Spectroscopic studies of ligand and substrate binding to human indoleamine 2,3-dioxygenase

Article abstract of DOI:10.1021/bi1005078

Human indoleamine 2,3-dioxygenase (hIDO) is an intracellular heme-containing enzyme, which catalyzes the initial and rate-determining step of l-tryptophan (l-Trp) metabolism via the kynurenine pathway in nonhepatic tissues. Steady-state kinetic data showed that hIDO exhibits substrate inhibition behavior, implying the existence of a second substrate binding site in the enzyme, although so far there is no direct evidence supporting it. The kinetic data also revealed that the K_m of l-Trp (15 μM) is ～27-fold lower than the K_d of l-Trp (0.4 mM) for the ligand-free ferrous enzyme, suggesting that O₂ binding proceeds l-Trp binding during the catalytic cycle. With cyanide as a structural probe, we have investigated the thermodynamic and kinetic parameters associated with ligand and substrate binding to hIDO. Equilibrium titration studies show that the cyanide adduct is capable of binding two l-Trp molecules, with K_d values of 18 μM and 26 mM. The data offer the first direct evidence of the second substrate binding site in hIDO. Kinetic studies demonstrate that prebinding of l-Trp to the enzyme retards cyanide binding by ～13-fold, while prebinding of cyanide to the enzyme facilitates l-Trp binding by ～22-fold. The data support the view that during the active turnover of the enzyme it is kinetically more favored to bind O₂ prior to l-Trp.

Full text of DOI:10.1021/bi1005078

5
028 Biochemistry 2010, 49, 5028–5034
DOI: 10.1021/bi1005078
Spectroscopic Studies of Ligand and Substrate Binding to Human
Indoleamine 2,3-Dioxygenase
Changyuan Lu, Yu Lin, and Syun-Ru Yeh*
Department of Physiology and Biophysics, Albert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, New York 10461
Received April 3, 2010; Revised Manuscript Received May 13, 2010
ABSTRACT: Human indoleamine 2,3-dioxygenase (hIDO) is an intracellular heme-containing enzyme, which
catalyzes the initial and rate-determining step of
L
-tryptophan ( -Trp) metabolism via the kynurenine
L
pathway in nonhepatic tissues. Steady-state kinetic data showed that hIDO exhibits substrate inhibition
behavior, implying the existence of a second substrate binding site in the enzyme, although so far there is no
direct evidence supporting it. The kinetic data also revealed that the K of
L
-Trp (15 μM) is ∼27-fold lower
m
than the K of
L-Trp (0.4 mM) for the ligand-free ferrous enzyme, suggesting that O binding proceeds -Trp
L
d
2
binding during the catalytic cycle. With cyanide as a structural probe, we have investigated the thermo-
dynamic and kinetic parameters associated with ligand and substrate binding to hIDO. Equilibrium titration
studies show that the cyanide adduct is capable of binding two -Trp molecules, with K values of 18 μM and
L
d
26 mM. The data offer the first direct evidence of the second substrate binding site in hIDO. Kinetic studies
demonstrate that prebinding of
cyanide to the enzyme facilitates
turnover of the enzyme it is kinetically more favored to bind O prior to
L
-Trp to the enzyme retards cyanide binding by ∼13-fold, while prebinding of
L
-Trp binding by ∼22-fold. The data support the view that during the active
L
-Trp.
2
1
L
-Tryptophan (
L
-Trp) is the scarcest essential amino acid in
simulations reveal that the dioxygenase reactions carried out by
IDO and TDO follow a common mechanism, by which the two
atoms of the heme-bound dioxygen are inserted into the substrate
one at a time via a ferryl intermediate (11). Despite the structural
and functional similarities, IDO exhibits several unique proper-
ties distinguishing itself from TDO. (i) IDO forms relatively
stable dioxygen adducts regardless of Trp (12), whereas the
dioxygen complex of TDO is stable only in the presence of
Trp (13). (ii) The ternary complex of IDO can be readily oxidized
mammals. Most of our dietary Trp is metabolized in the liver
along the kynurenine pathway, which ultimately leads to the
biosynthesis of NAD (1-4). The conversion of L-Trp to N-
formylkynurenine (NFK), the initial and rate-determining step of
the kynurenine pathway, is catalyzed by a hepatic heme-based
dioxygenase, tryptophan 2,3-dioxygenase (TDO), via insertion of
both atoms of dioxygen into the C dC bond of the indole
2
3
moiety of L-Trp (Scheme 1). In addition to TDO, a second heme-
based dioxygenase, indoleamine 2,3-dioxygenase (IDO), discov-
by releasing O as superoxide, while that of TDO is much less
2
ered by Hayaishi et al. in 1967 (5), catalyzes the same oxidative
susceptible to autoxidation (14). (iii) The binding of L-Trp in IDO
ring cleavage reaction of
L
-Trp to NFK. In contrast to the hepatic
retards CO binding, while that in TDO accelerates it (12, 13). (iv)
The ferric derivative of IDO reacts with superoxide to form the
active oxy complex, while that of TDO does not bind super-
oxide (14, 15).
As illustrated in Scheme 2, it is generally believed that during
the active catalytic cycle of hIDO (indicated by the black arrows),
TDO, IDO is ubiquitously distributed in all tissues other than
liver. In addition, instead of regulating homeostatic serum Trp
concentrations like TDO, IDO can be induced by IFN-γ and
closely linked to immunomodulation by regulating T cell activity
via controlling
L-Trp catabolism (6, 7). Recently, IDO has
attracted a great deal of attention because of its potential as a
therapeutic target for cancer and neurological disorders (8, 9).
