|Home | About | Journals | Submit | Contact Us | Français|
The heme enzyme indoleamine 2,3-dioxygenase (IDO) is a key regulator of immune responses through catalyzing l-tryptophan (l-Trp) oxidation. Here, we show that hydrogen peroxide (H2O2) activates the peroxidase function of IDO to induce protein oxidation and inhibit dioxygenase activity. Exposure of IDO-expressing cells or recombinant human IDO (rIDO) to H2O2 inhibited dioxygenase activity in a manner abrogated by l-Trp. Dioxygenase inhibition correlated with IDO-catalyzed H2O2 consumption, compound I-mediated formation of protein-centered radicals, altered protein secondary structure, and opening of the distal heme pocket to promote nonproductive substrate binding; these changes were inhibited by l-Trp, the heme ligand cyanide, or free radical scavengers. Protection by l-Trp coincided with its oxidation into oxindolylalanine and kynurenine and the formation of a compound II-type ferryl-oxo heme. Physiological peroxidase substrates, ascorbate or tyrosine, enhanced rIDO-mediated H2O2 consumption and attenuated H2O2-induced protein oxidation and dioxygenase inhibition. In the presence of H2O2, rIDO catalytically consumed nitric oxide (NO) and utilized nitrite to promote 3-nitrotyrosine formation on IDO. The promotion of H2O2 consumption by peroxidase substrates, NO consumption, and IDO nitration was inhibited by l-Trp. This study identifies IDO as a heme peroxidase that, in the absence of substrates, self-inactivates dioxygenase activity via compound I-initiated protein oxidation. l-Trp protects against dioxygenase inactivation by reacting with compound I and retarding compound II reduction to suppress peroxidase turnover. Peroxidase-mediated dioxygenase inactivation, NO consumption, or protein nitration may modulate the biological actions of IDO expressed in inflammatory tissues where the levels of H2O2 and NO are elevated and l-Trp is low.
Indoleamine 2,3-dioxygenase (IDO)4 is an intracellular heme enzyme that catalyzes the initial and rate-limiting step of l-tryptophan (l-Trp) metabolism along the kynurenine pathway (1–5). IDO catalyzes the oxidative cleavage of the pyrrole ring of l-Trp by insertion of molecular oxygen (O2) to generate N-formylkynurenine, which hydrolyzes into kynurenine and formate. Expression of IDO is induced at sites of inflammation in vivo principally by interferon-γ (IFNγ) (1, 3, 5). Induction of IDO is considered to be part of the innate immune response of the host and is thought to limit the growth of certain pathogens and tumor cells via local l-Trp depletion, the least abundant of all essential amino acids (1, 3, 5, 6). Recent in vivo studies have also established a major role for IDO in immune regulation by promoting immune tolerance via suppression of local T cell responses under various physiological and pathophysiological conditions, including mammalian pregnancy, tumor resistance, auto-immunity, chronic inflammation, and chronic infections (3, 6, 7). In light of the increasing awareness of the important roles of the enzyme, it is critical to understand how IDO activity is controlled and the scope of reactions it catalyzes.
IDO is one of two known mammalian heme dioxygenases; the other is tryptophan 2,3-dioxygenase (TDO). Although these enzymes catalyze the same dioxygenase reaction and contain similar heme active sites (8–10), IDO and TDO are distinct enzymes that are monomeric and homotetrameric, respectively, share only 10% sequence identity (9), and show differences in tissue distribution, protein structure, substrate specificity and binding, and enzyme reactivity (11–13).
The active-site heme is essential for IDO dioxygenase activity; reduction of heme from the ferric (FeIII) to the ferrous (FeII) form facilitates binding of O2 and l-Trp to form the active ternary complex (14, 15). Early studies on the dioxygenase reaction mechanism proposed that an initial de-protonation of a substrate's indole amino group is followed by the formation of 3-hydroperoxyindolenine and dioxetane intermediates (15, 16). The first resonance Raman spectroscopy (17) and x-ray crystallography (8) studies of recombinant human IDO (rIDO) supported this mechanism. However, recent resonance Raman data support a role for a ferryl-oxo (FeIV=O), compound II-type species (18, 19). These findings along with other recent experimental and theoretical data support a new mechanism where the initial ternary complex (FeIII-superoxo-Trp) reacts via a radical 2-alkyl-peroxo-transition state to form compound II and indole 2,3-epoxide (Trp-epoxide). The next step is postulated to involve proton transfer from the NH3+ group of l-Trp to the epoxide oxygen, triggering ring opening. The final step yielding N-formylkynurenine involves a concerted nucleophilic attack of the FeIV=O oxygen to C2 of l-Trp, C2–C3 bond cleavage, and back proton transfer from the C3 oxygen to the NH3+ group of l-Trp (15, 18, 20–23).
