Crystal structure of the Homo sapiens kynureninase-3-hydroxyhippuric acid inhibitor complex: Insights into the molecular basis of kynureninase substrate specificity

Article abstract of DOI:10.1021/jm8010806

Homo sapiens kynureninase is a pyridoxal-5'-phosphate dependent enzyme that catalyzes the hydrolytic cleavage of 3-hydroxykynurenine to yield 3-hydroxyanthranilate and L-alanine as part of the tryptophan catabolic pathway leading to the de novo biosynthesis of NAD⁺. This pathway results in quinolinate, an excitotoxin that is an NMDA receptor agonist. High levels of quinolinate have been correlated with the etiology of neurodegenerative disorders such as AIDS-related dementia and Alzheimer's disease. We have synthesized a novel kynureninase inhibitor, 3-hydroxyhippurate, cocrystallized it with human kynureninase, and solved the atomic structure. On the basis of an analysis of the complex, we designed a series of His- 102, Ser-332, and Asn-333 mutants. The H102W/N333T and H102W/S332G/N333T mutants showed complete reversal of substrate specificity between 3-hydroxykynurenine and L-kynurenine, thus defining the primary residues contributing to substrate specificity in kynureninases.

Full text of DOI:10.1021/jm8010806

J. Med. Chem. 2009, 52, 389–396
389
Crystal Structure of the Homo sapiens Kynureninase-3-Hydroxyhippuric Acid Inhibitor
Complex: Insights into the Molecular Basis Of Kynureninase Substrate Specificity†
Santiago Lima,‡, Sunil Kumar,§, Vijay Gawandi,§ Cory Momany,| and Robert S. Phillips*,‡,§
Department of Biochemistry and Molecular Biology, UniVersity of Georgia, Athens, Georgia 30602, Department of Chemistry, UniVersity of
Georgia, Athens, Georgia 30602, Department of Pharmaceutical and Biomedical Sciences, College of Pharmacy, UniVersity of Georgia,
Athens, Georgia 30602
ReceiVed August 30, 2008
Homo sapiens kynureninase is a pyridoxal-5′-phosphate dependent enzyme that catalyzes the hydrolytic
cleavage of 3-hydroxykynurenine to yield 3-hydroxyanthranilate and L-alanine as part of the tryptophan
catabolic pathway leading to the de novo biosynthesis of NAD+. This pathway results in quinolinate, an
excitotoxin that is an NMDA receptor agonist. High levels of quinolinate have been correlated with the
etiology of neurodegenerative disorders such as AIDS-related dementia and Alzheimer’s disease. We have
synthesized a novel kynureninase inhibitor, 3-hydroxyhippurate, cocrystallized it with human kynureninase,
and solved the atomic structure. On the basis of an analysis of the complex, we designed a series of His-
102, Ser-332, and Asn-333 mutants. The H102W/N333T and H102W/S332G/N333T mutants showed
complete reversal of substrate specificity between 3-hydroxykynurenine and L-kynurenine, thus defining
the primary residues contributing to substrate specificity in kynureninases.
In mammals, the essential amino acid L-tryptophan is a
precursor for metabolites such as the neurotransmitter serotonin,
the hormone melatonin, nicotinic acid, and NAD+, with the latter
two being the primary metabolic fate of dietary tryptophan. The
catabolic cascade that leads to the de novo biosynthesis of
nicotinic acid and NAD+, commonly known as the kynurenine
pathway,1 is notable for intermediates with important neuroac-
tive properties. Kynureninase, the third enzyme along the
kynurenine pathway, is a pyridoxal-5′-phosphate (PLP)a depen-
dent enzyme that catalyzes the hydrolytic cleavage of 3-hydroxy-
L-kynurenine (3-OH-Kyn) to yield 3-hydroxyanthranilic acid and
L-alanine (eq 1, R ) OH). One further kynurenine pathway
enzyme, 3-hydroxyanthranilate-3,4-dioxygenase, metabolizes
3-hydroxyanthranilic acid, the product of which undergoes a
nonenzymatic rearrangement leading to the formation of quino-
linate.2 Quinolinate has endogenous neurotoxic activity3 that is
mediated via its agonist effect on glutamate sensitive NMDA
ionotropic receptors,4 and its excitotoxic effect has been
correlated with the etiology of many neurodegenerative diseases
including AIDS-related dementia,5,6 Huntington’s disease,7,8
amyotrophic lateral sclerosis,9 and Alzheimer’s disease.10,11 The
increasing body of evidence implicating changes in normal
concentrations of quinolinate in CNS tissues with neurotoxicity
has led to the idea that altering the ratio of quinolinic acid to
other kynurenine pathway metabolites could provide a neuro-
protective effect during the onset of such maladies.12-15 Thus,
our goal was to provide a robust template that can be used for
rational and in silico drug design as well as complex pharma-
cophore library screening. Herein we report the crystal structure
of Homo sapiens kynureninase (HKynase) complexed with
3-hydroxyhippuric acid (5), a novel competitive inhibitor of
kynureninase. The complex reveals important enzyme-ligand
interactions that should be considered for the design and
screening of novel kynureninase inhibitors.
Two functional orthologues of kynureninase are known, a
“constitutive”, primarily eukaryotic enzyme16 that preferentially
catalyzes the hydrolysis of 3-OH-Kyn (eq 1, R ) OH) and a
primarily prokaryotic “inducible”16 enzyme that catalyzes the
hydrolysis of L-kynurenine (L-Kyn) to yield anthranilic acid and
L-Ala (eq 1, R ) H). Although both orthologues are able to
catalyze the hydrolysis of their noncognate substrates, they do
so at about 100-fold lower efficiency.17,18 Indeed, a number of
organisms, especially fungi, are known to contain both ortho-
logues.16 To date, the molecular basis of the substrate specificity
that allows these enzymes to differentiate between their
substrates has not been elucidated. Some insight into this
mechanism was gained through docking studies with the native
structure of H. sapiens kynureninase and 3-OH-L-Kyn as the
target ligand.18 These results revealed that two active site
residues, His-102 and Asn-333, could play an important role in
substrate binding and specificity.18 We present herein additional
experimental evidence supporting this hypothesis.
†
PDB accession code: 3E9K.