IDO is present only in mammals. The overall structure of the
human isoform (hIDO), resolved by Sugimoto et al. in 2006 (10),
displays two R-helical domains sandwiching the prosthetic heme
group. The structure of the large domain of IDO is highly similar
with that of TDO, suggesting that the two dioxygenases carry out
the dioxygen chemistry with a similar mechanism. Consistently,
recent resonance Raman studies combined with MD and QM/MM
the ferrous enzyme first binds L-Trp, followed by O binding, to
2
generate the ternary complex, which turns over to produce the
product, NFK, leaving the enzyme in the catalytically competent
ferrous state. During the reaction, a small amount of enzyme may
leak out of the active cycle via autoxidation to the inactive ferric
state, which can return to the active cycle only by accepting an
electron or by binding a superoxide (see the gray arrows). The
sequential binding of Trp and O was first proposed on the basis
2
of the observation that the apparent Michaelis-Menten constant
(
K ) of
m
L-Trp (13 μM) for the rabbit isoform of the enzyme
*To whom correspondence should be addressed: Department of
(
rIDO) matches the equilibrium dissociation constant, Kd (
L-Trp),
Physiology and Biophysics, Albert Einstein College of Medicine,
Bronx, NY 10461. Telephone: (718) 430-4234. Fax: (718) 430-4230.
of the ligand-free ferrous enzyme (14, 16). However, recent data
show that the Km of the human isoform (∼15 μM) is ∼40-fold
E-mail: syun-ru.yeh@einstein.yu.edu.
1
Abbreviations: IDO, indoleamine 2,3-dioxygenase; TDO, tryptophan
lower than the Kd
(L-Trp) of the ligand-free enzyme (0.53
2
,3-dioxygenase; hIDO, human indoleamine 2,3-dioxygenase; rIDO,
rabbit indoleamine 2,3-dioxygenase; IFN-γ, interferon-γ; NFK, N-for-
mylkynurenine; -Trp, -tryptophan.
mM) (10, 17), implying that in hIDO, O2 binding proceeds
L
L
L-Trp binding as illustrated by the red arrows in Scheme 2.
pubs.acs.org/Biochemistry
Published on Web 05/17/2010
r 2010 American Chemical Society

Article
Biochemistry, Vol. 49, No. 24, 2010 5029
Scheme 1: Dioxygenation Reaction of Tryptophan Catalyzed
by IDO and TDO
Scheme 2: Catalytic Cycle of Human Indoleamine 2,3-Dioxy-
genase
F
IGURE 1: Absorption spectra of ferrous hIDO (a) and equilibrium
titration curve of ferrous hIDO with L-Trp (b). The absorption
spectra in panel a were recorded in the absence (black curve) or
presence (red curve) of 20 mM L-Trp. The ΔA values in panel b were
measured at 428 nm. All the data were obtained with 5.5 μM hIDO in
1
00 mM Tris buffer (pH 7.4) at room temperature.
of analytical reagent grade without further purification. All
solutions were prepared with deionized (Millipore) water. Col-
umns and media for protein purification were from Amersham
Pharmacia Biotech.
Protein Expression and Purification. The recombinant
hIDO protein was prepared as described previously (20). The
protein concentration was estimated by using the Soret absorp-
Although the structural origin of the differences observed in
hIDO with respect to rIDO remains unknown, the data suggest
that the kinetic and thermodynamic properties associated with O2
and Trp binding are isoform-specific and evolutionarily optimized.
It has been recognized several decades ago that the steady-
state activity of rIDO follows Michaelis-Menten behavior at
-
1
-1
tion at 404 nm with an extinction coefficient of 172 mM cm
for the ferric form (18). The catalytic activity of hIDO was
confirmed by monitoring the increase in optical absorption at 321
nm, indicative of product (N-formylkynurenine) formation, with
-
1
-1
an extinction coefficient of 3.75 mM cm (13, 17). The protein
was stored at -80 °C until it was used. We prepared the ferrous
ligand-free sample by first flushing the sample with Ar gas and
then reducing it with a 5-fold molar excess of sodium dithionite.
Unless otherwise indicated, the hIDO samples were buffered with
<
0.2 mM
leads to a decrease in the activity (5, 16). It was initially believed
that, at high concentrations of -Trp, the substrate binds to the
L-Trp, yet a further increase in the L-Trp concentration
L
ferric enzyme, thereby inhibiting the turnover of the enzyme by
retarding its reduction to the active ferrous state (16). However,
recent studies suggest that this scenario is not operative in hIDO
1
00 mM Tris-HCl and 50 μM EDTA (pH 7.4).
Optical Absorption Measurements. The optical absorption
on the basis of two new observations. (i)
ferric enzyme promotes, instead of retards, its reduction (from a
thermodynamic point of view) (18), and (ii) the Kd ( -Trp) of the
L-Trp binding to the
spectra were recorded by using a Shimadzu spectrophotometer
Shimadzu Co., Kyoto, Japan). All spectra were recorded at
room temperature (∼25 °C). The protein concentrations were
(
L
ferric enzyme (0.9 mM) is much higher than the self-inhibition
∼
1-10 μM. The equilibrium ligand and substrate binding data
constant (K , 0.17 mM) (17). Additional data for hIDO indicate
si
were fitted with eq 1 by using Origin version 6.1 (Microcal, Inc.,
that, at high concentrations, the substrate can bind to an inhi-
bitory substrate binding site in addition to the active site, thereby
accounting for the substrate inhibition behavior (17). Although
recent ligand rebinding studies of CO-bound hIDO at cryogenic
temperatures provide kinetic evidence supporting the two sub-
strate binding sites in hIDO (19), so far there are no direct
experimental data demonstrating the existence of the inhibitory
substrate binding site in hIDO.