The involvement of a FeIV=O species is akin to the reaction mechanism of heme peroxidases. Although some studies support that IDO can exhibit a peroxidase activity to oxidize synthetic substrates or melatonin (24–27), the role of IDO as a catalyst of physiologically relevant peroxidase reactions and the implications of its peroxidase activity for the dioxygenase activity of the enzyme have received little attention. Heme peroxidases are a diverse enzyme family that performs a wide range of biological functions via coupling the reduction of hydrogen peroxide (H2O2) with the oxidation of low molecular weight substrates. Heme peroxidases exhibit a three-step catalytic cycle (Scheme 1). The first reaction is between FeIII heme and H2O2 to form a two-electron oxidized intermediate, compound I, that consists of a FeIV=O iron center and a reactive radical cation centered on the heme porphyrin ring. In certain cases, the radical in compound I resides on an amino acid, e.g. Trp in cytochrome c peroxidase (28). One-electron reduction of this radical yields compound II, referring to the remaining FeIV=O iron center. One-electron reduction of compound II forms the FeIII enzyme. In heme peroxidases, the reduction of the FeIV=O center in compound II by H2O2 or reducing substrates is considered the rate-limiting step for enzyme turnover (29). Heme peroxidases oxidize a wide array of substrates, commonly organic, low molecular weight reductants. The nature of the preferred substrate depends on the local environment of the enzyme, e.g. physiological substrates for mammalian heme peroxidases expressed within inflammatory tissues include amino acids (tyrosine, l-Trp), low molecular weight antioxidants (ascorbate, urate), nitric oxide (NO) and its oxidation product, nitrite (NO2−) (30–36).
For bona fide heme peroxidases, formation of compounds I and II is inherent in their catalytic cycle. However, the respiratory hemoproteins, hemoglobin and myoglobin, exhibit a “rogue” pseudo-peroxidase activity where reaction of their FeIII form with H2O2 forms compound I, which rapidly reacts via the porphyrin radical cation with adjacent amino acids to form protein-centered radicals, leading to oxidative protein damage (37). In vivo evidence supports a deleterious role for the heme peroxidase activity of myoglobin and hemoglobin in inflammatory disorders such as ischemia reperfusion and rhabdomyolysis (37–39).
IDO is commonly expressed in inflammatory cells and tissues (1, 3, 5, 6) characterized by increased production of H2O2 and NO that during inflammation can be produced at micromolar concentrations over time (40–42), e.g. activated leukocytes at blood concentrations (1.5–6 million cells/ml) produce H2O2 concentrations of 80–480 μm/h (40), whereas a steady-state concentration of ~400 nm NO has been estimated in local inflammatory environments (41, 42). In this study, we show for the first time that H2O2 engages the pseudo-peroxidase activity of IDO to mediate irreversible oxidative self-inactivation of its dioxygenase activity and to react with various physiological substrates, including l-Trp, ascorbate, tyrosine, nitrite, and NO.
H2O2 was purchased from Merck. NOC-9 was from Santa Cruz Biotechnology. High purity 5,5-dimethyl-1-pyrroline-N-oxide (DMPO) was from Enzo Life Sciences and was purified over activated charcoal as described previously (43). Rabbit polyclonal anti-DMPO nitrone adduct antibody was from Cayman Chemicals. Unless indicated, all other materials were from Sigma and of the highest purity available. The concentrations of l-Trp and H2O2 were determined by their extinction coefficients ϵ280 = 5.5 mm−1 cm−1 and ϵ240 = 43.6 mm−1 cm−1, respectively.