* To whom correspondence should be addressed. Phone: (706) 542-1996.
Fax: (706) 542-9454. E-mail: rsphillips@chem.uga.edu. Address: Depart-
ment of Chemistry, University of Georgia, Athens, GA 30602.
‡
Department of Biochemistry and Molecular Biology, University of
Georgia.
Results and Discussion
§
Department of Chemistry, University of Georgia.
Department of Pharmaceutical and Biomedical Sciences, College of
|
Synthesis of 5. Compound 5 was designed as an analogue
of 3-OH-Kyn that would bind to HKynase and be stable during
cocrystallization for up to several weeks. The synthesis of 5
started from readily available 3-hydroxybenzaldehyde (1), as
shown in Scheme 1. After protection of the hydroxyl group in
Pharmacy, University of Georgia.
S.L. and S.K. contributed equally to this work.
a Abbreviations: PLP, pyridoxal-5′-phosphate; L-Kyn, L-kynurenine;
3-OH-Kyn, 3-hydroxy-L-kynurenine; HKynase, Homo sapiens kynureninase;
5, 3-hydroxyhippuric acid.
10.1021/jm8010806 CCC: $40.75
2009 American Chemical Society
Published on Web 12/17/2008

390 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 2
Lima et al.
Table 1. Summary of Crystallographic Analysis
Scheme 1
data collection and processing statistics
space group
C2
resolution range (outer shell), Å
no. of reflections (outer shell)
average I/σI (outer shell)
redundancy (outer shell)
completeness (outer shell), %
Rmerge (outer shell)
89.09-1.65 (1.68-1.65)
55449 (2096)
24.1 (1.9)
3.5 (2.3)
96.3 (68.4)
0.050 (0.362)
refinement statistics
Rfactor (outer shell)
0.154 (0.251)b
0.190 (0.342)
16.65
a
Rfree (outer shell)
mean B value, Å2
rmsd from ideal geometry
bond angles, deg
1.067
bond distances, Å
0.007
Ramachandran plot residues in
favored/allowed/disallowed regions, %
97.06/2.71/0.23
a Rfree calculated with 5.1% of the total data that were excluded from
refinement. b The outer shell resolution range used in refinement was
1.69-1.65 Å.
substituted hippurates tested and was thus chosen for the
cocrystallization experiments. This compound was found to be
a moderate competitive inhibitor of HKynase with a Ki of 60
µM. The fact that 5 is the most potent inhibitor of the
compounds tested reflects the contribution of the 3-OH moiety
to substrate specificity observed between kynurenine and
3-hydroxykynurenine among kynureninases.18 Inhibitors that
contain an R-amino group capable of forming an external
aldimine with the PLP do not require the 3-OH to bind to
HKynase because compounds such as 2-methoxybenzoylala-
nine20 are also modest kynureninase inhibitors. However, the
most potent inhibitor known for HKynase is 2-amino-4-[3′-
hydroxyphenyl]-4-hydroxybutanoic acid, which contains both
an R-amino group and the 3-OH, with a Ki value of 100 nM.21
Human Kynureninase-5 Inhibitor Complex. Statistics for
the refined model are presented in Table 1. The structure (PDB
accession number 3E9K) contained one monomer per asym-
metric unit and had good geometry with 441/1 residues in the
allowed/disallowed regions of the Ramachandran plot. The final
model was refined to an Rvalue of 0.152, Rfree value of 0.189,
and covers 96% of the predicted amino acid sequence with 669
water molecules included in the final structure. The biologically
active unit (dimer) can be generated by applying the crystal-
lographic symmetry operator 1-x, -y, -z to the monomer. The
regions between residues 1-5, 377-387, and 461-465 could
not be modeled due to poorly ordered electron density. These
regions had disordered electron density in the native HKynase
structure as well.18 Electron density could be seen extending
between the ε amino group of Lys-276 and the C-4′ of PLP, an
observation consistent with a PLP internal aldimine covalent
adduct. Positive density for the 5 molecule could be clearly
observed in a difference electron density map (Fo - Fc)
calculated from the unrefined molecular replacement solution.
Upon docking the molecule of 5, electron density for the atoms
improved and became continuous (Figure 1). Further refinement
revealed the presence of two water molecules, water 669 and
water 671, in difference electron density maps within a short
distance of the C2 atom of the aromatic ring of 5. Refinement
of these waters and compound 5 at full occupancy produced
density for both molecules at negative contour levels in a
difference electron density map. Upon adjustment of occupan-
cies to 20% for water 669 and 671 and 80% for compound 5,
no residual positive or negative density (<3σ) could be observed
for either molecule. Interestingly, water 669 occupies a position
relative to 5 similar to that which would be occupied by a
Scheme 2
1 with ethylchloroformate to obtain 3-ethoxycarbonyloxyben-
zaldehyde (2), the aldehyde was oxidized to 3-ethoxycarbony-
loxybenzoic acid (3) with oxone in DMF. The carboxylic acid
(3) was then converted to 3-ethoxycarbonyloxybenzoyl chloride
(4) with oxalyl chloride. Finally, deprotection of the hydroxyl
group of 4 and formation of 5 was achieved in a single step by
the reaction of 4 with glycine in the presence of 0.5 M sodium
hydroxide. The product 5 is at least 98% pure based on HPLC
1
and H NMR (see Supporting Information).