Northampton, MA) (21).
2
ΔA ¼ ΔAmaxf½hIDOꢀ þ ½Lꢀ þ K - ½ð½hIDOꢀþ½LꢀþK Þ
d
d
1=2
- 4½hIDOꢀ½Lꢀꢀ g=ð2½hIDOꢀÞ
ð1Þ
where [L] and [hIDO] are the ligand (or substrate) and enzyme
concentrations, respectively, Kd is the ligand (or substrate)
dissociation constant, and ΔA and ΔAmax are the observed and
maximum changes in absorbance, respectively.
Stopped-Flow Measurements. The transient kinetics ex-
periments were performed with a π*180 stopped-flow instrument
from Applied Photophysics Inc. (Leatherhead, U.K.) (22). The
progression of the reaction was monitored in optical absorption
mode with a photodiode array detector at 20 °C. The data were
analyzed using Pro-K software provided by Applied Photophys-
Recent resonance Raman data of the O adduct of hIDO show
2
2þ
3þ
-
that, instead of Fe -O2 character, it exhibits Fe -O2
character (11), suggesting that the O adduct can be mimicked
2
3þ
-
by the Fe -CN complex. In this work, we sought to use
cyanide as a structural probe to investigate the thermodynamic
and kinetic properties associated with substrate and ligand
binding to hIDO. Our data provide the first direct evidence
supporting the second substrate binding site in hIDO; they also
ics. The concentration of L-Trp or cyanide was at least 5-fold
higher than that of the protein, so that all the reactions were
conducted under pseudo-first-order conditions.
revealed that prebinding of
subsequent binding of cyanide, while prebinding of cyanide
facilitates the subsequent binding of -Trp, both observations
L-Trp to the enzyme retards the
L
being consistent with the view that Trp binding is substantially
promoted by binding of the small dioxygen ligand.
RESULTS AND DISCUSSION
K (L-Trp) of Ferrous and Ferric hIDO. The optical
d
MATERIALS AND METHODS
absorption spectrum of the substrate-free ferrous hIDO exhibits
a Soret maximum at 429 nm (Figure 1a) and a Q-band at 558 nm
(as listed in Table 1), characteristics of five-coordinate high-spin
Materials.
L-Tryptophan (L-Trp) and potassium cyanide
(
KCN) were purchased from Aldrich. All chemicals used were

5
030 Biochemistry, Vol. 49, No. 24, 2010
Lu et al.
curve fitting of the data with a single-binding site model described
in eq 1. The slight decrease in Kd (L-Trp) from 0.9 to 0.4 mM upon
Table 1: Soret Bands and Q-Bands of the Various Derivatives of hIDO
λ
Soret (nm)
λvis (nm)
reduction is consistent with the observation that substrate
binding in hIDO leads to a small increase in its redox potential
3
3
3
þ
þ
þ
Fe
Fe
404
410
419
416
429
425
499, 533, 633
540, 576
541, 570
541, 570
558
(
∼46 mV) (18). The substrate-induced redox potential change in
-L-Trp
-
hIDO is substantially weaker than those observed in P450 and
NOS [157 mV (14, 26) and 84 mV (27) for P450cam and iNOS,
respectively]. It is generally believed that the substrate binding-
induced increase in the redox potential of P450 and NOS plays a
critical role in preventing the formation and release of super-
oxide, which is toxic to cells, under substrate-deficient conditions
Fe -CN
3þ
2þ
2
þ
-Trp-CN-
Fe
Fe
Fe
-
L
-
L-Trp
555
in vivo. For hIDO, the autoxidation rate of the O complex of
2
hIDO is relatively slow in the absence of substrate; this type of
cellular protection mechanism, hence, might not be vital. How-
ever, it is noteworthy that, in contrast to P450 and NOS, IDO
exhibits a much higher autoxidation rate in the substrate-bound
state (14) (for reasons yet to be resolved); consequently, an exoge-
nous electron source is required to sustain the multiple turnovers
of the enzyme in vitro, despite the fact that the dioxygenase
reaction does not consume any electron (Scheme 1).
F
IGURE 2: Absorption spectra of ferric hIDO (a) and equilibrium
titration curve of ferric hIDO with L-Trp (b). The absorption spectra
in panel a were recorded in the absence (black curve) or presence (red
curve) of 20 mM L-Trp. The ΔA values in panel b were measured at
404 nm. Allthe datawereobtainedwith5.5μM hIDOin100 mMTris
buffer (pH 7.4) at room temperature.
Kd (L-Trp) of Cyanide-Bound Ferric hIDO. Cyanide
binding to the substrate-free ferric enzyme causes the shift of
the Soret maximum to 419 nm and Q-bands to 541 and 570 nm
(
Figure 3a). Titration studies show a high affinity of cyanide for
hIDO, with a K of 2.4 ( 0.2 μM on the basis of the data fitting
d
ferrous heme species. The binding of
L-Trp causes the shift of the
with eq 1 (Figure 3b).
Soret and Q-bands to 425 and 555 nm, respectively.
L-Trp titra-
The addition of 1 mM L-Trp to the cyanide-bound enzyme
tion studies show a typical hyperbolic behavior (Figure 1b), with
leads to the shift of the Soret maximum to 416 nm, as well as a
small change in the Q-bands (Figure 3a). The spectral change is
consistent with the published resonance Raman data showing
that the cyanide-related vibrational modes (including νFe-CN and
a K of 0.40 ( 0.07 mM based on curve fitting of the data with a
d
single-binding site model described in eq 1, which is consistent
with the reported value (0.53 mM) (10). The data confirm that the
K of
L
-Trp for the active ferrous enzyme is much higher than the
-Trp (15 μM) observed during steady-state reactions (17),
-Trp binding
does not proceed O binding. It is noteworthy that the spectral
δ
Fe-C-N modes) are significantly perturbed by
implying the proximity of -Trp to the heme-bound cyanide (20).
