The phosphate of pyridoxal-5′-phosphate is an acid/base catalyst in the mechanism of Pseudomonas fluorescens kynureninase

Article abstract of DOI:10.1111/febs.12671

Kynureninase (l-kynurenine hydrolase, EC 3.7.1.3) catalyzes the hydrolytic cleavage of l-kynurenine to l-alanine and anthranilic acid. The proposed mechanism of the retro-Claisen reaction requires extensive acid/base catalysis. Previous crystal structures showed that Tyr226 in the Pseudomonas fluorescens enzyme (Tyr275 in the human enzyme) hydrogen bonds to the phosphate of the pyridoxal-5′-phosphate (PLP) cofactor. This Tyr residue is strictly conserved in all sequences of kynureninase. The human enzyme complexed with a competitive inhibitor, 3-hydroxyhippuric acid, showed that the ligand carbonyl O is located 3.7 A from the phenol of Tyr275 (Lima, S., Kumar, S., Gawandi, V., Momany, C. & Phillips, R. S. (2009) J. Med. Chem. 52, 389-396). We prepared a Y226F mutant of P. fluorescens kynureninase to probe the role of this residue in catalysis. The Y226F mutant has approximately 3000-fold lower activity than wild-type, and does not show the pK_a values of 6.8 on k_cat and 6.5 and 8.8 on k_cat/K_m seen for the wild-type enzyme (Koushik, S. V., Moore, J. A. III, Sundararaju, B. & Phillips, R. S. (1998) Biochemistry 37, 1376-1382). Wild-type kynureninase shows a resonance at 4.5 ppm in ³¹P-NMR, which is shifted to 5.0, 3.3 and 2.0 ppm when the potent inhibitor 5-bromodihydrokynurenine is added. However, Y226F kynureninase shows resonances at 3.6 and 2.5 ppm, and no change in the peak position is seen when 5-bromodihydrokynurenine is added. Taken together, these results suggest that Tyr226 mediates proton transfer between the substrate and the phosphate, which accelerates formation of external aldimine and gem-diol intermediates. Thus, the phosphate of PLP acts as an acid/base catalyst in the mechanism of kynureninase.

Full text of DOI:10.1111/febs.12671

The phosphate of pyridoxal-5′-phosphate is an acid/base
catalyst in the mechanism of Pseudomonas fluorescens
kynureninase
1
,2
1
2
2
1
Robert S. Phillips , Israel Scott , Riya Paulose , Akshay Patel and Taylor Colt Barron
1
2
Department of Biochemistry and Molecular Biology, University Georgia, Athens, GA, USA
Department of Chemistry, University Georgia, Athens, GA, USA
Keywords
1
Kynureninase (L-kynurenine hydrolase, EC 3.7.1.3) catalyzes the hydrolytic
cleavage of L-kynurenine to L-alanine and anthranilic acid. The proposed
mechanism of the retro-Claisen reaction requires extensive acid/base cataly-
sis. Previous crystal structures showed that Tyr226 in the Pseudomonas
fluorescens enzyme (Tyr275 in the human enzyme) hydrogen bonds to the
phosphate of the pyridoxal-5′-phosphate (PLP) cofactor. This Tyr residue
is strictly conserved in all sequences of kynureninase. The human enzyme
complexed with a competitive inhibitor, 3-hydroxyhippuric acid, showed
3
P-NMR; acid–base catalysis; enzyme
mechanism; mutagenesis; pyridoxal-5′-
phosphate
Correspondence
R. S. Phillips, Department of Chemistry,
University of Georgia, Athens, GA 30602-
2
556, USA
Fax: +1 706 542 9454
Tel: +1 706 542 1996
E-mail: plp@uga.edu
ꢀ
that the ligand carbonyl O is located 3.7 A from the phenol of Tyr275
(
Lima, S., Kumar, S., Gawandi, V., Momany, C. & Phillips, R. S. (2009)
J. Med. Chem. 52, 389–396). We prepared a Y226F mutant of P. fluores-
cens kynureninase to probe the role of this residue in catalysis. The Y226F
mutant has approximately 3000-fold lower activity than wild-type, and
(
Received 18 July 2013, revised 27
November 2013, accepted 2 December
013)
2
does not show the pK values of 6.8 on kcat and 6.5 and 8.8 on kcat/K
a m
seen for the wild-type enzyme (Koushik, S. V., Moore, J. A. III, Sun-
dararaju, B. & Phillips, R. S. (1998) Biochemistry 37, 1376–1382). Wild-
doi:10.1111/febs.12671
3
1
type kynureninase shows a resonance at 4.5 ppm in P-NMR, which is
shifted to 5.0, 3.3 and 2.0 ppm when the potent inhibitor 5-
bromodihydrokynurenine is added. However, Y226F kynureninase shows
resonances at 3.6 and 2.5 ppm, and no change in the peak position is seen
when 5-bromodihydrokynurenine is added. Taken together, these results
suggest that Tyr226 mediates proton transfer between the substrate and the
phosphate, which accelerates formation of external aldimine and gem-diol
intermediates. Thus, the phosphate of PLP acts as an acid/base catalyst in
the mechanism of kynureninase.
Introduction
Kynureninase (L-kynurenine hydrolase, EC 3.7.1.3) is
a pyridoxal-5′-phosphate (PLP)-dependent enzyme that
catalyzes the hydrolytic C –C cleavage of L-kynure-
b
c
nine to alanine and anthranilic acid (Eqn 1) [1]:
and is also found in a number of bacteria. However,
the eukaryotic enzyme has a reaction preference for 3-
hydroxykynurenine, while most bacterial enzymes react
preferentially with kynurenine [1].