Inhibition of Hkynase by 5. Our previous efforts to complex
HKynase crystals with substrate and analogues had proven
unsuccessful due to severe crystal cracking immediately upon
exposure to the compounds in solution. Substituted hippurates
were chosen for complexation because these compounds are
structural analogues of L-Kyn/3-OH-Kyn (Scheme 2), which can
form a Michaelis complex with HKynase. Because the hippu-
rates do not have a free amino group, they bind without forming
a covalent adduct with PLP, thus avoiding the structural
rearrangements associated with substrate binding,19 common
among PLP-dependent aminotransferases, which can lead to
crystal cracking during complexation. The parent compound,
hippuric acid, does not show any detectable inhibition of
HKynase at 1 mM. Of the other substituted hippurates tested,
2-amino-3-hydroxyhippuric acid (synthesis not shown) oxidizes
rapidly in solution and thus could not be used in incubations to
grow crystals. 2-Aminohippuric acid proved to be unsuitable
because it also did not inhibit HKynase at a concentration of 1
mM. Compound 5 was the most stable inhibitor of the

HKynase-3-Hydroxyhippuric Acid Inhibitor Complex
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 2 391
stituent and the side chain δ oxygen of Asn-333* (3.54 Å). Also,
the carbonyl oxygen of 5 is involved in a weak hydrogen bond
with the hydroxyl oxygen of Tyr-275 (3.77 Å) and the amide
nitrogen of 5 is within hydrogen bonding distance to the Ser-
75 γ oxygen (3.63 Å). On the other side of the active site cavity,
and near the Lys-276-PLP internal aldimine imine bond, the
carboxylate of 5 forms an ion pair with Arg-434 (2.98 and 3.06
Å), and hydrogen bonds with the ε nitrogen of His-253 (2.74
Å), and with the γ oxygen of Ser-75 (2.88 Å). The interaction
between the carboxylate of 5 (which is equivalent to the
R-carboxylate of L-enantiomer substrates) with the strictly
conserved Arg-434 is well characterized among members of
the PLP-dependent aspartate aminotransferase superfamily and
is crucial to catalysis.19,23,24
A structural superposition of native (PDB accession code
2HZP) and 5 complexed HKynase reveals no major structural
conformational changes. However, the side chains of residues
Tyr-226, Arg-428, and Arg-434 showed conformational differ-
ences associated with the presence of 5. Specifically, Arg-434
forms an electrostatic bond with the carboxylate oxygens of 5.
To do so, the Arg-434 guanidino nitrogens must move 7.7 Å
from the position observed in native HKynase.18 Consistent with
the partial occupancy for 5, the difference electron density map
suggests two conformational states for the side chain of Arg-
434: a high occupancy state (80%) in which the side chain
interacts with the carboxylate oxygens of 5, and a low occupancy
state (20%) in which no ligand is present in the active site and
the Arg-434 side chain conformation resembles that of native
HKynase. Similarly, Tyr-226 and Arg-428 showed additional
density in a difference electron density map, suggesting that
these residues must adopt different conformations in order to
accommodate 5 in the active site cavity. These residues were
modeled to have 80% and 20% conformations as well. Interest-
ingly, these residues do not directly contact ligand atoms, yet
they line the surface of an interdomain channel that leads to
the active site cavity. Thus, these residues appear to be
dynamically involved in substrate binding by allowing access
to the active site. The position of the side chains of these residues
in ligand bound kynureninase places them at 4.4 Å (between
hydroxyl oxygens of Tyr-226) and 3.7 Å (between guanidino
nitrogens of Arg-428) from those in native HKynase.
Figure 1. Electron density map showing continuous electron density
for atoms of 5 (green carbon atoms, CPK coloring) in the active site
of Hkynase.
substituent at the 2-position of the aromatic ring of 5. Water
669 is stabilized by hydrogen bonds to the hydroxyl group of
Tyr-275 (2.90 Å) and the δ NH2 group of Asn-333* (3.19 Å)
(* denotes residues on the 2-fold symmetry related monomer).
Thus, this partial occupancy water molecule reveals some of
the interactions that are likely to stabilize the 2-amino substituent
on the aromatic ring in 3-OH-Kyn and L-Kyn.
The molecule of 5 is stabilized within the HKynase active
site by interactions with residues from both monomer chains.
The aromatic ring moiety of 5 lies in the active site pocket
between the side chains of Ile-110*, His-102*, Tyr-275, Trp-
305*, Phe-306*, Phe-314*, and Asn-333* (Figure 2). Trp-305*
and Phe-314* stabilize 5 via alignment of their molecular
quadrupole moments toward the aromatic ring (3.53 and 3.73
Å, respectively) and sandwich it with a π-stacking interaction
against the side chain of His-102*. Trp-305 is strictly conserved
in kynureninases of both eukaryotic and prokaryotic origin.
Interestingly, Trp-305* also participates in cofactor binding by
hydrogen bonding through its side chain pyrrole nitrogen with
a PLP phosphate oxygen. Among PLP-dependent aminotrans-
ferases, the external aldimine scissile bond must be aligned
perpendicular to the plane of the cofactor ring system in order
to minimize the energy of the upcoming transition state.22 Thus,
maintaining the appropriate geometry between cofactor and
substrate in order to maximize σ-π overlap of the breaking
bond with the cofactor π-system is crucial to catalysis. This
complex reveals that Trp-305* likely plays a key role in
maintaining the geometry of the external aldimine covalent
adduct and subsequent intermediates by interacting with both
substrate and cofactor. Moreover, Trp-305* is held firmly in
position, relative to both cofactor and ligand, by two flanking
quadrupole π-stacking interactions with the side chains of Phe-
165 and Phe-306*. Phe-165 is a crucial active site residue
because it participates in cofactor binding by π-stacking with
the pyridine ring moiety of PLP. Phe-165 is stabilized in this
position by one further quadrupole π-stacking interaction with
the side chain of Phe-225. Phe-165, Trp-305*, and Phe-314*
are strictly conserved among kynureninases regardless of their
substrate specificity. Furthermore, all kynureninases have either
a histidine or tryptophan residue at the bottom of the active
site cavity to provide a π-stacking interaction to the substrate
ring moiety.
The structure of the HKynase-5 complex provides new
insights into the catalytic mechanism of kynureninase. In
previous studies on the mechanism of kynureninase from
Pseudomonas fluorescens, we determined the stereochemistry
of inhibition by transition state analogues and proposed that a
general acid-base catalyst was involved in hydration of the
substrate carbonyl and cleavage of the resulting gem-diol to give
the carboxylic acid product.25 The stereochemistry of inhibition
by dihydrokynurenine and retro-aldol reactions catalyzed by
kynureninase suggested that the water addition occurred from
the re-face of the carbonyl.25 In the present structure, the face
of the inhibitor carbonyl equivalent to the re-face of the substrate
is exposed, and the opposite face is covered by the ring of Tyr-
275 and the side chain of Ser-75. Thus, the structure of the
5-HKynase complex is consistent with the stereochemistry of
inhibition by dihydrokynurenine and suggests Tyr-275 may be
an auxiliary acid-base catalyst for the reaction. Supporting the
idea that Tyr-275 may play a critical role in catalysis is the
observation that this Tyr residue is strictly conserved in all
known kynureninase sequences (Figure 3). In addition, Ser-75
could potentially donate a hydrogen bond to the carbonyl, and
it is also strictly conserved (Figure 3). A plausible mechanism
for HKynase, taking into account the new structural data, is
Other interactions that stabilize 5 in the HKynase active site
include hydrogen bonding between the 3-hydroxyl ring sub-

392 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 2
Lima et al.