A further increase in the -Trp concentration to 44 mM surpris-
L-Trp binding,
d
K of
L
L
m
supporting the view that during the catalytic cycle
L
L
ingly induces partial recovery of the intensity of the Soret band
without affecting its maximum position. To understand the
spectral transitions, the cyanide-bound enzyme was titrated with
2
feature of the
distinct from that reported by Chauhan et al. (with λmax at
13, 551, and 575 nm) (23), as in the latter case the spectrum is
L-Trp-bound ferrous hIDO shown in Figure 1a is
4
L-Trp. As shown in Figure 3c, two transitions with K
values of
d
likely derived from a mixture of deoxy and O -bound enzyme (the
latter exhibits λmax at 412, 542, and 576 nm) (11).
The optical absorption spectrum of the substrate-free ferric
hIDO, on the other hand, exhibits a Soret maximum at 404 nm
and Q-bands at 499 and 533 nm (Figure 2a), as well as a charge
transfer band at 633 nm, characteristics of a water-bound ferric
18 ( 3 μM and 26 ( 8 mM are revealed.
The clear two-step transition shown in Figure 3c provides the
first experimental evidence demonstrating the existence of two
2
substrate binding sites in hIDO. The K value for the second
d
substrate binding site, 26 mM, is much higher than the substrate
inhibition constant (K ), 0.17 mM, observed during the steady-
si
species. The binding of
L-Trp causes the shift of the Soret maximu
state turnover of the enzyme, indicating that the affinity of L-Trp
to 410 nm and Q-bands to 540 and 576 nm, which is accompanied
by the disappearance of the charge transfer band, characteristic
of a hydroxide-bound ferric species, as revealed by resonance
Raman studies (20), suggesting substrate-induced deprotonation
of the heme-bound water.
In the P450 or NOS class of enzymes (with thiolate as the
proximal heme ligand), substrate binding typically results in the
displacement of the heme-bound water (24, 25), due to steric
hindrance introduced by substrate occupying the distal heme
pocket. The substrate binding-induced deprotonation of the
heme-bound water in hIDO indicates that the indoleamine or
for the second binding site is sensitive to the redox and/or ligation
state of the heme iron. Likewise, the K value of the first binding
d
site, 18 μM, is much lower than that of the CO-bound form
[0.12 mM (C. Lu et al., results to be published)], as well as those of
the ligand-free ferric enzyme (0.9 mM) and ferrous enzyme
(0.4 mM). Taken together, the data indicate that the redox-state
change and/or ligand binding to the heme iron introduces signi-
ficant structural perturbations into the protein matrix housing
the two substrate binding sites and that the second substrate
binding site cannot be observed in the ligand-free ferric and
ferrous enzyme (see Figures 1b and 2b).
amino group of the
L-Trp is located in the proximity of the water,
L-Trp Binding Kinetics. To investigate the structural per-
thereby perturbing its pK . The enzyme-specific substrate-ligand
turbation introduced by cyanide binding to the heme iron, L-Trp
a
interactions are plausibly optimized for each of these oxygen-
binding to the ligand-free and cyanide-bound hIDO was studied
with stopped-flow measurements. As shown in Figure 4a, the
kinetic trace at 403 nm obtained from the reaction of the ligand-
utilizing enzymes to selectively enhance specific functions.
L-Trp titration of the ferric species shows typical hyperbolic
behavior with a K value of 0.90 ( 0.09 mM (Figure 2b), based on
free enzyme with 0.75 mM L-Trp follows single-exponential
d

Article
Biochemistry, Vol. 49, No. 24, 2010 5031
F
(
1
IGURE 3: Absorption spectra of cyanide-bound ferric hIDO (a) and equilibrium titration curve of the substrate-free ferric hIDO with cyanide
b) or that of the cyanide-bound hIDO with L-Trp (c). The absorption spectra in panel a were recorded in the absence (black curve) or presence of
mM L-Trp (red curve) or 44 mM L-Trp (blue curve). The ΔA values in panels b and c were measured at 403 and 423 nm, respectively. The data
were obtained with 4.7 μM hIDO (a and c) and 1.5 μM hIDO (b) in 100 mM Tris buffer (pH 7.4) at room temperature. The cyanide concentration
used for panels a and c was 1 mM.
F
IGURE 4: L-Trp binding kinetics of cyanide-free and cyanide-bound ferric hIDO (a) and plot of the observed rate constants as a function of L-Trp
concentration (b). The kinetic traces of the cyanide-free hIDO (2.5 μM) and cyanide-bound hIDO (4.0 μM) shown in panel a were obtained with
50 μM L-Trp at 403 nm and 75 μM Trp at 421 nm, respectively. The solid lines in panel a are the best-fit curves with single-exponential functions.
All the samples were prepared in 100 mM Tris buffer (pH 7.4) at 20 °C, in the presence or absence of 1 mM cyanide.