This is one of a small group of enzymes that cata-
lyze a retro-Claisen reaction. The enzyme is part of the
major eukaryotic pathway for tryptophan catabolism,
Abbreviation
PLP, pyridoxal-5′-phosphate.
1
100
FEBS Journal 281 (2014) 1100–1109 ª 2013 FEBS

R. S. Phillips et al.
Acid/base catalysis in kynureninase
The reaction of kynureninase is relatively unusual
among PLP-dependent enzymes in that it occurs with
electrophilic rather than nucleophilic substitution at
The crystal structure of the complex of human ky-
nureninase bound to a competitive inhibitor, 3-hydrox-
yhippuric acid, shows the carbonyl oxygen of the
inhibitor located near the strictly conserved residues
C . The proposed mechanism of kynureninase, shown
b
ꢀ
Tyr275 and Ser75, with 3.7 A between the carbonyl O
in Scheme 1, requires extensive acid/base catalysis [2–
6
]. Initially, trans-aldimination of the substrate with
and the Tyr275 phenolic O [7]. Tyr275 also donates a
hydrogen bond to the phosphate of the PLP [8], as
does the homologous Tyr226 in Pseudomonas fluores-
cens kynureninase [9]. We hypothesized from the struc-
tural data that this Tyr may act as a shuttle to
transfer a proton from PLP to the substrate c-car-
bonyl oxygen, to accelerate formation of the gem-diol.
Hence, we mutated Tyr226 in P. fluorescens kynure-
ninase to phenylalanine to assess the role of this
the internal aldimine gives the external aldimine. De-
protonation of Ca then forms a quinonoid structure,
which is reprotonated at C-4′ to give the kynurenine
ketimine. Subsequent addition of water to the c-car-
bonyl of the ketimine intermediate forms a gem-diol,
which then undergoes C –C bond cleavage to give
the carboxylic acid product and an enamine intermedi-
ate. Protonation of the enamine on Cb gives a pyru-
vate ketimine, which is deprotonated on C-4′ to give
the L-alanine quinonoid structure. Finally, protonation
of the L-alanine quinonoid intermediate on Ca gives
the L-alanine aldimine, which releases the L-alanine
product.
b
c
3
1
tyrosine in catalysis, and we performed P-NMR on
wild-type and Y226F kynureninase to examine the ion-
ization state of the phosphate in the resting enzyme
and complexed with a potent inhibitor. The results of
these experiments are reported here.
Scheme 1. Proposed mechanism for kynureninase.
FEBS Journal 281 (2014) 1100–1109 ª 2013 FEBS
1101

Acid/base catalysis in kynureninase
R. S. Phillips et al.
Results
k
curve, with a base with a pK of 6.5 and an acid with
cat/K
m
for wild-type kynureninase has a bell-shaped
a
Kinetic properties of Y226F kynureninase
a pK of 8.8 [3]. The pH dependence of kcat/K for the
a m
reaction with 3-hydroxykynurenine, a slower substrate,
gave a pK of 6.8 for the basic group [3]; hence, ky-
Y226F mutant kynureninase has very low catalytic
activity compared to wild-type enzyme, with a kcat that
is reduced approximately 2800-fold, and a kcat/Km that
is reduced approximately 376-fold (Table 1). The bind-
ing of PLP to Y226F kynureninase is not strongly
affected, as the isolated enzyme contains 1 mol PLP/
subunit. However, the apparent Km of Y226F kynure-
ninase for PLP under assay conditions is 21 lM,
compared with 0.6 lM for the wild-type enzyme [9].
Wild-type kynureninase normally shows no significant
increase in initial rate with added PLP, with the func-
tion of the PLP being primarily to increase the length
of the linear portion of the reaction because of com-
peting abortive trans-amination of the L-alanine prod-
uct [10]. The reaction of Y226F kynureninase shows
a
nurenine is a ‘sticky substrate’. Although we attempted
to fit the data in Fig. 1 to a pH dependence similar to
wild-type kynureninase, the results were not reliable,
and a simple linear regression fit is more reasonable,
as may be seen from the lines in Fig. 1A,B. Thus, it
may be concluded that the Y226F mutation strongly
affects the pH dependence of the reaction, and the
conjugate base with a pK of 6.5 in wild-type kynure-
a
ninase must be either Tyr226 or a group closely associ-
ated with it.
We also examined the reaction of wild-type and
Y226F kynureninase with O-benzoyl-L-serine. O-ben-
zoyl-L-serine is an alternate substrate for kynureninase,
which undergoes a normal b-elimination reaction to
give benzoate, pyruvate and ammonium, as previously
shown for tyrosine phenol lyase and tryptophan indole
lyase [11]. The formation of pyruvate was followed
no significant pH dependence for either Vmax or Vmax/
Km (Fig. 1). It was shown previously that wild-type
kynureninase has a single base, with a pK of 6.8, in
a
the pH dependence of kcat [3]. The pH dependence of
Table 1. Kinetic parameters for wild-type and Y226F kynureninase.