Figure 2. Relaxed-eye stereo representation of the binding interactions of 5 (colored green, ball and stick atoms) within the HKynase active site.
Residues contributed from the symmetry related monomer are labeled with an asterisk (*).
Figure 3. Sequence alignments of kynureninases from various organisms. Numbers at the bottom refer to the human sequence.
shown in Scheme 3. In the 3-OH-Kyn Schiff’s base, Lys-276
removes the R-proton to form a quinonoid species, which is
rapidly reprotonated at C-4′ to form a ketimine.26 Lys-276 can
then act as a general base to deprotonate a water molecule so
that it can add to the carbonyl. A potential donor for the gem-
diol of the reaction intermediate is water-657 hydrogen bonded
to the side chain of Asp-426. We note that 5 is not quite
positioned like a substrate would be because the R-hydrogen
would have to be closer to Lys-276 for proton abstraction. The
reaction of substrate with the PLP would likely result in
reorganization of the active site bringing the catalytic residues
into position. Interestingly, the complex shows that Tyr-275 also
hydrogen bonds with the PLP phosphate, which suggests a
possible involvement of its side chain in the catalytic cycle by
mediating a proton transfer from the phosphate to the carbonyl.
In this scenario, Tyr-275 could initiate Cγ-Cꢀ cleavage by
deprotonation after formation of the gem-diol intermediate. The
proposed mechanistic model resembles that of tyrosine phenol-
lyase, where we found that Tyr-71 forms a hydrogen bond with
the PLP phosphate in the internal aldimine and donates a proton
to the phenol leaving group during catalysis.27
Kinetic Characterization of HKynase Mutants. The struc-
ture of 5 bound in the active site, shown in Figure 2, shows
that Asn-333* accepts a hydrogen bond from the 3-OH of the
bound ligand. As shown in the sequence alignments in Figure
3, the corresponding residue in bacterial kynureninase is a Thr.
Characterization of the kinetic properties of the HKynase N333T
mutant revealed a 9-fold decrease in kcat for L-Kyn, but no
change in Km (Table 2). This mutant has weaker 3-OH-Kyn
binding (6-fold higher Km) and kcat at least 1100 times slower

HKynase-3-Hydroxyhippuric Acid Inhibitor Complex
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 2 393
Scheme 3
Table 2. L-Kynurenine and 3-OH-DL-Kynurenine Kinetic Constants for Wild Type and Mutant HKynases
L-kynurenine 3-OH-DL-Kynurenine
L-Kyn/3-OH-DL-Kyn
enzyme
wild type
N333T
S332G/N333T
H102W/N333T
H102W/S332G/ N333T
Km (µM)
kcat (s-1
)
kcat/Km (M-1 s-1
)
Km (µM)
kcat (s-1
)
kcat/Km (M-1 s-1
)
kcat/Km ratio
495
499
540
327
143
0.230
0.026
0.017
0.007
0.018
465
28.3
3.5
1.23 × 105
3.78 × 10-3
52.1
31.4
21.4
125.6
>200
>510
N/A
<3.9 × 10-3
<10.9 × 10-3
N/A
19.8
2.6
21.5
<0.0165
<0.0165
1.5
>1.36 × 103
N/A
N/A
>7.61 × 103
than wild type HKynase. Thus, this mutation affects 3-OH-Kyn
binding and reduces its catalytic efficiency (kcat/Km) by a factor
of more than 6 × 103, whereas the kcat/Km for L-Kyn is reduced
only 9-fold. Hence, the hydrogen bond formed between the
3-OH group in 3-OH-Kyn and the carbonyl oxygen of Asn-
333* appears to be an important determinant for the 3-OH-
Kyn reaction. It is important to note that the Km and Vmax values
of 3-OH-Kyn for N333T and S332G/N333T mutants could not
be accurately determined, as rates for high concentrations of
3-OH-Kyn showed significant substrate inhibition. Thus, Km
values for N333T and S332G/N333T HKynase are estimated
as greater than the highest 3-OH-Kyn concentration at which
hydrolysis rates were still increasing in a linear fashion. The
kcat/Km values for these mutants were calculated from the second-
order rate constant obtained from the linear portion of a V vs
[S] plot. Subsequently, the estimated Km and calculated kcat/Km
values were used to estimate the upper limit of kcat. N333T and
S332G/N333T mutant HKynases show a very slight preference
for L-Kyn over 3-OH-Kyn (Table 2).
soluble cellular extract and appeared entirely in the insoluble
protein fraction. We did not attempt to solubilize and refold
protein from the inclusion bodies that resulted from these
cultures. This protein in the insoluble pellet was of identical
apparent SDS-PAGE molecular weight as wild-type HKynase.
However, combination of the H102W and N333T mutations in
the H102W/N333T double mutant restored the expression of
soluble protein to levels similar to that of wild-type HKynase.