7
-
1
-1
decay kinetics, with an observed rate constant of 10 s . The
observed rate constant depends on the -Trp concentration in a
linear fashion (Figure 4b). The bimolecular binding rate constant
decay kinetics, with an observed rate constant of 0.38 s . Con-
L
centration-dependent studies show that the k and k values are
on
off
3
-1 -1
-1
(4.1 ( 0.1) ꢁ 10 M
s
and 0.018 ( 0.006 s , respectively
(
1
k ) obtained from the slope of the best-fit line is (5.5 ( 0.1) ꢁ
(Figure 5b). The K
d
calculated from the koff/kon (4.4 μM) is
on
3
-1 -1
0 M
s
, while the dissociation rate constant (k ) obtained
consistent with that observed from equilibrium measurements
(2.4 μM) and is similar to the published value (3.6 μM) (23).
In the presence of L-Trp, the cyanide binding kinetics also
follows single-exponential kinetics (Figure 5a). It is important
to note that the reaction was conducted in the presence of 25 mM
off
-
1
from the intercept is 5.6 ( 0.1 s . The K calculated from the
d
koff/kon (1.0 mM) is consistent with that observed from equilib-
rium measurements (0.9 mM), confirming the reliability of the
kinetic measurements. It is noteworthy that there is a typo in the
4
-1 -1
unit of the previously published k , 1.55 ꢁ 10 μM
s
(23),
L
-Trp to ensure that all the protein molecules are in the
bound state during the reaction. With 25 mM
-Trp, ∼50% of
the molecules are in the two-Trp-bound state at the end of the
reaction (considering the fact that the K for the second sub-
L-Trp-
on
-
1 -1
which should be M
s
(Emma Raven, private communica-
L
3
-1 -1
tion), hence the kon we determined, 5.5 ꢁ 10 M
s , is similar
to the reported value.
d
In the presence of cyanide, the
found to be much faster (Figure 4a,b).
dependent data show that the kon and koff determined from
L
-Trp binding kinetics were
strate binding site is 26 mM). As such, two parallel reactions
illustrated below have to be taken into account (assuming that the
ferric ligand-free enzyme comprises only one substrate binding site).
L-Trp concentration-
5
the slope and intercept of the best-fit line are (1.2 ( 0.1) ꢁ 10
-
1
-1
-1
k
on
3þ
-
IDO - CN- ðTrpÞ
3þ
M
s
and 2.8 ( 0.4 s , respectively (Figure 4b). The Kd
IDO ðTrpÞ þ CN
s
k
f
3
r
s
3
calculated from the koff/kon (23 μM) again is consistent with that
observed from equilibrium measurements (18 μM). It is impor-
off
k
on
3þ
-
IDO - CN- ðTrpÞ
3þ
IDO ðTrpÞ þ CN
s
f
r
s
tant to note that the
than 0.3 mM; hence, only the first Trp binding site is occupied.
Kinetic studies conducted with higher -Trp concentrations
revealed that the binding of -Trp to the second binding site is
L-Trp concentrations employed here are less
3
3
k
off
0
k
on
L
IDO3þ- CN- ðTrpÞ2
þ Trp
s
k
f
r
s
3
L
0
off
too fast to be measured with our instrument.
Cyanide Binding Kinetics. To examine how the
L
-Trp
As the second reaction is limited in rate by cyanide binding
-
0
bound in the distal heme pocket perturbs ligand binding kinetics,
cyanide binding to the substrate-free and -Trp-bound enzyme
was studied with stopped-flow measurements. As shown in Figure
a, the kinetic trace at 404 nm obtained from the reaction of the
substrate-free enzyme with 87 μM cyanide follows single-exponential
(i.e., k [CN ] , k
[L-Trp]), the overall reaction follows single-
exponential kinetics with a rate constant of kon as demonstrated
on
on
L
in Figure 5a. Concentration-dependent studies show that the k
on
2
-1 -1
5
is (3.2 ( 0.1) ꢁ 10 M
s
while the koff is too slow to be
precisely determined (Figure 5b).

5
032 Biochemistry, Vol. 49, No. 24, 2010
Lu et al.
F
IGURE 5: Cyanide binding kinetics of substrate-free and L-Trp-bound ferric hIDO (a) and plot of the observed rate constants as a function of
cyanide concentration (b). The kinetic tracesofthe substrate-freehIDO(1.5μM)and L-Trp-boundhIDO(3.1μM)shown inpanel awereobtained
with 87 μM cyanide at 404 and 410 nm, respectively. The solid lines in panel a are the best-fit curves with single-exponential functions. All the
samples were prepared in 100 mM Tris buffer (pH 7.4) at 20 °C, in the presence or absence of 25 mM L-Trp.
Table 2: Kinetic and Thermodynamic Parameters Associated with Cyanide or
hIDO
L
-Trp Binding Reactions of hIDO, rIDO, and TDO
rIDO
TDO
-1
-1
-1
b
eq
d
eq
d
eq
d
k
on (M
s
)
k
off (s
)
K
d
(μM)
K
(μM)
K
(μM)
K
(μM)
2
þ
2
2
d
g
Fe þ Trp
ND
ND
(4.0 ( 0.7) ꢁ 10
(9.0 ( 0.9) ꢁ 10
13
4.12
3
þ
3
5
3
3d
3g
Fe þ Trp
(5.5 ( 0.1) ꢁ 10
(1.2 ( 0.1) ꢁ 10
5.6 ( 0.1
1.0 ꢁ 10
23
5.8 ꢁ 10
3.8 ꢁ 10
4f
3
.0 ꢁ 10
Fe -CN- þ Trp
3
þ
2.8 ( 0.4
18 ( 3
10-20
e
f
16
3
þ
-
3
2
f
Fe þ CN
(4.1 ( 0.1) ꢁ 10
(3.2 ( 0.1) ꢁ 10
0.018 ( 0.006
3.2 ꢁ 10
4.4
2.4 ( 0.2
11
3
þ
-
-5a
a
a
f
Fe -Trp þ CN
0.1
0.05
0.5
3þ
-
4
Fe -Trp- CN þ Trp
(μM)
(2.6 ( 0.8) ꢁ 10
c
d
g
K
m
15 ( 2
12
114
a
b
c
d
e
f
Calculated from the thermodynamic cycle illustrated in Figure 6. Calculated from kon/koff
.
