kubiseski TJ & Flynn TG (1995) Studies on human aldose reductase: probing the role
of arginine 268 by site-directed mutagenesis. J Biol Chem 270, 16911–16917.
Rechlin C, Scheer F, Terwesten F, Wulsdorf T, Pol E, Fridh V, Toth P, Diederich
WE, Heine A & Klebe G (2017) Price for opening the transient specificity pocket in
human aldose reductase upon ligand binding: structural, thermodynamic, kinetic, and
computational analysis. ACS Chem Biol 12, 1397–1415.
27.
28.
29.
Kavanagh KL, Klimacek M, Nidetzky B & Wilson DK (2002) The structure of apo
and holo forms of xylose reductase, a dimeric aldo-keto reductase from Candida
tenuis. Biochemistry 41, 8785–8795.
Kavanagh KL, Klimacek M, Nidetzky B & Wilson DK (2003) Structure of xylose
reductase bound to NAD+ and the basis for single and dual co-substrate specificity in
family 2 aldo-keto reductases. Biochem J 373, 319–326.
Leitgeb S, Petschacher B, Wilson DK & Nidetzky B (2005) Fine tuning of coenzyme
specificity in family 2 aldo-keto reductases revealed by crystal structures of the
Lys274Arg mutant of Candida tenuis xylose reductase (AKR2B5) bound to NAD+
and NADP+. FEBS Lett 579, 763–767.
30.
Petschacher B, Leitgeb S, Kavanagh KL, Wilson DK & Nidetzky B (2005) The
coenzyme specificity of Candida tenuis xylose reductase (AKR2B5) explored by
site-directed mutagenesis and X-ray crystallography. Biochem J 385, 75–83.
This article is protected by copyright. All rights reserved.

31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
Kratzer R, Wilson D & Nidetzky B (2006) Catalytic mechanism and substrate
selectivity of aldo-keto reductases: Insights from structure-function studies of
Candida tenuis xylose reductase. IUBMB Life 58, 499–507.
Kratzer R, Leitgeb S, Wilson DK & Nidetzky B (2006) Probing the substrate binding
site of Candida tenuis xylose reductase (AKR2B5) with site-directed mutagenesis.
Biochem J 393, 51–58.
Olsen JG, Pedersen L, Christensen CL, Olsen O & Henriksen A (2008) Barley
aldose reductase: structure, cofactor binding, and substrate recognition in the
aldo/keto reductase 4C family. Proteins: Struct Funct Bioinf 71, 1572–1581.
Khurana S, Powers DB, Anderson S & Blaber M (1998) Crystal structure of 2,5-
diketo-D-gluconic acid reductase A complexed with NADPH at 2.1 Å resolution.
Proc Natl Acad Sci 95, 6768–6773.
Harrison DH, Bohren KM, Ringe D, Petsko GA & Gabbay KH (1994) An anion
binding site in human aldose reductase: mechanistic implications for the binding of
citrate, cacodylate, and glucose 6-phosphate. Biochemistry 33, 2011–2020.
Kumar S & Gummadi SN (2011) Metabolism of glucose and xylose as single and
mixed feed in Debaryomyces nepalensis NCYC 3413: production of industrially
important metabolites. Appl Microbiol Biotechnol 89, 1405–1415.
Paidimuddala B, Krishna AG & Gummadi SN (2017) A halotolerant aldose reductase
from Debaryomyces nepalensis: gene isolation, overexpression and biochemical
characterization. RSC Adv 7, 20384–20393.
Neuhauser W, Haltrich D, Kulbe KD & Nidetzky B (1997) NAD(P)H-dependent
aldose reductase from the xylose-assimilating yeast Candida tenuis. Isolation,
characterization and biochemical properties of the enzyme. Biochem J 326, 683–692.
Almarsson O & Bruice TC (1993) Evaluation of the factors influencing reactivity
and stereospecificity in NAD(P)H dependent dehydrogenase enzymes. J Am Chem
Soc 115, 2125–2138.
Tsybovsky Y, Donato H, Krupenko NI, Davies C & Krupenko SA (2007) Crystal
structures of the carboxyl terminal domain of rat 10-formyltetrahydrofolate
dehydrogenase: implications for the catalytic mechanism of aldehyde
dehydrogenases. Biochemistry 46, 2917–2929.
41.
42.