These results are consistent with the observation that kynureni-
nases contain pairings of residues at these positions that are
consistently either His/Asn or Trp/Thr, as can be seen in Figure
3. Characterization of the HKynase mutant H102W/N333T
revealed a small decrease in kcat and Km values for L-Kyn so
that the kcat/Km value only decreases about 2.5-fold. However,
the H102W/N333T mutant HKynase did not exhibit any
measurable activity with 600 µM 3-OH-Kyn and 37.2 µM
HKynase. Thus, these values were used to estimate a theoretical
upper limit for kcat/Km by assuming 3-OH-Kyn Km g 600 µM
and Vmax e 0.0001 absorbance min-1, the minimal rate that could
be measured. The net effect of the H102W/N333T double
mutation is a decrease in catalytic efficiency toward 3-OH-Kyn
Isolation and characterization of H102W mutant HKynase
was not possible because it could not be purified from the

394 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 2
Lima et al.
by a factor of at least 7.5 × 106 when compared to that of wild-
type HKynase. Thus, removing the hydrogen bonding partner
at the δ-position of Asn-333, increasing the local nonpolar
environment, and introducing a large bulky side chain (Trp) at
position 102 effectively act to exclude any 3-OH-Kyn binding
at the concentrations measured. This suggests that residues Asn-
333 and His-102 are the primary substrate specificity contacts
for the 3-substituent on the kynurenine aromatic ring. Although
the overall ratio of substrate specificity for 3-OH-Kyn/L-Kyn
is reversed in the H102W/N333T mutant, this mutant is 22-
fold less efficient at hydrolyzing L-Kyn than the wild-type
enzyme. This suggests that a surrounding shell of residues
contribute to catalytic efficiency. To further probe which
residues contribute to catalytic efficiency, we introduced a third
mutation, S332G, a residue showing a conserved pattern among
orthologues utilizing L-Kyn or 3-OH-Kyn similar to that of His-
102 and Asn-333 (Figure 3). In most 3-OH-Kyn hydrolyzing
kynureninases, a Ser residue is conserved at this position,
whereas the enzymes from organisms lacking the 3-OH-Kyn
metabolite contain a conserved Gly (Figure 3). In the HKynase-5
complex, Ser-332 does not form hydrogen bonds or have
electrostatic contacts with any atoms of 5. Yet, Ser-332 forms
a hydrogen bond through the backbone amide nitrogen with
one of the PLP phosphate oxygens. Characterization of the
H102W/S332G/N333T triple mutant showed it has the strongest
activity and catalytic efficiency toward L-Kyn of all Hkynase
mutants studied. Furthermore, it showed no detectable activity
with 3-OH-Kyn (Table 2) and has a 6-fold improvement in kcat/
Km over the H102W/N333T mutant Hkynase. Thus, this
mutation restores catalytic efficiency with L-Kyn to one-third
of that measured with wild-type enzyme. In contrast, the S332G/
N333T double mutant shows no improvement in kcat/Km over
that measured in the N333T single mutant. From these results,
we infer that the increased backbone flexibility provided by Gly-
332 facilitates the interaction between Thr-333 and the bulky
side chain of Trp-102, allowing a better fit of L-Kyn within the
mutant active site cavity. This result suggests that two shells of
residues surrounding the ligand atoms determine catalytic
efficiency in kynureninases. One shell of residues directly
contacts ligand atoms, providing essential acid/base, steric, and
electrostatic interactions, and the other contributes remotely to
the geometric arrangement of the active site and plays a dynamic
role in catalysis. Interestingly, kynureninase from Ralstonia
metallodurans was reported to have comparable activity with
both L-Kyn and 3-OH-Kyn.28 Yet, as shown in Figure 3, this
kynureninase has an active site that contains the Trp/Gly/Thr
pairing, further emphasizing the role played by outer shell
residues in substrate specificity and catalysis. It will be of interest
to use the HKynase H102W/S332G/N333T triple mutant for
directed evolution experiments to identify residues remote from
the active site which contribute to specificity and catalysis.
of Ile-110*, Trp-305*, Phe-306*, and Phe-314*. HKynase-5
interactions clearly show that a strong quadrupole molecular
alignment occurs between the side chains of Trp-305*, Phe-
314*, and the aromatic ring moiety of 5, as well as ligand
π-stacking against the side chain of His-102*. Third, electrostatic
and/or hydrogen bonding interactions should be assigned
between ligand atoms and the guanidino nitrogens of Arg-434
and the imidazole nitrogen atoms of His-253. Fourth, our results
clearly show that His-102 and Asn-333 are the residues involved
in substrate specificity for 3-OH-Kyn. As such, the interaction
between these residues and ligand atoms should be used as a
one of the primary forms of binding mode evaluation in docking
studies. Furthermore, scoring of docked poses should consider
the interaction between ligand atoms and Arg-434 side chain
atoms as a strong indicator of acceptable docking because this
form of stabilization is known to be crucial in ligand binding
and stabilization throughout the PLP-dependent aspartate ami-
notransferase family of enzymes. Finally, we note that 5 is an
ideal lead compound for combinatorial synthesis of libraries of
possible HKynase inhibitors.
Experimental Section
Crystallization, Data Collection, and Molecular Replacement.
Human kynureninase was expressed and purified as previously
described.18 Complexed kynureninase-3-hydroxyhippuric acid crys-
tals were grown using the modified microbatch under oil technique
at 25 °C by mixing (1:1 ratio) an HKynase solution (9 mg mL-1 in
50 mM HEPES, pH 5.1, and 0.1 mM PLP) with a crystallization
solution containing 0.05 M MgCl2, 0.1 M Tris (pH 8.0), 25% PEG
3000, and 350 µM 5. Dark-yellow crystals appeared after 4-5
weeks and grew to dimensions of 0.075 mm × 0.05 mm × 0.015
mm. These crystals were flash frozen in liquid nitrogen with
cryoprotectant containing 1 mM 5, 55 mM MgCl2, 110 mM Tris
(pH 8), 33% PEG 3000. These crystals had space group C2 with
cell constants a ) 74.44 Å, b ) 77.12 Å, c ) 94.55 Å, ꢀ ) 109.35.
X-ray synchrotron data were collected (λ ) 1.007 Å, 200 frames,
1° oscillations) at the Advanced Photon Source beamline 19-ID
and were processed and scaled with HKL3000.34 The merged
SCALEPACK34 intensities were used as input for the molecular
replacement program Phaser35 using the HKynase (PDB accession
code 2HZP) (HKynase) coordinates as the phasing model. All water,
hetero, and PLP coordinates were deleted from the input model.
The molecular replacement solution and Hkynase-5 complex were
refined with Refmac5.36 Water addition was done with ARP/
wARP.37 The PLP cofactor and compound 5 were manually
introduced with COOT.38 TLS39 refinement was performed with
the following groups: 6-37, 38-45, 46-120, 121-190, 191-213,
214-357, 358-375, 376-412, 413-460. The model of 5 and
library description were created with the CCP4i Monomer Library
Sketcher.40 Molprobity41 was used to evaluate the quality of the
final model.