From ref 17. From ref 16. From ref 34. From
g
ref 35. From ref 33.
F
IGURE 6: Thermodynamic cycle describing the cyanide and L-Trp binding reactions of hIDO. The parameters labeled in red were obtained from
equilibrium measurements. The parameters listed in the shaded boxes were obtained from kinetic measurements. The free energy (ΔG) values
associated with ligand or substrate binding were calculated from K , on the basis of the relationship ΔG = RT ln(K ). The side reaction labeled in
green is associated with the binding of L-Trp to the second substrate binding site in the cyanide-bound enzyme.
d
d
Implication for the Catalytic Cycle of hIDO. The thermo-
dynamic and kinetic parameters associated with binding of -Trp
and cyanide to hIDO obtained from this work are summarized in
Table 2. On the basis of the thermodynamic cycle illustrated in
560-fold, respectively, indicating that L-Trp bound in the distal
L
heme pocket restricts the access of ligand to the heme iron and
escape of ligand from the heme iron. The net result is tighter
cyanide binding, with a ΔΔG of -2.3 kcal/mol. On the other
Figure 6, the ΔG for cyanide binding to
L-Trp-bound ferric hIDO
hand, prebinding of cyanide to the enzyme facilitates L-Trp
is calculated to be -9.4 kcal/mol (or -9.8 kcal/mol based on the
equilibrium data). The K and k calculated from the ΔG are
binding by 22-fold but retards its dissociation by 2-fold, indicat-
ing that cyanide binding to the heme iron introduces structural
changes to the protein matrix allowing faster access of the
substrate to the active site and slower dissociation from it. The
net result is tighter substrate binding, with a ΔΔG of -2.3 kcal/mol.
d
-1
off
-
5
0
.1 μM and 3.2 ꢁ 10
s
, respectively.
Taken together, the data show that prebinding of
L-Trp to
the enzyme retards cyanide binding and dissociation by 13- and

Article
Biochemistry, Vol. 49, No. 24, 2010 5033
The positive cooperativity between cyanide and
L
-Trp binding
substrates, thereby producing the products. The catalytic cycle of
hIDO, on the other hand, is much less clear. Nonetheless, it is
generally believed that the dioxygenase reaction is initiated in the
ferrous state. Our data reported here suggest that the ferrous
(
i.e., cyanide binding increases the binding affinity of
vice versa) highlights the plasticity of the enzyme.
Steady-state kinetic studies show that the Km of
hIDO is 15 μM (17), which is 27-fold lower than the Kd(
of the ligand-free ferrous enzyme (0.4 mM), while the Km of O2
42 μM) (17) is similar to the K (O ) of the substrate-free ferrous
L-Trp, and
L-Trp for
L-Trp)
enzyme first binds O , followed by L-Trp binding, to generate the
2
ternary complex. The heme-bound O in the ternary complex is
2
(
then inserted into the C dC bond of the indole ring of the
d
2
2
3
enzyme (13 μM) (23), suggesting that during multiple turnovers
hIDO preferentially binds O prior to -Trp. The physiological
-Trp and O in tissues and plasma are typically
substrate, leading to a ferryl intermediate and a Trp epoxide (via
an alkylperoxo transition state), which subsequently recombine
to generate the product (11). We postulate that the sequential
L
2
concentrations of
in the ranges of 50-100 μM (28, 29) and 50-76 μM (30, 31),
respectively. On the basis of the Kd( -Trp) and K (O ) of the
ferrous enzyme (400 and 13 μM, respectively), 50 μM
L
2
binding of O and substrate during the catalytic cycle of hIDO is
2
L
important for downregulating the activity of the enzyme at high
L-Trp concentrations via its binding to the inhibitory substrate
d
2
L-Trp
forces ∼5% enzyme into the substrate-bound form, while 50 μM
binding site, which can be observed only in the ligand-bound
state.
O leads to >85% O -bound enzyme; as such, under physiolo-
2
2
gical conditions, prebinding of O followed by L-Trp binding, as
2
indicated by the red arrows in Scheme 2, is thermodynamically
favored. On the other hand, the data present in this work
demonstrate that ligand binding-induced structural changes in
ACKNOWLEDGMENT
We thank Dr. Denis L. Rousseau for valuable discussions.
the enzyme facilitate
binding-induced structural transition retards cyanide binding by
3-fold, suggesting that a sequential binding of O and -Trp is
L-Trp binding by 22-fold, while the L-Trp
REFERENCES
1. Kotake, Y., and Masayama, T. (1936) Uber den mechanismus der
kynurenin-bildung aus tryptophan. Z. Physiol. Chem. 243, 237–244.
2. Greengard, O., and Feigelsonp, P. (1962) The purification and proper-
ties of liver tryptophan pyrrolase. J. Biol. Chem. 237, 1903–1907.
3. Schutz, G., Chow, E., and Feigelson, P. (1972) Regulatory properties
of hepatic tryptophan oxygenase. J. Biol. Chem. 247, 5333–5337.
4. Ren, S., and Correia, M. A. (2000) Heme: A regulator of rat hepatic
tryptophan 2,3-dioxygenase? Arch. Biochem. Biophys. 377, 195–203.
1
L
2
also kinetically favored during the catalytic cycle.