rIDO encoded by the pQE9-IDO plasmid vector was expressed in Escherichia coli as a hexahistidyl fusion protein and purified as detailed (17, 44). IDO concentration was expressed as the enzyme's heme concentration determined using the extinction coefficient ϵ405 = 159 mm−1 cm−1. For experiments, rIDO was prepared in 50–100 mm potassium phosphate buffer, pH 7.4, supplemented with 50–100 μm diethylenetriaminepentaacetic acid (DETAPAC) to prevent spurious trace-metal redox chemistry. Typically, reactions were initiated by addition of H2O2 (0–100 mol eq) in the absence or presence of experimental treatments (i.e. l-Trp, cyanide, tempol, ascorbate, tyrosine, NO2−, NO, and DMPO) at 22 °C. For studies examining l-Trp oxidation by rIDO in the presence of H2O2, reactions were initiated by the simultaneous addition of l-Trp and H2O2 as a mixture to FeIII-rIDO. For l-Trp oxidation experiments under O2-depleted conditions, solutions of enzyme and reactants were degassed with argon for ~30 min under an air-tight septum seal prior to reaction initiation. Heat inactivation of rIDO was achieved by incubating the enzyme at 100 °C for 15 min. Where required, unreacted H2O2 was removed by the addition of catalase (250 μg/ml) or superoxide anion radical (O2) produced and removed by the addition of superoxide dismutase (200 units/μl). In some experiments, reaction solutions were gel-filtered using NAP5 columns (GE Healthcare) in 50 mm potassium phosphate buffer, pH 7.4. For immuno-spin trapping, rIDO was exposed to H2O2 in the presence of 100 mm DMPO, incubated for 30 min, and analyzed by Western blot.
The dioxygenase activity of rIDO or IDO in cell lysates was determined by the ascorbate/methylene blue assay under standard assay conditions (45). For rIDO activity measurements, rIDO was nontreated (control) or treated with H2O2 for 10 min prior to dilution of rIDO to a final concentration of 50–100 nm into enzyme assay buffer (i.e. 50 mm potassium phosphate buffer, pH 7.4, containing 0.5 mm EDTA, 25 μm methylene blue, 10 mm sodium ascorbate, 250 μg/ml catalase and 400 μm l-Trp) and incubated at 37 °C for 15 min. For cellular IDO, cell lysates isolated from vehicle-treated (control) or H2O2-treated cells were diluted 1:1 (v/v) with enzyme assay buffer and incubated at 37 °C for 30 min. The reaction was then terminated by addition of 20% (w/v) trichloroacetic acid (1:4 v/v), and the mixture was stored overnight at 4 °C to allow the N-formylkynurenine formed to hydrolyze into the more stable product kynurenine. IDO dioxygenase activity was determined by measuring the amount of l-Trp converted into kynurenine by HPLC.
l-Trp, kynurenine, and oxindolylalanine were measured by reversed phase HPLC (Agilent 1200 system) with a Hypersil 3-μm ODS C18 column (Phenomenex) and eluted with 100 mm chloroacetic acid, 9% v/v acetonitrile, pH 2.4, at 0.5 ml/min. l-Trp, kynurenine, and oxindolylalanine were detected by UV-visible absorbance at 280, 364, and 256 nm, respectively. The concentrations of l-Trp, kynurenine, or oxindolylalanine were calculated by comparing the peak area of the sample with the peak area of a known concentration of authentic standards. Authentic standards of oxindolylalanine were synthesized from l-Trp and purified as described previously with the UV spectrum showing a characteristic λmax (H2O) of 250 nm (46). The concentration of the oxindolylalanine standard was determined by its extinction coefficient: ϵ250 = 7250 m−1 cm−1 (47).
Optical absorption spectra of rIDO (5–10 μm) were recorded in the absence or presence of H2O2 treatment under atmospheric conditions in 50 mm potassium phosphate buffer supplemented with 50–100 μm DETAPAC, pH 7.4, in the absence or presence of l-Trp (200 μm, 1 mm) using a Varian-Cary 300 spectrophotometer and quartz cuvettes (1-cm path length).