Inhibition Kinetics. Compound 5 was evaluated for its ability
to inhibit the reaction of HKynase with its cognate substrate, 3-OH-
Kyn. Assays were performed by following the absorbance change
at 370 nm as 3-OH-Kyn (DL mixture, USBiochemical Corp.) is
converted to 3-hydroxyanthranilic acid and L-Alanine (∆ε ) -4500
M-1 cm-1) with a Cary 1 UV/visible spectrophotometer equipped
with a Peltier temperature controlled cell 6 × 6 cell compartment.
Assays contained 30 mM potassium phosphate, pH 8.0, and 40 µM
PLP at 37 °C together with variable amounts of 3-OH-Kyn and
four different concentrations of 5. Data were fit to a competitive
inhibition model with Cleland’s COMPO program.42
Site Directed Mutagenesis. Mutants were created using the
QuickChange site-directed mutagenesis kit (Stratagene). The fol-
lowing mutants were created and sequenced to confirm the desired
codon change: H102W, N333T, S332G/N333T (double mutant),
H102W/N333T (double mutant), H102W/S332G/N333T (triple
mutant). The mutagenesis primers used were (only coding strand
Currently, a number of kynureninase inhibitors have been
reported in the literature.20,21,25,29-33 Until now, their design
has been based primarily on mechanistic considerations due to
a lack of knowledge regarding the primary substrate binding
mode and substrate-enzyme interactions required for in silico
directed drug design. The HKynase-5 complex reveals impor-
tant restraints for the design and screening of novel substrate
analogues and inhibitors. First, the following atoms should be
designated as potential hydrogen bonding partners for ligand
atoms: the side chain δ-oxygen of Asn-333*, the side chain
hydroxyl oxygen of Tyr-275, and the γ-oxygen of Ser-75.
Second, hydrophobic interactions should be maintained between
aromatic substituents on screened ligands and the side chains

HKynase-3-Hydroxyhippuric Acid Inhibitor Complex
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 2 395
written, mutated bases bold underlined): H102W: 5′-CCA AAA
TAG CAG CCT ATG GTT GGG AAG TGG GGA AGC GTC
CTT GG-3′; S332G/N333T: 5′-GGG GTC TGT GGA TTC CGA
ATT GGC ACC CCT CCC ATT TTG TTG GTC-3′; N333T: 5′-
GTC TGT GGA TTC CGA ATT TCA ACC CCT CCC ATT TTG
TTG G-3′.
References
(1) Soda, K.; Tanizawa, K. Kynureninases: enzymological properties and
regulation mechanism. AdV. Enzymol. Relat. Area Mol. Biol. 1979,
49, 1–40.
(2) Colabroy, K. L.; Begley, T. P. The pyridine ring of NAD is formed
by a nonenzymatic pericyclic reaction. J. Am. Chem. Soc. 2005, 127,
840–841.
(3) Schwarcz, R.; Whetsell, W. O., Jr.; Mangano, R. M. Quinolinic acid:
an endogenous metabolite that produces axon-sparing lesions in rat
brain. Science 1983, 219, 316–318.
(4) Stone, T. W.; Perkins, M. N. Quinolinic acid: a potent endogenous
excitant at amino acid receptors in CNS. Eur. J. Pharmacol. 1981,
72, 411–412.
(5) Heyes, M. P.; Brew, B.; Martin, A.; Markey, S. P.; Price, R. W.; Bhalla,
R. B.; Salazar, A. Cerebrospinal fluid quinolinic acid concentrations
are increased in acquired immune deficiency syndrome. AdV. Exp. Med.
Biol. 1991, 294, 687–690.
(6) Achim, C. L.; Heyes, M. P.; Wiley, C. A. Quantitation of human
immunodeficiency virus, immune activation factors, and quinolinic
acid in AIDS brains. J. Clin. InVest. 1993, 91, 2769–2775.
(7) Heyes, M. P.; Swartz, K. J.; Markey, S. P.; Beal, M. F. Regional brain
and cerebrospinal fluid quinolinic acid concentrations in Huntington’s
disease. Neurosci. Lett. 1991, 122, 265–269.
Synthesis of 3-Hydroxyhippuric Acid (5). 3-Ethoxycarbon-
yloxybenzaldehyde (2).43 3-Hydroxybenzaldehyde (1; 15.0 g, 0.123
mol) was dissolved in 100 mL dry pyridine, cooled to 4 °C, and
ethyl chloroformate (20 mL) was added dropwise over a period of
30 min. The resulting solution was stirred for 2 h at room
temperature. The solvent was evaporated in vacuo, and water (150
mL) was added to the residue. The product was extracted into ether,
and the extract was washed consecutively with water, 5% HC1,
5% cold NaOH, and again with water. The dried organic extract
was evaporated in vacuo to give a dark-red viscous product (15.5
1
g, 65%). H NMR (400 MHz, CDCl3) δ 1.29 (t, 3H), 4.21 (m,
2H), 7.96 (dd, 1H), 7.57 (d, 1H), 7.72 (d, 1H), 7.74 (m, 1H), 9.61
(s, 1H).
3-Ethoxycarbonyloxybenzoic acid (3).44 3-Ethoxycarbonyloxy-
benzaldehyde (6 g, 0.03 mol) was dissolved in 30 mL DMF. Oxone
(19 g, 0.03 mol) was added, and the mixture was stirred at room
temperature for 3 h. The reaction was monitored by TLC and, after
completion, 1 N HCl was added to dissolve the salts. The crude
product was extracted into ethyl acetate (50 mL), the organic extract
washed with 1N HCl (3 × 15 mL), brine (2 × 10 mL), and then
dried over Na2SO4. Evaporation of the solvent in vacuo gave
3-ethoxycarbonyloxybenzoic acid (3), which was purified by silica
gel column chromatography (60:40 EtOAc/hexane). Yield, 3.86 g
(63%). 1H NMR (400 MHz, CDCl3) δ 1.29 (t, 3H), 4.21 (m, 2H),
7.93 (dd, 1H), 7.63 (d, 1H), 8.02 (d, 1H), 8.06 (m, 1H), 12.54 (s,
1H).