One unique feature of the dioxygenase family of enzymes
with respect to other oxygen-utilizing heme proteins, such as
P450 and NOS, is the relatively weak substrate affinity in the
ferric state. As listed in Table 2, the substrate affinities of ferric
hIDO, rIDO, and xcTDO are in the millimolar window, which
are much lower than the affinity of ∼2 μM observed in P450 and
NOS (16, 17, 27, 32, 33). In P450 and NOS, the reduction of the
heme iron only slightly perturbs the substrate affinity (24, 27),
while that in rIDO and xcTDO leads to an ∼500-1000-fold
increase in substrate affinity (16, 33). In contrast, reduction of
hIDO improves the substrate affinity by only ∼2-fold. The low
substrate affinity of the ferrous hIDO is presumably important
5. Yamamoto, S., and Hayaishi, O. (1967) Tryptophan pyrrolase of
rabbit intestine. - and -tryptophan-cleaving enzyme or enzymes.
J. Biol. Chem. 242, 5260–5266.
D
L
6
7
8
. Muller, A. J., and Prendergast, G. C. (2007) Indoleamine 2,3-dioxy-
genase in immune suppression and cancer. Curr. Cancer Drug Targets
7
, 31–40.
. Katz, J. B., Muller, A. J., and Prendergast, G. C. (2008) Indoleamine
,3-dioxygenase in T-cell tolerance and tumoral immune escape.
2
Immunol. Rev. 222, 206–221.
. Uyttenhove, C., Pilotte, L., Theate, I., Stroobant, V., Colau, D.,
Parmentier, N., Boon, T., and Van den Eynde, B. J. (2003) Evi-
dence for a tumoral immune resistance mechanism based on trypto-
phan degradation by indoleamine 2,3-dioxygenase. Nat. Med. 9, 1269–
1274.
for ensuring that the enzyme binds O prior to substrate binding
2
during multiple turnovers. The physiological importance of this
unique property of hIDO remains to be investigated further;
9
. Muller, A. J., DuHadaway, J. B., Donover, P. S., Sutanto-Ward, E.,
and Prendergast, G. C. (2005) Inhibition of indoleamine 2,3-dioxy-
genase, an immunoregulatory target of the cancer suppression gene
Bin1, potentiates cancer chemotherapy. Nat. Med. 11, 312–319.
nonetheless, the sequential binding of O and
L-Trp may be
2
critical for downregulating the hIDO activity at high -Trp
L
concentrations via the substrate inhibition mechanism (as the
inhibitory substrate binding site can be observed only in the
ligand-bound state).
1
1
0. Sugimoto, H., Oda, S., Otsuki, T., Hino, T., Yoshida, T., and Shiro,
Y. (2006) Crystal structure of human indoleamine 2,3-dioxygenase:
Catalytic mechanism of O incorporation by a heme-containing
2
dioxygenase. Proc. Natl. Acad. Sci. U.S.A. 103, 2611–2616.
1. Lewis-Ballester, A., Batabyal, D., Egawa, T., Lu, C., Lin, Y., Marti,
M. A., Capece, L., Estrin, D. A., and Yeh, S. R. (2009) Evidence for a
ferryl intermediate in heme-based dioxygenases: Mechanistic implica-
tions. Proc. Natl. Acad. Sci. U.S.A. 106, 17371–17376.
CONCLUSIONS
Dioxygen-activating oxygenases can be divided into two
groups: (i) monooxygenases (such as P450s), which incorporate
one atom of the dioxygen into an organic substrate and reduce
the other atom of the dioxygen to water, and (ii) dioxygenases
12. Taniguchi, T., Sono, M., Hirata, F., Hayaishi, O., Tamura, M.,
Hayashi, K., Iizuka, T., and Ishimura, Y. (1979) Indoleamine 2,3-
dioxygenase. Kinetic studies on the binding of superoxide anion and
molecular oxygen to enzyme. J. Biol. Chem. 254, 3288–3294.
(
such as IDO and TDO), which insert both atoms of dioxygen
into the substrate. In contrast to the wide spectrum of P450-type
enzymes, hIDO and hTDO are the only two heme-based
dioxygenases found in humans. The catalytic cycle of P450 has
been studied in great detail (14, 24). In general, the substrate
binds the enzyme in the ferric state, which leads to the displace-
ment of the heme-bound water. It increases the redox potential of
the enzyme and triggers the transfer of the electron from the
reductase to the heme iron. The resulting ferrous heme iron binds
1
3. Ishimura, Y., Nozaki, M., and Hayaishi, O. (1970) The oxygenated
form of -tryptophan 2,3-dioxygenase as reaction intermediate.
J. Biol. Chem. 245, 3593–3602.
4. Sono, M., Roach, M. P., Coulter, E. D., and Dawson, J. H. (1996)
Heme-Containing Oxygenases. Chem. Rev. 96, 2841–2888.
L
1
15. Hirata, F., Ohnishi, T., and Hayaishi, O. (1977) Indoleamine 2,3-
-
dioxygenase. Characterization and properties of enzyme O
J. Biol. Chem. 252, 4637–4642.
2
complex.
3
1
1
6. Sono, M., Taniguchi, T., Watanabe, Y., and Hayaishi, O. (1980)
Indoleamine 2,3-dioxygenase. Equilibrium studies of the tryptophan
binding to the ferric, ferrous, and CO-bound enzymes. J. Biol. Chem.
O , leading to the ternary complex, in which the O-O bond is
2
255, 1339–1345.
heterolytically cleaved upon accepting an additional electron and
a proton. The cleavage reaction generates a key ferryl intermedi-
ate, which is capable of inserting a single oxygen into the organic
7. Lu, C., Lin, Y., and Yeh, S. R. (2009) Inhibitory substrate binding site
of human indoleamine 2,3-dioxygenase. J. Am. Chem. Soc. 131,
12866–12877.