Wild-type
Y226F
ꢀ
1
ꢀ1 ꢀ1
ꢀ1
ꢀ1 ꢀ1
(M
Compound
k
cat (s
a
)
k
cat/K
m
(M
s
)
K
i
(nM)
k
cat (s
)
k
cat/K
m
s
)
K
i
(nM)
5 a
ꢀ3
L-kynurenine
16
2.0 9 10
–
–
(5.8 ꢁ 0.3) 9 10
530 ꢁ 85
41 ꢁ 11
–
–
–
4
O-benzoyl-L-serine
S-(2-aminophenyl)-L-cysteine
dioxide
2.5 ꢁ 0.2
(1.1 ꢁ 0.1) 9 10
0.10 ꢁ 0.02
a
–
–
30
–
920 ꢁ 240
a
Data from reference 13.
A
B
Fig. 1. pH dependence of Y226F
kynureninase. (A) pH dependence of Vmax
values. The line is a linear regression fit to
the data. (B) pH dependence of Vmax/K
m
values. The line is a linear regression fit to
the data.
1
102
FEBS Journal 281 (2014) 1100–1109 ª 2013 FEBS

R. S. Phillips et al.
Acid/base catalysis in kynureninase
using a coupled assay with lactate dehydrogenase and
NADH. b-substituted amino acids with good leaving
groups, such as b-chloroalanine, were previously
shown to undergo b-elimination reactions with kynure-
ninase [12]. However, O-benzoyl-L-serine is a better
substrate for kynureninase than the compounds stud-
ied previously, probably due to its closer structural
similarity to kynurenine. The reaction of O-benzoyl-L-
serine is less affected than L-kynurenine by the Y226F
mutation, with the kcat value being reduced only 25-
fold and kcat/Km being reduced 280-fold (Table 1).
Finally, binding of a potent transition state analog
inhibitor, S-(2-aminophenyl)-L-cysteine S,S-dioxide
result in significant changes in the mutant enzyme
spectrum, in contrast to wild-type kynureninase.
31
P-NMR of kynureninase
3
1
The P-NMR spectrum of P. fluorescens kynureninase
exhibits a peak at 4.6 ppm (Fig. 3A, solid line) consis-
tent with a dianionic phosphate ester group in the rest-
ing enzyme, as is characteristic of PLP-dependent
enzymes [15,16]. The line width is approximately
23 Hz, consistent with restricted motion of the phos-
phate around the C–O bond. The peak position is at a
lower field than normal for PLP-dependent enzymes,
but is quite similar to what is observed for aspartate
aminotransferase and O-acetyl serine sulfhydrylase
[13,14], is weakened approximately 30-fold for the
Y226F mutant kynureninase (Table 1).
[
15,17]. However, when a potent transition-state ana-
log inhibitor, 5-bromodihydrokynurenine [13], is
added, the spectrum becomes more complex, with
peaks at approximately 2.0, 3.3 and 5.0 ppm (Fig. 3A,
dashed line). The upfield shift of the 2 ppm peak in
the ligand complex by approximately 3 ppm is consis-
tent with formation of a monoanionic phosphate upon
Spectrum of Y226F kynureninase
The UV-vis spectrum for Y226F kynureninase as iso-
lated shows an absorbance peak at 416 nm (Fig. 2),
i.e. slightly blue shifted compared to wild-type kynure-
ninase. Addition of the product, L-alanine, to Y226F
does not result in any immediate changes in the spec-
trum, in contrast to wild-type enzyme, but, upon incu-
bation, pyridoxamine-5′-phosphate, absorbing at
3
1
ligand binding [15,16]. In contrast, the P-NMR spec-
trum of Y226F kynureninase exhibits peaks at 2.5 and
3.6 ppm (Fig. 3B, solid line), which do not change sig-
nificantly upon addition of 5-bromodihydrokynurenine
(Fig. 3B, dashed line), although there is a change in
relative intensity.
3
25 nm, is formed slowly in an abortive trans-amina-
tion reaction (Fig. 2), as also seen with the wild-type
enzyme. The rate constant for the abortive trans-ami-
ꢀ
5
ꢀ1
nation, 7.3 9 10
s , is approximately 30-fold slower
for Y226F than for the wild-type enzyme, which shows
a t½ of 3–5 min for the abortive trans-amination of
alanine [10]. Addition of other substrates and inhibi-
tors, i.e. kynurenine [4], b-benzoylalanine [5] and S-(2-
aminophenyl)-L-cysteine S,S-dioxide [13,14] did not
Discussion
The mechanism proposed for kynureninase requires
nucleophilic addition of water to the c-carbonyl of the
substrate (Scheme 1), so it is likely that the enzyme will
Fig. 2. Visible spectra for Y226F
kynureninase. The solution contained
ꢀ1
1
mgꢂmL Y226F kynureninase in 0.05 M
potassium phosphate, pH 8.0, 100 mM L-
alanine. Spectra were collected at 30 min
intervals. Inset: Absorbance at 416 nm as
a function of incubation time. The line is
the result of fitting to a single exponential
decay.
FEBS Journal 281 (2014) 1100–1109 ª 2013 FEBS
1103

Acid/base catalysis in kynureninase
R. S. Phillips et al.
A
31
Fig. 3. P-NMR spectra for kynureninase.
A) Wild-type kynureninase. The
concentration of the enzyme was 1 mM in
.05 M bicine-Na, pH 8.0. Solid line, wild-
B
(
0
type enzyme; dashed line, 1 mM wild-type
enzyme with 2 mM 5-
bromodihydrokynurenine. (B) Y226F
kynureninase. The concentration of the
enzyme was 1.1 mM in 0.05 M bicine-Na,
pH 8.0. Solid line, Y226F mutant enzyme;
dashed line, 1 mM Y226F mutant enzyme
with 2 mM 5-bromodihydrokynurenine.