(8) Giorgini, F.; Moller, T.; Kwan, W.; Zwilling, D.; Wacker, J. L.; Hong,
S.; Tsai, L. C.; Cheah, C. S.; Schwarcz, R.; Guidetti, P.; Muchowski,
P. J. Histone deacetylase inhibition modulates kynurenine pathway
activation in yeast, microglia, and mice expressing a mutant huntingtin
fragment. J. Biol. Chem. 2008, 283, 7390–7400.
(9) Guillemin, G. J.; Meininger, V.; Brew, B. J. Implications for the
kynurenine pathway and quinolinic acid in amyotrophic lateral
sclerosis. Neurodegener. Dis. 2005, 2, 166–176.
(10) Leonard, B. E. Inflammation, depression and dementia: are they
connected. Neurochem. Res. 2007, 32, 1749–1756.
(11) Guillemin, G. J.; Brew, B. J.; Noonan, C. E.; Takikawa, O.; Cullen,
K. M. Indoleamine 2,3 dioxygenase and quinolinic acid immunore-
activity in Alzheimer’s disease hippocampus. Neuropathol. Appl.
Neurobiol. 2005, 31, 395–404.
(12) Stone, T. W.; Mackay, G. M.; Forrest, C. M.; Clark, C. J.; Darlington,
L. G. Tryptophan metabolites and brain disorders. Clin. Chem. Lab.
Med. 2003, 41, 852–859.
(13) Schwarcz, R.; Pellicciari, R. Manipulation of brain kynurenines: glial
targets, neuronal effects, and clinical opportunities. J. Pharmacol. Exp.
Ther. 2002, 303, 1–10.
(14) Stone, T. W.; Addae, J. I. The pharmacological manipulation of
glutamate receptors and neuroprotection. Eur. J. Pharmacol. 2002,
447, 285–296.
(15) Tankiewicz, A.; Pawlak, D.; Buczko, W. Enzymes of the kynurenine
pathway. Postepy. Hig. Med. Dosw. 2001, 55, 715–731.
(16) Shetty, A. S.; Gaertner, F. H. Kynureninase-Type enzymes of
Penicillum roqueforti, Aspergillus niger, Rhizopus stolonifer, and
Pseudomonas fluorescens: further evidence for distinct kynureninase
and hydroxykynureninase activities. J. Bacteriol. 1975, 122, 235–244.
(17) Toma, S.; Nakamura, M.; Tone, S.; Okuno, E.; Kido, R.; Breton, J.;
Avanzi, N.; Cozzi, L.; Speciale, C.; Mostardini, M.; Gatti, S.; Benatti,
L. Cloning and recombinant expression of rat and human kynureninase.
FEBS Lett. 1997, 408, 5–10.
3-Ethoxycarbonyloxybenzoyl chloride (4). To 3.0 g (0.15 mol)
of 3-ethoxycarbonyloxybenzoic acid in dry dichloromethane, one
equiv of 0.2 M oxalyl chloride in dichloromethane was added
dropwise, and the mixture was stirred for 5 h at room temperature.
The crude 3-methoxycarbonyloxybenzoyl chloride (2.35 g, 68%)
1
was obtained after evaporation of excess solvent and reagent. H
NMR(400 MHz, CDCl3) δ 1.29 (t, 3H), 4.21 (m, 2H), 7.93 (dd,
1H), 7.66 (d, 1H), 7.96 (d, 1H), 8.00 (m, 1H).
3-Hydroxyhippuric acid (5).45 The finely powdered 3-ethoxy-
carbonyloxybenzoyl chloride (1.7 g, 0.007 mol) was added to 0.6 g
of glycine in 20 mL of 0.5 N NaOH. Subsequently, 1N NaOH was
continuously added to the solution over a period of 15 min to
maintain the pH above 10. The reaction mixture was then acidified
with conc. HCl, and 5 was extracted with ethyl acetate (3 × 15
mL). The organic extracts were dried and evaporated, and the crude
product was recrystallized from 50% ethanol-water. Yield, 0.8 g
(18) Lima, S.; Khristoforov, R.; Momany, C.; Phillips, R. S. Crystal
structure of Homo sapiens kynureninase. Biochemistry 2007, 46, 2735–
2744.
1
(48.3%), mp 184-187 °C. H NMR (DMSO-d6) δ 3.87 (d, 2H,
CH2), 6.88-6.91 (m, 3H, Ar-H), 7.21-7.25 (m, 1H, Ar-H), 8.68
(t, 1H, NH), 9.60 (s, 1H, OH), 12.60 (s, 1H, CO2H). MS, anion
mode, direct injection ESI (M - H)- ) 194.0 (100%); (2M -
H)- ) 389.2 (36%).
(19) Jager, J.; Moser, M.; Sauder, U.; Jansonius, J. N. Crystal structures
of Escherichia coli aspartate aminotransferase in two conformations.
Comparison of an unliganded open and two liganded closed forms. J.
Mol. Biol. 1994, 239, 285–305.
(20) Chiarugi, A.; Carpenedo, R.; Molina, M. T.; Mattoli, L.; Pellicciari,
R.; Moroni, F. Comparison of the neurochemical and behavioral effects
resulting from the inhibition of kynurenine hydroxylase and/or
kynureninase. J. Neurochem. 1995, 65, 1176–1183.
(21) Walsh, H. A.; Leslie, P. L.; O’Shea, K. C.; Botting, N. P. 2-Amino-
4-[3′-hydroxyphenyl]-4-hydroxybutanoic acid; a potent inhibitor of rat
and recombinant human kynureninase. Bioorg. Med. Chem. Lett. 2002,
12, 361–363.
(22) Dunathan, H. C. Conformation and reaction specificity in pyridoxal
phosphate enzymes. Proc. Natl. Acad. Sci. U.S.A. 1966, 55, 712–716.
(23) Delle Fratte, S.; Iurescia, S.; Angelaccio, S.; Bossa, F.; Schirch, V.
The function of arginine 363 as the substrate carboxyl-binding site in
Escherichia coli serine hydroxymethyltransferase. Eur. J. Biochem.
1994, 225, 395–401.
Acknowledgment. Results shown in this report are derived
from work performed at Argonne National Laboratory, Struc-
tural Biology Center at the Advanced Photon Source. Argonne
is operated by University of Chicago Argonne, LLC, for the
U.S. Department of Energy, Office of Biological and Environ-
mental Research under contract DE-AC02-06CH11357. Use of
the Advanced Photon Source was supported by the U.S.