Díaz-Sánchez ÁG, González-Segura L, Rudiño-Piñera E, Lira-Rocha A, Torres-
Larios A & Muñoz-Clares RA (2011) Novel NADPH-cysteine covalent adduct
found in the active site of an aldehyde dehydrogenase. Biochem J 439, 443–455.
Park J & Rhee S (2013) Structural basis for a cofactor-dependent oxidation
protection and catalysis of cyanobacterial succinic semialdehyde dehydrogenase. J
Biol Chem 288, 15760–15770.
43.
44.
Van Eys J, & Kaplan NO (1957) The addition of sulfhydryl compounds to
diphosphopyridine nucleotide and its analogues. J Biol Chem 228, 305–314.
Lee YS, Hodoscek M, Brooks BR & Kador PF (1998) Catalytic mechanism of
aldose reductase studied by the combined potentials of quantum mechanics and
molecular mechanics. Biophys Chem 70, 203–216.
45.
Várnai P, Richards WG & Lyne PD (1999) Modelling the catalytic reaction in
human aldose reductase. Proteins: Struct Funct Genet 37, 218–227.
This article is protected by copyright. All rights reserved.

46.
47.
Várnai P & Warshel A (2000) Computer simulation studies of the catalytic
mechanism of human aldose reductase. J Am Chem Soc 122, 3849–3860.
Sotriffer CA, Krämer O & Klebe G (2004) Probing flexibility and “induced-fit”
phenomena in aldose reductase by comparative crystal structure analysis and
molecular dynamics simulations. Proteins: Struct Funct Bioinf 56, 52–66.
Kolb P, Ferreira RS, Irwin JJ & Shoichet B. K (2009) Docking and chemoinformatic
screens for new ligands and targets. Curr Opin Biotechnol 20, 429–436.
Dréanic MP, Edge CM & Tuttle T (2017) New insights into the catalytic mechanism
of aldose reductase: a QM/MM study. ACS Omega 2, 5737–5747.
Bohren KM, Grimshaw CE, Lai CJ, Harrison DH, Ringe D, Petsko GA & Gabbay
KH (1994) Tyrosine-48 is the proton donor and histidine-110 directs substrate
stereochemical selectivity in the reduction reaction of human aldose reductase:
enzyme kinetics and crystal structure of the Y48H mutant enzyme. Biochemistry 33,
2021–2032.
48.
49.
50.
51.
52.
53.
Cui Q, Elstner M & Karplus M (2002) A theoretical analysis of the proton and
hydride transfer in liver alcohol dehydrogenase (LADH). J Phys Chem B 106, 2721–
2740.
Dzierlenga MW, Antoniou D & Schwartz SD (2015) Another look at the
mechanisms of hydride transfer enzymes with quantum and classical transition path
sampling. J Phys Chem Lett 6, 1177–1181.
Blakeley MP, Ruiz F, Cachau R, Hazemann I, Meilleur F, Mitschler A, Ginell S,
Afonine P, Ventura ON, cousido-Siah A, Haertlein M, Joachimiak A, Myles D &
Podjarny (2008) Quantum model of catalysis based on a mobile proton revealed by
subatomic x-ray and neutron diffraction studies of h-aldose reductase. Proc Natl
Acad Sci 105, 1844–1848.
54.
Ye Q, Hyndman D, Green NC, Li L, Jia Z & Flynn TG (2001) The crystal structure
of an aldehyde reductase Y50F mutant-NADP complex and its implications for
substrate binding. Chem Biol Interact 130, 651–658.
55.
56.
57.
58.
Dittmer DC & Kolyer JM (1963) Addition compounds of thiols and 1-substituted
nicotinamides. J Org Chem 28, 1720–1722.
Misra HP (1974) Generation of superoxide free radicals during the autoxidation of
thiols. J Biol Chem 249, 2151–2155.
Baccanari DP (1978) Coupled oxidation of NADPH with thiols at neutral pH. Arch
Biochem Biophys 191, 351–357.
Biadene M, Hazemann I, Cousido A, Ginell S, Joachimiak A, Sheldrick GM,
Podjarny A & Schneider TR (2007) The atomic resolution structure of human aldose
reductase reveals that rearrangement of a bound ligand allows the opening of the
safety-belt loop. Acta Crystallogr D Biol Crystallogr 63, 665–672.