10
8
6
4
2
0
–2
–4
–6
–8
–10
ppm
polarize the carbonyl to increase its reactivity. In the
serine proteases, the reactive carbonyl of substrates is
located in an ‘oxyanion hole’, which is a pocket of
hydrogen bond donors that polarizes the carbonyl and
stabilizes the tetrahedral intermediate [18]. The phenolic
group of Tyr275 is located near the carbonyl oxygen of
3-hydroxyhippurate, a competitive inhibitor, that is
bound at the active site of human kynureninase [7], with
ꢀ
an O–O distance of 3.7 A (Fig. 4). This tyrosine residue
is located in the active site of kynureninase, in the
sequence just prior to the PLP-binding lysine, is strictly
conserved (Fig. 5), and was found to form a hydrogen
bond with the PLP phosphate in the crystal structures
of P. fluorescens and human kynureninase, with a tyro-
ꢀ
sine O–PO distance of 2.6 A (Fig. 4) [6,8]. This hydro-
4
gen bond is maintained in the structure of human
kynureninase with 3-hydroxyhippuric acid bound, as
may be seen from the position of the phenolic O of
Tyr275 in Fig. 4, located between the phosphate O and
the 3-hydroxyhippuric acid amide carbonyl O. Thus,
we postulated that this tyrosine residue (Tyr226 in
P. fluorescens kynureninase) may serve as part of an
oxyanion hole to facilitate addition of water to the
substrate c-carbonyl, by analogy with serine proteases.
To test this hypothesis, we prepared the Y226F mutant
of P. fluorescens kynureninase, and evaluated its physi-
cal and catalytic properties. The P. fluorescens enzyme
was chosen for these studies, rather than the human
enzyme, because of its much higher expression in
recombinant systems, ease of purification, and the fact
that the majority of previous kinetic and mechanistic
data have been obtained using it.
Fig. 4. 3-hydroxyhippurate bound in the active site of human
kynureninase. The hydrogen bonds are indicated by blue dashed
lines, and 3-hydroxyhippurate is shown as a space-filling model.
ꢀ
The hydrogen bond distances in A are indicated. The figure was
created using PDB structure 3E9K with Pymol.
Y226F P. fluorescens kynureninase has low, but
measurable, activity with L-kynurenine, with a kcat
value almost 2800-fold lower and a kcat/Km ratio
almost 375-fold lower than those of the wild-type
enzyme (Table 1). Thus, the effect of the Y226F muta-
tion on kcat is approximately 7.5-fold greater. We have
shown that the rate-determining step for wild-type
kynureninase is release of the second product, L-alanine
1
104
FEBS Journal 281 (2014) 1100–1109 ª 2013 FEBS

R. S. Phillips et al.
Acid/base catalysis in kynureninase
#
*
Xanthomonas
Cytophaga
Human
IVRLARAQGA AVGCDLAHAV GNIPLTLHDD GVDFAVWCNY KYLNAGPGAV
ISQATHLAGA FVGFDLAHAI GNIPLHLHDW KVDFAVWCSY KYLNGGPGAV
ITKAGQAKGC YVGFDLAHAV GNVELYLHDW GVDFACWCSY KYLNAGAGGI
ITRAGQAKGC FVGFDLAHAV GNVELHLHDW GVDFACWCSY KYLNSGAGGL
ITQAGHAKGC FVGFDLAHAV GNVELHLHDW DVDFACWCSY KYLNSGAGGL
ITKAGHAKGC FVGFDLAHAV GNVELRLHDW GVDFACWCSY KYLNSGAGGL
ITKAGQKKGC FVGFDLAHAV GNVELCLHDW GVDFACWCTY KYLNSGAGGL
ITTAGHSKGC FVGFDCAHAV GNAELRLHDW NVDFACWCTY KYMNSGAGGL
ITEAGHRKGC FVGFDLAHAF ANVPLHLHWW DVDFACWCSY KYGCTGAGSI
ITEVGHEIGA IVGWDLAHAA GNVELSLHDW NVDFACWCTY KYLNSGPGCI
ITAHAQSKGI LVGWDCAHAA GNVDLQLHDW NVDFAAWCTY KYLNSGPGGT
LTALSHECGA LAIWDLAHSA GAVPVDLHQA GADYAIGCTY KYLNGGPGSQ
Dog
Rat
Fig. 5. Sequence alignments of
Mouse
Chicken
kynureninases. The active site Lys227 is
indicated by an asterisk, the PLP-binding
Asp203 residue is indicated by a hash
symbol (#), and the Tyr226 residue is
shown with a black background.
Zebra fish
Caenorhabditis
Dictyostelium
Aspergillus
P. fluorescens
[3,6], but based on the mechanism in Scheme 1, the
Y226F mutant (Fig. 1B). The real pK of the enzyme
a
mutation is more likely to affect the rate of formation
of the first product, anthranilate, which occurs at 54
is 6.8, as that is the pK of the reaction with 3-hydrox-
a
ykynurenine, a slow substrate. It seems unlikely that
the Tyr226 phenol has a pK of 6.8, as it is hydrogen-
ꢀ
1
s
. If the rate-determining step for the Y226F mutant
a
enzyme is anthranilate formation, then the effect of
the mutation is an 8600-fold reduction. In contrast,
the Y226F mutation has a much smaller effect on both
PLP binding and alanine trans-amination. The effect
of the Y226F mutation on kcat for the O-benzoyl-L-ser-
ine reaction is only 25-fold, approximately 100-fold
less than the effect on kyurenine (Table 1). This is
consistent with a different rate-determining step for
the two substrates. However, the effect of the Y226F
mutation on kcat/Km for the elimination reaction of
O-benzoyl-L-serine, a 268-fold decrease, is similar to
that for L-kynurenine (Table 1). The mechanism for
reaction of O-benzoyl-L-serine is simpler, as the
enzyme needs only to form the external aldimine and
quinonoid intermediates, as shown in Scheme 2.