Department of Energy, O. o. S., Office of Basic Energy
Sciences, under contract no. W-31-109-Eng-38.
(24) Almo, S. C.; Smith, D. L.; Danishefsky, A. T.; Ringe, D. The structural
basis for the altered substrate specificity of the R292D active site
mutant of aspartate aminotransferase from E. coli. Protein Eng. 1994,
7, 405–412.
Supporting Information Available: 1H NMR, MS, and HPLC
of 5. This material is available free of charge via the Internet at
http://pubs.acs.org.

396 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 2
Lima et al.
(25) Dua, R.; Phillips, R. S. Stereochemistry and mechanism of aldol
reactions catalyzed by kynureninase. J. Am. Chem. Soc. 1991, 113,
7385–7388.
(26) Phillips, R. S.; Sundararaju, B.; Koushik, S. V. The Catalytic
Mechanism of Kynureninase from Pseudomonas fluorescens: Evidence
for Transient Quinonoid and Ketimine Intermediates from Rapid-
scanning Stopped-flow Spectrophotometry. Biochemistry 1998, 37,
8783–8789.
(27) Chen, H. Y.; Demidkina, T. V.; Phillips, R. S. Site-directed Mutagen-
esis of Tyrosine-71 to Phenylalanine in Citrobacter freundii Tyrosine
Phenol-lyase: Evidence for Dual Roles of Tyrosine 71 as a General
Acid Catalyst in the Reaction Mechanism and in Cofactor Binding.
Biochemistry 1995, 34, 12276–12283.
(28) Kurnasov, O.; Goral, V.; Colabroy, K.; Gerdes, S.; Anantha, S.;
Osterman, A.; Begley, T. P. NAD biosynthesis: identification of the
tryptophan to quinolinate pathway in bacteria. Chem. Biol. 2003, 10,
1195–1204.
(29) Dua, R. K.; Taylor, E. W.; Phillips, R. S. S-Aryl-L-cysteine S,S-
Dioxides: Design and evaluation of a new class of mechanism
based inhibitors of kynureninase. J. Am. Chem. Soc. 1993, 115, 1264–
1270.
(30) Walsh, H. A.; O’Shea, K. C.; Botting, N. P. Comparative inhibition
by substrate analogues 3-methoxy- and 3-hydroxydesaminokynurenine
and an improved 3 step purification of recombinant human kynureni-
nase. BMC Biochem. 2003, 4, 13.
(31) Heiss, C.; Anderson, J.; Phillips, R. S. Differential effects of
bromination on substrates and inhibitors of kynureninase from
Pseudomonas fluorescens. Org. Biomol. Chem. 2003, 1, 288–295.
(32) Fitzgerald, D. H.; Muirhead, K. M.; Botting, N. P. A comparative
study on the inhibition of human and bacterial kynureninase by novel
bicyclic kynurenine analogues. Bioorg. Med. Chem. 2001, 9, 983–
989.
(33) Carpenedo, R.; Chiarugi, A.; Russi, P.; Lombardi, G.; Carla, V.;
Pellicciari, R.; Mattoli, L.; Moroni, F. Inhibitors of kynurenine
hydroxylase and kynureninase increase cerebral formation of kynurenate
and have sedative and anticonvulsant activities. Neuroscience 1994,
61, 237–243.
(34) Otwinowski, Z.; W., M. Processing of X-ray Diffraction Data Collected
in Oscillation Mode. Methods Enzymol. 1997, 276, 307–326.
(35) McCoy, A. J.; Grosse-Kunstleve, R. W.; Storoni, L. C.; Read, R. J.
Likelihood-enhanced fast translation functions. Acta Crystallogr., Sect.
D: Biol. Crystallogr. 2005, 61, 458–464.
(36) Murshudov, G. N.; Vagin, A. A.; Dodson, E. J. Refinement of
macromolecular structures by the maximum-likelihood method. Acta
Crystallogr., Sect. D: Biol. Crystallogr. 1997, 53, 240–255.
(37) Lamzin, V. S.; Perrakis, A.; Wilson, K. S. The ARP/WARP suite for
automated construction and refinement of protein models. In Inter-
national Tables for Crystallography, Volume F; Kluwer Academic
Publishers: Dordrecht, The Netherlands, 2001.
(38) Emsley, P.; Cowtan, K. Coot: model-building tools for molecular
graphics. Acta Crystallogr., Sect, D: Biol. Crystallogr. 2004, 60, 2126–
2132.
(39) Winn, M. D.; Isupov, M. N.; Murshudov, G. N. Use of TLS parameters
to model anisotropic displacements in macromolecular refinement. Acta
Crystallogr., Sect. D: Biol. Crystallogr. 2001, 57, 122–133.
(40) Collaborative Computational Project, N. The CCP4 Suite: Programs
for Protein Crystallography Acta Crystallogr., Sect. D: Biol. Crystal-
logr. 1994, 50, 760-763.
(41) Davis, I. W.; Leaver-Fay, A.; Chen, V. B.; Block, J. N.; Kapral, G. J.;
Wang, X.; Murray, L. W.; BryanArendall, W., III.; Snoeyink, J.;
Richardson, J. S.; Richardson, D. C. MolProbity: all-atom contacts
and structure validation for proteins and nucleic acids. Nucleic Acids
Res. 2007, 35, W375–W383.
(42) Cleland, W. W. Statistical analysis of enzyme kinetic data. Methods
Enzymol. 1979, 63, 103–138.
(43) Dalgliesh, C. E.; Mann, F. G. Synthesis of ꢀ-(3,4-dihydroxyphenyl)-
N-methylserine (adrenalinecarboxylic acid). J. Chem. Soc. 1947, 658–
662.
(44) Travis, B. R. S., M.; Hollist, G. O.; Borhan, B. Facile Oxidation of
Aldehydes to Acids and Esters with Oxone. Org. Lett. 2003, 5, 1031–
1034.
(45) Acheson, R. M. B. D. A.; Brettle, R.; Harris, A. M. Synthesis of some
acylglycines and related oxazolones. J. Chem. Soc. 1960, 3457–3461.
JM8010806