Giuseppe PO, de Santos MLD, Sousa SM., de Koch KE, Yunes JA, Aparicio R &
Murakami MT (2016) A comparative structural analysis reveals distinctive features
of co-factor binding and substrate specificity in plant aldo-keto reductases. Biochem
Biophys Res Commun 474, 696–701.
59.
This article is protected by copyright. All rights reserved.

60.
61.
62.
63.
64.
65.
Zhang M, Jiang ST, Zheng Z, Li XJ, Luo SZ & Wu XF (2015) Cloning, expression,
and characterization of a novel xylose reductase from Rhizopus oryzae. J Basic
Microbiol 55, 907–921.
Khoury GA, Fazelinia H, Chin JW, Pantazes RJ, Cirino PC & Maranas CD (2009)
Computational design of Candida boidinii xylose reductase for altered cofactor
specificity. Protein Sci 18, 2125–2138
Banta S, Swanson BA, Wu S, Jarnagin A & Anderson S (2002) Alteration of the
specificity of the cofactor-binding pocket of Corynebacterium 2,5-diketo-D-gluconic
acid reductase A. Protein Eng 15, 131–140.
Hoog SS, Pawlowski JE, Alzari PM, Penning TM & Lewis M (1994) Three-
dimensional structure of rat liver 3α-hydroxysteroid/dihydrodiol dehydrogenase: a
member of the aldo-keto reductase superfamily. Proc Natl Acad Sci 91, 2517–2521.
Asako H, Shimizu M & Itoh N (2008) Engineering of NADPH-dependent aldo-keto
reductase from Penicillium citrinum by directed evolution to improve thermostability
and enantioselectivity. Appl Microbiol Biotechnol 80, 805–812.
Kratzer R & Nidetzky B (2007) Identification of Candida tenuis xylose reductase as
highly selective biocatalyst for the synthesis of aromatic alpha-hydroxy esters and
improvement of its efficiency by protein engineering. Chem Commun 14, 1047–
1049.
66.
Wang Z, Zhou S, Zhang S, Zhang S, Zhu F, Jin X, Chen Z & Xu X (2017) Semi-
rational engineering of a thermostable aldo-keto reductase from Thermotoga
maritima for synthesis of enantiopure ethyl-2-hydroxy-4-phenylbutyrate (EHPB). Sci
Rep 7, 4007.
67.
68.
Krump C, Vogl M, Brecker L, Nidetzky B & Kratzer R (2014) Acceleration of an
aldo-keto reductase by minimal loop engineering. Protein Eng Des Sel 27, 245–248.
Smith PK, Krohn RI, Hermanson GT, Mallia AK, Gartner FH, Provenzano MD,
Fujimoto EK, Goeke NM, Olson BJ & Klenk DC (1985) Measurement of protein
using bicinchoninic acid. Anal Biochem 150, 76–85.
69.
70.
71.
Cook PF & Cleland WW (2007) Enzyme kinetics and mechanism, Garland science,
NY.
Kabsch W (1988) Evaluation of single-crystal X-ray diffraction data from a position-
sensitive detector. J Appl Crystallogr 21, 916–924.
Battye TGG, Kontogiannis L, Johnson O, Powell HR & Leslie AGW (2011)
iMOSFLM: a new graphical interface for diffraction-image processing with
MOSFLM. Acta Crystallogr D Biol Crystallogr 67, 271–281.
72.
Winn MD, Ballard CC, Cowtan KD, Dodson EJ, Emsley P, Evans PR, Keegan RM,
Krissinel EB, Leslie AGW, McCoy A, McNicholas SJ, Murshudov GN, Pannu NS,
Potterton EA, Powell HR, Read RJ, Vagin A & Wilson KS (2011) Overview of the
CCP 4 suite and current developments. Acta Crystallogr D Biol Crystallogr 67, 235–
242.
73.
74.
McCoy AJ (2007) Solving structures of protein complexes by molecular replacement
with Phaser. Acta Crystallogr D Biol Crystallogr 63, 32–41.
Adams PD, Afonine PV, Bunkóczi G, Chen VB, Davis IW, Echols N, Headd JJ,
Hung LW, Kapral GJ, Grosse-Kunstleve RW, McCoy AJ, Moriarty NW, Oeffner R,
This article is protected by copyright. All rights reserved.