bonded to the dianionic phosphate, so we believe it is
more likely that this ionization corresponds to that of
the phosphate group of PLP. Furthermore, the wild-
type enzyme shows no changes in the UV-vis absorp-
tion spectrum with pH, so the basic group is not the
Schiff’s base of the PLP. Phosphate esters, including
PLP, are well known to exhibit pK values close to pH
a
3
1
6.8. The P-NMR of wild-type kynureninase shows a
peak of 5.0 ppm at pH 8.0, consistent with a dianionic
phosphate, as is seen with all PLP enzymes [15]. We
attempted to obtain the NMR spectrum of kynurenin-
ase at pH 6.0 to determine whether an upfield shift
may be seen that may be attributed to a phosphate
monoanion, but the enzyme precipitated. However, in
the presence of a potent competitive inhibitor, 5-bro-
modihydrokynurenine, at pH 8, we observed two new
peaks with an upfield shift in position, at 3.3 and
2.0 ppm (Fig. 3A). This 3 ppm upfield shift to
The ionization of a catalytic base with an apparent
pK of 6.5 in the pH dependence of kcat/Km for kynur-
enine with wild-type kynureninase is not seen in the
a
Scheme 2. Proposed mechanism for
reaction of O-benzoyl-L-serine.
FEBS Journal 281 (2014) 1100–1109 ª 2013 FEBS
1105

Acid/base catalysis in kynureninase
R. S. Phillips et al.
2
.0 ppm suggests that ligand binding results in a com-
bind. There must be an alternative base for deproto-
nation of the zwitterion, possibly the PLP 3′-O, in the
Y226F mutant enzyme. Furthermore, deprotonation
of the amino acid substrate and subsequent trans-aldi-
mination results in release of Lys227 as the free base,
which is required in subsequent steps to form the qui-
nonoid intermediate and the ketimine. The rate-limit-
ing step in the reaction of wild-type kynureninase is
deprotonation of C-4′ of the pyruvate ketimine [3].
plex with partial formation of the monoanionic phos-
phate ester even at pH 8. A change in the ionization
state of the PLP phosphate on ligand binding has not
been observed for other PLP-dependent enzymes [15].
3
1
The P-NMR spectrum of Y226F kynureninase shows
resonances at 3.7 and 2.5 ppm, suggesting that the
phosphate is in the monoanionic form, and there is no
further shift upon addition of 5-bromodihydrokynure-
nine (Fig. 3B). The loss of basic ionization in the pH
Hence, it is likely that the basic group with a pK of
a
3
1
dependence and the effect of mutation on the P-
NMR spectrum of the Y226F mutant enzyme suggests
that Tyr226 plays a role in the protonation state of the
PLP phosphate group. Furthermore, the short hydro-
gen bond between Tyr226 and the phosphate oxygen
demonstrates that Tyr226 cannot be a proton donor
independently of the phosphate.
6.8 seen in the pH dependence of kcat is that of
Lys227, rather than the PLP phosphate. This ioniza-
tion is not seen in the pH dependence of kcat of
Y226F kynureninase, consistent with a change in the
rate-limiting step. Lysine residues in enzyme active
sites are often found to have pK values reduced by
a
3–4 pH units, for example, in the case of acetoacetate
decarboxylase, with a pK of 5.9 [19]. At the kynure-
We propose that Tyr226, which is hydrogen-bonded
to the PLP phosphate in the free enzyme, accepts a
proton from the incoming substrate ammonium group
to initiate formation of the external aldimine
a
nine ketimine stage, Tyr226 donates a hydrogen bond
to the c-carbonyl oxygen of the substrate, resulting in
polarization of the carbonyl, and mediates transfer of
a proton to it from the phosphate when hydroxide
attacks the carbonyl, assisted by Lys227 as a general
base (Scheme 3). It is interesting that an off-pathway
vinylogous amide intermediate, with kmax = 348 nm,
accumulates during turnover, apparently arising from
accidental deprotonation of the rather acidic b-car-
bon, located between a carbonyl and an iminium ion,
either by Lys227 or by the nascent hydroxide, in com-
petition with carbonyl hydration [5,6]. This intermedi-
(Scheme 3), and concomitantly releases a proton to
the phosphate, resulting in the monoprotonated phos-
phate hydrogen-bonded to the phenol of Tyr226. This
is also supported by the pH dependence of kcat/Km
for wild-type kynureninase, which shows an acidic
group with a pKa of 8.8, which we assigned to the
substrate zwitterion [3]. The lack of such ionization in
the Y226F mutant enzyme suggests that both the
zwitterionic and anionic forms of kynurenine may
Scheme 3. Proposed role of Y226 in the mechanism of kynureninase.