5
034 Biochemistry, Vol. 49, No. 24, 2010
Lu et al.
1
8. Papadopoulou, N. D., Mewies, M., McLean, K. J., Seward, H. E.,
Svistunenko, D. A., Munro, A. W., and Raven, E. L. (2005) Redox
and Spectroscopic Properties of Human Indoleamine 2,3-Dioxygen-
ase and A His303Ala Variant: Implications for Catalysis. Biochem-
istry 44, 14318–14328.
27. Presta, A., Weber-Main, A. M., Stankovich, M. T., and Stuehr, D. J.
(1998) Comparative effects of substrates and pterin cofactor on the
heme midpoint potential in inducible and neuronal nitric oxide
synthases. J. Am. Chem. Soc. 120, 9460–9465.
28. Terness, P., Kallikourdis, M., Betz, A. G., Rabinovich, G. A., Saito,
S., and Clark, D. A. (2007) Tolerance signaling molecules and
pregnancy: IDO, galectins, and the renaissance of regulatory T cells.
Am. J. Reprod. Immunol. 58, 238–254.
29. Torres, M. I., Lopez-Casado, M. A., Lorite, P., and Rios, A. (2007)
Tryptophan metabolism and indoleamine 2,3-dioxygenase expression
in coeliac disease. Clin. Exp. Immunol. 148, 419–424.
30. Bunn, H. F., and Poyton, R. O. (1996) Oxygen sensing and molecular
adaptation to hypoxia. Physiol. Rev. 76, 839–885.
1
2
9. Nickel, E., Nienhaus, K., Lu, C., Yeh, S. R., and Nienhaus, G. U.
(2009) Ligand and substrate migration in human indoleamine 2,3-
dioxygenase. J. Biol. Chem. 284, 31548–31554.
0. Terentis, A. C., Thomas, S. R., Takikawa, O., Littlejohn, T. K.,
Truscott, R. J., Armstrong, R. S., Yeh, S. R., and Stocker, R. (2002)
The heme environment of recombinant human indoleamine 2,3-
dioxygenase. Structural properties and substrate-ligand interactions.
J. Biol. Chem. 277, 15788–15794.
2
1. Sohl, C. D., Isin, E. M., Eoff, R. L., Marsch, G. A., Stec, D. F., and
Guengerich, F. P. (2008) Cooperativity in oxidation reactions cata-
lyzed by cytochrome P450 1A2: Highly cooperative pyrene hydro-
xylation and multiphasic kinetics of ligand binding. J. Biol. Chem.
31. McCormick, C. C., Li, W. P., and Calero, M. (2000) Oxygen tension
limits nitric oxide synthesis by activated macrophages. Biochem. J.
350, 709–716.
32. Kim, D., Heo, Y. S., and Ortiz de Montellano, P. R. (2008) Efficient
catalytic turnover of cytochrome P450(cam) is supported by a T252N
mutation. Arch. Biochem. Biophys. 474, 150–156.
33. Forouhar, F., Anderson, J. L., Mowat, C. G., Vorobiev, S. M.,
Hussain, A., Abashidze, M., Bruckmann, C., Thackray, S. J.,
Seetharaman, J., Tucker, T., Xiao, R., Ma, L. C., Zhao, L., Acton,
T. B., Montelione, G. T., Chapman, S. K., and Tong, L. (2007)
Molecular insights into substrate recognition and catalysis by trypto-
phan 2,3-dioxygenase. Proc. Natl. Acad. Sci. U.S.A. 104, 473–478.
34. Uchida, K., Shimizu, T., Makino, R., Sakaguchi, K., Iizuka, T.,
Ishimura, Y., Nozawa, T., and Hatano, M. (1983) Magnetic and
2
83, 7293–7308.
2. Lu, C., Mukai, M., Lin, Y., Wu, G., Poole, R. K., and Yeh, S. R.
2007) Structural and functional properties of a single domain
2
(
hemoglobin from the food-borne pathogen Campylobactor jejuni. J.
Biol. Chem. 282, 25917–25928.
2
3. Chauhan, N., Basran, J., Efimov, I., Svistunenko, D. A., Seward, H.
E., Moody, P. C., and Raven, E. L. (2008) The role of serine 167 in
human indoleamine 2,3-dioxygenase: A comparison with tryptophan
2,3-dioxygenase. Biochemistry 47, 4761–4769.
2
4. Denisov, I. G., Makris, T. M., Sligar, S. G., and Schlichting, I. (2005)
Structure and chemistry of cytochrome P450. Chem. Rev. 105, 2253–2277.
5. Rousseau, D. L., Li, D., Couture, M., and Yeh, S. R. (2005) Ligand-
protein interactions in nitric oxide synthase. J. Inorg. Biochem. 99,
natural circular dichroism of
L-tryptophan 2,3-dioxygenases and
2
indoleamine 2,3-dioxygenase. II. Spectra of their ferric cyanide and
ferrous carbon monoxide complexes and an oxygenated form. J. Biol.
Chem. 258, 2526–2533.
3
6. Das, A., Grinkova, Y. V., and Sligar, S. G. (2007) Redox potential
06–323.
2
35. Sono, M. (1990) Spectroscopic and equilibrium studies of ligand and
organic substrate binding to indolamine 2,3-dioxygenase. Biochemi-
stry 29, 1451–1460.
control by drug binding to cytochrome P450 3A4. J. Am. Chem. Soc.
1
29, 13778–13779.