Samples of rIDO (125 μl, 30–40 μm enzyme in 100 mm potassium phosphate buffer, pH 7.4) for RR measurements were prepared in a septum-sealed cylindrical quartz cell. To measure changes to the distal heme pocket environment, RR spectra of CO complexes of rIDO and H2O2-treated rIDO were performed as described previously (48). For this, the FeII-deoxy form of rIDO was first prepared by flushing the enzyme sample with argon and then injecting a 100-fold mole excess of buffered sodium dithionite solution. FeII-deoxy samples were then flushed with carbon monoxide (CO) gas (Airgas Specialty Gases) to form the FeII-CO complex. For H2O2-treated rIDO samples, the enzyme was first treated with a 10-fold mole excess of H2O2 for 10 min under a normal atmosphere and then flushed with argon and reduced with dithionite and finally CO. The final concentration of l-Trp when added was 1 mm. l-Trp was added either before or after H2O2 treatment, as indicated. The rotating sample cell was irradiated with 5 milliwatts of 413.1 nm light using a mixed krypton/argon ion laser (Spectra Physics, Beamlok 2060). The spectral acquisition time was 10 min. The scattered light was collected at right angles to the incident beam and focused onto the entrance slit (125 μm) of a 0.8-m spectrograph, where it was dispersed by a 600-groove/mm grating and detected by a liquid N2-cooled CCD camera (Horiba-JY). Spectral calibration was performed against the lines of mercury. For the detection of the FeIV-heme intermediate, RR measurements were performed under a normal atmosphere. The rIDO sample with or without l-Trp (1 mm) was treated with a 10-fold mole excess of H216O2 or H218O2 (Icon Isotopes, Summit, NJ), and spectral acquisition was started immediately. Spectra were recorded every 20 s. Typically, the first two or three spectra in the time course were combined and averaged to calculate the H216O2-H218O2 difference spectrum.
CD spectra were recorded with a Jasco 810 spectropolarimeter. A sample of rIDO (2 μm) in 50 mm potassium phosphate buffer, pH 7.4, was placed in a 1-mm path length quartz cuvette. H2O2 was cumulatively added to the buffered enzyme solution at the specified mole ratios. After each H2O2 addition, the sample was incubated for 10 min before the CD spectrum was recorded (parameter settings: 25 °C; 320 to 185 nm; scan rate 100 nm/min; bandwidth 1 nm; response time 1 s; sensitivity of 100 millidegrees). Spectra were acquired from an average of 8 scans, and all final spectra were corrected for buffer background CD signals. In certain experiments, 1.6 μm rIDO in phosphate buffer with 1 mm l-Trp was treated with 10 mol eq of H2O2 for 10 min before the sample was filtered with a PD Minitrap G-25 column, and then the CD spectrum of the enzyme (now diluted to ~450 nm) was measured. In this case, a 1-cm path length cuvette was used, and six scans were taken. All other measurement parameters were the same. All CD spectra were modeled using the CDPro software package employing the SDP42 and SDP48 data sets (49, 50). The reported percent helix, β-form, and unresolved structure values were determined by averaging the results from the SELCON3, CDSSTR and CONTINLL programs.
Binding of l-Trp to rIDO was determined as the magnitude of absorbance change at 404.5 nm in response to increasing amounts of l-Trp, as described previously (48). Absorption spectra were recorded with a PerkinElmer Life Sciences Lambda EZ210 spectrophotometer in dual beam mode using quartz cuvettes. For experiments, rIDO (5 μm) in 100 mm potassium phosphate buffer, pH 7.4, was nontreated (control) or treated with H2O2 (10 mol eq) for 10 min, ascorbate (1 mm) added (to reduce any residual IDO compound II to FeIII-IDO) for 5 min, and the reaction solution passed through PD Minitrap G-25 columns to remove residual H2O2 and ascorbate. The filtered rIDO was used immediately for the titration experiment. Aliquots of concentrated l-Trp stocks were cumulatively added to the rIDO solution (~3 μm) in quartz cuvettes. Absorbance changes were corrected for the small dilution factors obtained due to the addition of the concentrated l-Trp stock. Binding curves were fitted with a standard Langmuir binding isotherm function corresponding to either one or two independent binding sites using Origin 8 software.
The rates of consumption of H2O2 and NO or production of O2 in real time by rIDO in 100 mm phosphate buffer (pH 7.4 supplemented with 50 μm DETAPAC) were measured using H2O2- (ISO-HPO-2), NO- (ISO-NOP), or O2-specific (ISO-OXY2) electrodes interfaced with an Apollo 4000 free radical analyzer at 22 °C (World Precision Instruments). For H2O2 experiments, 20 μm of the oxidant was added to the buffer followed by the addition of rIDO (0.625–2.5 μm) in the absence or presence of l-Trp (50–1000 μm), ascorbate (50 μm), or tyrosine (50 μm), and the rate of H2O2 consumption was monitored. In some experiments, consumption of H2O2 by rIDO was measured using the Amplex Red Assay kit according to the manufacturer's instructions (Molecular Probes). For O2 experiments, the time-dependent change in O2 concentration was monitored in buffer containing H2O2 (200 μm), before and after addition of rIDO (5 μm) or catalase (1 μm). For NO experiments, the NO donor NOC-9 (t½ ~3 min, pH 7.4, 22 °C) was added to the buffer with rapid mixing to achieve a maximal steady-state concentration of ~3 μm NO. H2O2 (20 μm) and rIDO (0.1–1 μm) were added sequentially in the absence or presence of l-Trp (50 μm), and the rate of NO consumption was measured after rIDO addition.