1
106
FEBS Journal 281 (2014) 1100–1109 ª 2013 FEBS

R. S. Phillips et al.
Acid/base catalysis in kynureninase
ꢀ
ate is in rapid equilibrium with the ketimine, so it
does not impair catalysis. The 30-fold weaker binding
of the potent transition state analog inhibitor, S-
P. fluorescens) with the c-O located 4.5 A from the 3-
hydroxyhippurate carbonyl oxygen. Although the
geometry of Ser35 is not ideal for hydrogen bonding
in the 3-hydroxyhippurate complex, the distance may
be reduced after trans-aldimination and the geometry
may improve. Once the anthranilate product is
released, Tyr226 appears to have little or no role in
the subsequent enamine to ketimine conversion steps
(Scheme 1). The principle of microscopic reversibility
suggests that the proton on the phosphate is ulti-
mately donated via Tyr226 back to the amino group
of the L-alanine product as it leaves the active site,
thus re-setting the enzyme in the proper ionization
state for another catalytic cycle. This explains the low
reactivity of Y226F in the abortive trans-amination
with alanine.
(2-aminophenyl)-L-cysteine S,S-dioxide, to Y226F
kynureninase (Table 1) may be due to loss of a hydro-
gen bond from the tyrosine to the sulfone oxygen,
which structurally resembles the gem-diol intermediate.
Furthermore, this mechanism is consistent with the ste-
reochemical properties of the dihydrokynurenines,
whereby the (aS,4R) diastereomer is a slow substrate
for a retro-aldol reaction, and the (aS,4S) isomer is a
potent competitive inhibitor [2,13]. Based on this
mechanism, the low kcat of the (aS,4S) isomer of
dihydrokynurenine may be due to the active form of
the enzyme being the low-concentration dianionic pro-
tonation state of the phosphate, as proton abstraction
from the alcohol is required to initiate the retro-aldol
reaction, rather than proton donation (Scheme 4). In
the reaction of kynurenine, the proton is donated to
the carbonyl oxygen from the phosphate as the
gem-diol is formed, through mediation of Tyr226, and
then returned to the phosphate, initiating the retro-
There are very few examples of PLP-dependent
enzymes that catalyze a reaction with amino acid sub-
strates using the phosphate group of the cofactor as
an auxiliary acid/base catalyst. The phosphate group
of PLP is generally assumed to be only a ‘handle’ for
the enzyme to hold the cofactor during catalysis, when
the covalent attachment with the lysine is lost. Inter-
estingly, for threonine synthase, the phosphate product
from elimination of c-phosphohomoserine was shown
to be catalytically active in water addition to the am-
inocrotonate intermediate [20]. Glycogen phosphory-
lase contains PLP, and the phosphate group has been
shown to be the catalytically essential part of the PLP
molecule [21], and the pyridine ring provides the non-
catalytic ‘handle’ for the enzyme. It is interesting that
the phosphate group must be dianionic in glycogen
phosphorylase, and recent evidence suggests that it
does not function via simple acid/base catalysis [21]. It
has been proposed on the basis of X-ray crystal struc-
tures that L-serine dehydratase [22] and D-serine dehy-
dratase [23] use the phosphate of the PLP both as a
base to facilitate external aldimine formation and as
an acid to assist in elimination of water from the
substrate. However, in contrast to this hypothesis,
Claisen reaction resulting in C –C
cleavage
b
c
(Scheme 3).
If the mechanism in Scheme 3 is correct, how does
the Y226F mutant enzyme have any remaining activ-
ity? There is another conserved tyrosine residue,
Tyr228 in P. fluorescens and Tyr277 in human ky-
nureninase, located just after the active-site lysine
(Fig. 5). However, in the structure of human kynure-
ninase with 3-hydroxyhippurate bound, the phenol of
ꢀ
Tyr277 is 11.3 A from the ligand carbonyl oxygen
ꢀ
and 15.0 A from the nearest phosphate oxygen, so
Tyr228 is unlikely to be able to substitute in the
Y226F mutant enzyme. In the Y226F mutant, it is
possible that a water molecule may form a bridge
between the phosphate and the substrate c-carbonyl
oxygen, replacing the Tyr OH and allowing reaction
to proceed more slowly. Alternatively, the gem-diol
intermediate may be partly stabilized by other resi-
dues. In this regard, it is interesting that there is
another strictly conserved residue, Ser75 (Ser35 in
3
1
P-NMR studies of D-serine dehydratase bound to a
substrate analog (isoserine) do not show formation of
Scheme 4. Proposed mechanism for reaction of dihydrokynurenine.
FEBS Journal 281 (2014) 1100–1109 ª 2013 FEBS
1107

Acid/base catalysis in kynureninase
R. S. Phillips et al.
the phosphate monoanion [15], but instead show
stabilization of the dianion, even at acidic pH values.
Thus, the evidence supports the hypothesis that the
reaction mechanism of kynureninase requires catalysis
using both the electronic properties of the pyridine
ring and the acid/base properties of the phosphate
moiety of PLP.
parameters were obtained by fitting the data to Eqns (2,3),
using HYPERO for kcat and kcat/K
[27]:
m
and COMPO for
K
i
V ¼ VmaxS=ðKm þ SÞ
(2)
V ¼ VmaxS=ðKmð1 þ I=KiÞ þ SÞ
(3)
The kcat and kcat/K parameters were obtained from Vmax
m
and Vmax/K using the enzyme molecular weight of 47 kDa
21]. The pH dependence data were obtained in 0.1 M
m
[
Experimental procedures
triethanolamine phosphate buffers adjusted to pH values
ranging from 6.5 to 9.5. The actual pH of the reaction mix-
tures was measured at the end of data collection.
Materials
5
4
-bromodihydrokynurenine was prepared by NaBH reduc-
3
1
tion of 5-bromo-L-kynurenine as previously described [13],
and used as the obtained 60:40 mixture of diastereomers.
O-benzoyl-L-serine [11] and S-(2-aminophenyl)-L-cysteine S,
S-dioxide [14] were prepared as described previously. Buf-
fers and reagents were purchased from Fisher Scientific Co
P-nmr
Wild-type P. fluorescens kynureninase was passed
through a PD-10 gel (GE Healthcare Life Sciences,
Pittsburgh, PA, USA) filtration column equilbrated
with 0.05 M bicine-Na, pH 8, to remove phosphate
buffer, and concentrated using a Centricon centrifugal
filter over a YM-30 membrane (Millipore, Billerica,
(Waltham, MA, USA).
Enzymes
MA, USA) to a final volume of 0.50 mL. D O
2
Y226F mutant kynureninase was prepared using a modifica-
tion of the QuikChange protocol (Stratagene, Santa Clara,
CA, USA), using plasmid pTZKyn [22] as the template with
mutagenic primers 5′-CTACGCCATCGGTTGCACTTTC
AAAT-3′ (forward) and 5′-GACCGCCATTGAGGTATTT
GAAAGTG-3 (reverse) (mutated positions are underlined).
The resulting Y226F mutant plasmid was transformed into
Escherichia coli TOP-10 cells (Invitrogen, Carlsbad, CA,
USA), colonies were selected, and plasmids were isolated,
sequenced to confirm the mutation, and transformed into
E. coli DH5a cells for expression. Cell were grown, and wild-
type and Y226F P. fluorescens kynureninases were purified
by phenyl-Sepharose chromatography as described previ-
ously [24]. The concentration of enzyme was estimated from
(50 lL) was then added for the field frequency lock.
The final concentration of the enzyme was 1 mM
(47 mgꢂmLꢀ ). P-NMR data were collected at ambi-
ent temperature (approximately 20 °C) on a Varian
Mercury 400 (Aligent Technologies, Santa Clara, CA,
USA) instrument operating at 161.984 Mhz, with pro-
ton decoupling. A total of 20 000 transients were col-
lected in 64 k data points with a spectral width of
1
31
4
5
2
0 486 Hz. The acquisition time was 0.809 s, with a
s relaxation delay. The data were multiplied with
0 Hz exponential line broadening before Fourier
transformation. For the inhibitor complex, the
enzyme was first exchanged into 0.05 M bicine-Na,
pH 8.0, then diluted with 6 mL of 2 mM 5-bro-
modihydrokynurenine in 0.05 M bicine-Na, pH 8.0,
and concentrated to 1 mM in a Centricon centrifugal
1
%
the absorbance at 280 nm, using a value for A of 14.3 [25].
2
80
The PLP content was determined by the absorption of the
protein at 388 nm in 0.1 M NaOH [26].
filter over a YM-30 membrane, and 10% v/v D O
2
was added. In this experiment, a total of 30 000 tran-
sients were collected, and the data were multiplied
with a 20 Hz exponential before Fourier transforma-
tion. The Y226F mutant kynureninase was prepared
similarly to wild-type kynureninase. A total of 38 000
transients were collected in 64 k data points with a
spectral width of 40 486 Hz, and the data were multi-
plied with 20 Hz exponential line broadening before
Fourier transformation. For the Y226F–inhibitor
complex, a total of 38 000 transients were collected,
and the data were multiplied with a 20 Hz exponen-
tial before Fourier transformation. Chemical shifts
are reported relative to an external standard of 85%
H PO .
Kinetic studies
Kinetics measurements were performed on a Cary 1 UV-vis
spectrophotometer (Agilent Technologies, Santa Clara, CA,
USA) equipped with a 6 9 6 Peltier temperature-controlled
cell compartment. The kinetic studies were performed
at 37 °C in 40 mM potassium phosphate, pH 8.0, 50 lM
PLP, with varying amounts of substrate, by following the
absorbance decrease at 360 nm (De = ꢀ4500 Mꢀ cm ) as
kynurenine is converted to anthranilic acid. For the reaction
of O-benzoyl-L-serine, formation of pyruvate was followed
at 340 nm by addition of 20 lgꢂmLꢀ rabbit muscle lactate
dehydrogenase (US Biochemicals, Cleveland, OH, USA)
1
ꢀ1
1
and 0.1 mM NADH (De =ꢀ6220 Mꢀ cm ). The kinetic
1
ꢀ1
3
4
1
108
FEBS Journal 281 (2014) 1100–1109 ª 2013 FEBS

R. S. Phillips et al.
Acid/base catalysis in kynureninase
1
1
3 Heiss C, Anderson J & Phillips RS (2003) Differential
effects of bromination on substrates and inhibitors of
kynureninase from Pseudomonas fluorescens. Org
Biomol Chem 1, 288–295.
Acknowledgements
We would like to thank Greg Wiley (Department of
Chemistry, University of Georgia) for his assistance in
3
collection of the P-NMR spectra.
1
4 Dua RK, Taylor EW & Phillips RS (1993) S-Aryl-L-
cysteine S,S-dioxides: design and evaluation of a new
class of mechanism based inhibitors of kynureninase.
J Am Chem Soc 115, 1264–1270.
References
31
1
2
3
Soda K & Tanizawa K (1979) Kynureninases:
enzymological properties and regulation mechanism.
Adv Enzymol Relat Areas Mol Biol 49, 1–40.