Chinese hamster ovary (CHO) cells were cultured in F-12 medium supplemented with 10% fetal bovine serum, glutamine, and penicillin/streptomycin. For transfection, CHO cells were seeded into 6-well tissue culture plates and grown to 90% confluence. Cells were then cultured in serum-free Opti-MEM medium and transfected with pcDNA3 encoding full-length human IDO cDNA (1 μg/well) using FuGENE 6 transfection reagent (Roche Applied Science) according to the manufacturer's instructions. After overnight transfection, IDO-expressing CHO cells were incubated in Hanks' balanced salt solution (HBSS) in the absence or presence of l-Trp (0–50 μm) and H2O2 (0–1 mm) for 30 min after which time the cells were washed and lysed in cell lysis buffer (Cell Signaling). Cell lysates were then assayed for IDO dioxygenase activity using the ascorbate/methylene blue assay or IDO protein expression by Western blot.
Cell proteins or rIDO were resolved on 10% NUPAGE BisTris acetate polyacrylamide gels (Invitrogen). Electrophoresed proteins were transferred onto nitrocellulose membranes using the i-Blot system (Invitrogen), and IDO was detected by Western blot using a mouse monoclonal antibody against human IDO (generously provided by Dr. Osamu Takikawa, National Institute for Longevity Sciences, Japan) as described (51). The formation of 3-nitrotyrosine or DMPO adducts on rIDO was assessed using a mouse monoclonal anti-3-nitrotyrosine antibody (Clone 1A6, Millipore) or rabbit polyclonal anti-DMPO nitrone-adduct antibody (Cayman Chemicals), respectively.
Initial studies tested the ability of H2O2 to alter cellular IDO activity. To address this, we exposed CHO cells expressing active human IDO to increasing H2O2 concentrations for 30 min before measurement of dioxygenase activity in cell lysates by the ascorbate/methylene blue assay and IDO protein levels by Western blot. Although H2O2 did not alter IDO protein levels, H2O2 exposure inhibited IDO dioxygenase activity in a dose-dependent manner (Fig. 1A). Prior incubation of IDO-transfected cells with micromolar levels of l-Trp (0–50 μm) dose-dependently abrogated the ability of H2O2 to inhibit cellular IDO dioxygenase activity (Fig. 1B).
Having demonstrated that H2O2 inhibits cellular IDO activity at the post-translational level, we next examined if the oxidant directly inhibited the dioxygenase activity of purified human rIDO. For this, we treated rIDO with H2O2 at low oxidant to enzyme molar ratios (1:1–100:1 mol/mol) for 10 min prior to the addition of the co-factors ascorbate and methylene blue in the presence of catalase and then measured dioxygenase activity. H2O2 pretreatment inhibited rIDO dioxygenase activity in a dose-dependent manner with ~50% inhibition achieved with 2.5–5 mol eq of H2O2 (Fig. 1C). Similar to IDO-expressing cells, supplementation of rIDO with micromolar concentrations of l-Trp prior to H2O2 addition significantly abrogated the inhibitory action of the oxidant on dioxygenase activity (Fig. 1D).
As well as binding avidly to the IDO active site, l-Trp is a known substrate for compound I, but not compound II, of peroxidases (32, 52). In light of this, we also examined if other peroxidase substrates could modulate H2O2-induced inhibition of dioxygenase activity. Prior addition of ascorbate or tyrosine, effective substrates for compounds I and II (31, 53), also protected rIDO against dioxygenase inhibition afforded by H2O2 exposure, with ascorbate more protective than tyrosine (Fig. 1E). These data indicate that H2O2 directly inhibits IDO dioxygenase activity in a manner abrogated by l-Trp, ascorbate, and tyrosine, which share in common their ability to rapidly react with compound I of peroxidases.