15 Schnackerz KD, Andi B & Cook PF (2011) P NMR
spectroscopy senses the microenvironment of the 5′-
phosphate group of enzyme-bound pyridoxal 5′-
Phillips RS & Dua RK (1991) Stereochemistry and
mechanism of aldol reactions catalyzed by
phosphate. Biochim Biophys Acta 1814, 1447–1458.
16 Schnackerz KD, Keller J, Phillips RS & Toney MD
(2006) Ionization state of pyridoxal 5′-phosphate in D-
serine dehydratase, dialkylglycine decarboxylase and
tyrosine phenol-lyase and the influence of monovalent
kynureninase. J Am Chem Soc 113, 7385–7388.
Koushik SV, Moore JA III, Sundararaju B & Phillips
RS (1998) The catalytic mechanism of kynureninase
from Pseudomonas fluorescens: insights from the effects
of pH and isotopic substitution on steady-state and
pre-steady-state kinetics. Biochemistry 37, 1376–1382.
Phillips RS, Sundararaju B & Koushik SV (1998) The
catalytic mechanism of kynureninase from
31
cations as inferred by P NMR spectroscopy. Biochim
Biophys Acta 1764, 230–238.
17 Schnackerz KD, Wahler G, Vincent MG & Jansonius JN
31
4
5
(1989) Evidence that P NMR is a sensitive indicator of
small conformational changes in the coenzyme of
aspartate aminotransferase. Eur J Biochem 185, 525–531.
18 Robertus JD, Kraut J, Alden RA & Birktoft JJ (1972)
Subtilisin. Stereochemical mechanism involving
transition-state stabilization. Biochemistry 11, 4293–4303.
19 Westheimer FH & Schmidt DE Jr (1971) pK of the
lysine amino group at the active site of acetoacetate
decarboxylase. Biochemistry 10, 1249–1253.
Pseudomonas fluorescens: evidence for transient
quinonoid and ketimine intermediates from rapid-
scanning stopped-flow spectrophotometry. Biochemistry
3
7, 8783–8789.
Gawandi VB, Liskey D, Lima S & Phillips RS (2004)
Reaction of Pseudomonas fluorescens kynureninase with
b-benzoyl-L-alanine: detection of a new reaction
intermediate and a change in rate-determining step.
Biochemistry 43, 3230–3237.
20 Murakawa T, Machida Y & Hayashi H (2011) Product-
assisted catalysis as the basis of the reaction specificity of
threonine synthase. J Biol Chem 286, 2774–2784.
21 Stirtan WG & Withers SG (1996) Phosphonate and
a-fluorophosphonate analogue probes of the ionization
state of pyridoxal 5′-phosphate (PLP) in glycogen
phosphorylase. Biochemistry 35, 15057–15064.
22 Yamada T, Komoto J, Takata Y, Ogawa H, Pitot HC &
Takusagawa F (2003) Crystal structure of serine
dehydratase from rat liver. Biochemistry 42, 12854–12865.
23 Urusova DV, Isupov MN, Antonyuk S, Kachalova GS,
Obmolova G, Vagin AA, Lebedev AA, Burenkov GP,
Dauter Z, Bartunik HD et al. (2012) Crystal structure
of D-serine dehydratase from Escherichia coli. Biochim
Biophys Acta 1824, 422–432.
6
7
Kumar S, Gawandi VB, Capito N & Phillips RS (2010)
Substituent effects on the reaction of b-benzoylalanines
with Pseudomonas fluorescens kynureninase.
Biochemistry 49, 7913–7919.
Lima S, Kumar S, Gawandi V, Momany C & Phillips
RS (2009) Crystal structure of the Homo sapiens
kynureninase-3-hydroxyhippuric acid inhibitor complex:
insights into the molecular basis of kynureninase
substrate specificity. J Med Chem 52, 389–396.
Lima S, Khristoforov R, Momany C & Phillips RS
8
9
(2007) Crystal structure of Homo sapiens kynureninase.
Biochemistry 46, 2735–2744.
Momany C, Levdikov V, Blagova L, Lima S & Phillips
RS (2004) Three-dimensional structure of
kynureninase from Pseudomonas fluorescens.
Biochemistry 43, 1193–1203.
24 Koushik SV, Sundararaju B, McGraw RA & Phillips
RS (1997) Cloning, sequence, and expression of
kynureninase from Pseudomonas fluorescens.
1
1
1
0 Moriguchi M & Soda K (1973) Transamination
reaction catalyzed by kynureninase and control of the
enzyme activity. Biochemistry 12, 2974–2980.
Arch Biochem Biophys 344, 301–308.
25 Moriguchi M, Yamamoto T & Soda K (1973)
Properties of crystalline kynureninase from
1 Phillips RS (1987) Reactions of O-acyl-L-serines with
tryptophanase, tyrosine phenol-lyase, and tryptophan
synthase. Arch Biochem Biophys 256, 302–310.
2 Kishore GM (1984) Mechanism-based inactivation of
bacterial kynureninase by b-substituted amino acids.
J Biol Chem 259, 10669–10674.
Pseudomonas marginalis. Biochemistry 12, 2969–2974.
26 Peterson EA & Sober HA (1954) Preparation of
crystalline phosphorylated derivatives of vitamin B6.
J Am Chem Soc 76, 169–175.
27 Cleland WW (1979) Statistical analysis of enzyme
kinetic data. Methods Enzymol 63, 103–138.
FEBS Journal 281 (2014) 1100–1109 ª 2013 FEBS
1109