We next measured the propensity of IDO to catalytically consume H2O2 via its peroxidase activity and the effect of l-Trp, ascorbate, and tyrosine on this, using a H2O2-specific electrode. Fig. 2 shows that rIDO consumed H2O2 in a time-dependent manner with the total amount of H2O2 consumed being supra-stoichiometric with respect to the concentration of rIDO added, i.e. ~12 μm of H2O2 was consumed by 2.5 μm rIDO after 8 min (Fig. 2A). Also, the initial rate of H2O2 consumption was directly proportional to the concentration of rIDO added (Fig. 2B). Although l-Trp did not affect the initial rapid period of H2O2 consumption by rIDO (i.e. 0–0.5 min), the amino acid significantly reduced H2O2 consumption over the ensuing time period (i.e. 0.5–8 min) (Fig. 2C). In contrast to l-Trp, addition of ascorbate or tyrosine promoted the consumption of H2O2 by rIDO (Fig. 2, D and E). Co-addition of equimolar concentrations of l-Trp attenuated the ability of ascorbate (Fig. 2D) and prevented the ability of tyrosine (Fig. 2E) to stimulate rIDO-mediated H2O2 consumption. These findings are indicative of IDO exhibiting a heme peroxidase function, where enzyme turnover and consequently H2O2 consumption is accelerated by effective substrates for compound II (i.e. ascorbate and tyrosine) (31, 53) and inhibited by l-Trp, a poor compound II reductant (32, 52). Spectroscopic studies (see below) support this proposal as IDO compound II is the major heme species detected in the presence of H2O2 and l-Trp.
In the absence of other available substrates, heme peroxidases can exhibit a pseudo-catalase activity, where H2O2 is converted into O2 (54, 55). Using an O2-electrode, we found that in contrast to catalase, addition of H2O2 to rIDO did not result in a net increase in O2 levels (Fig. 2F).
The data above support that the ability of H2O2 to inhibit IDO dioxygenase activity involves the activation of the enzyme's peroxidase activity and resultant formation of the highly reactive compound I, which in the pseudo-peroxidases myoglobin or hemoglobin, rapidly reacts via the porphyrin radical cation with adjacent amino acids to form protein-centered radicals, leading to oxidative protein damage (37). To test if exposure of rIDO to H2O2 affords compound I-mediated formation of protein-centered radicals, we employed the nitrone spin trap DMPO and performed immuno-spin trapping coupled to Western blotting with an anti-DMPO antibody (56). Addition of H2O2 to FeIII-rIDO, at low oxidant to enzyme ratios that inhibited dioxygenase activity, induced a dose-dependent increase in the formation of DMPO-protein adducts on rIDO (Fig. 3A). Protein-centered radical formation correlated with the appearance of higher molecular weight rIDO oligomers that are indicative of covalent cross-linking of rIDO monomers (Fig. 3A). No nitrone adduct was observed in rIDO incubated with DMPO in the absence of H2O2. The H2O2-mediated formation of protein-centered radicals was largely prevented by incubation of rIDO with the heme ligand cyanide, which prevents compound I formation, or by the compound I substrates and/or radical scavengers tempol, l-Trp, ascorbate, and tyrosine, prior to oxidant addition (Fig. 3B).
The finding that l-Trp inhibits H2O2-induced formation of protein-centered radicals supports that the amino acid is preferentially oxidized by compound I and that this protects the enzyme against protein oxidation and dioxygenase inactivation. We therefore next measured the extent to which l-Trp is oxidized by rIDO upon addition of H2O2 using HPLC. The addition of H2O2 and physiological concentrations of l-Trp (100 μm) simultaneously as a mixture to FeIII-rIDO initiated both time-dependent (Fig. 4, A and B) and enzyme concentration-dependent (Fig. 4C) consumption of l-Trp in a manner that was prevented by the heme ligand cyanide and inhibited by the compound I substrates and radical scavengers, tempol or ascorbate (Fig. 4D), but was not affected by superoxide dismutase (200 units/μl) (data not shown). l-Trp oxidation was not observed upon addition of H2O2 to heat-inactivated rIDO enzyme (data not shown).
Similar to H2O2 consumption (Fig. 2), oxidation of l-Trp was supra-stoichiometric with respect to rIDO concentration, i.e. ~30 μm of l-Trp was consumed by 5 μm rIDO after a 30-min reaction (Fig. 4A). Moreover, in the initial rapid phase (i.e. 0–2 min) of l-Trp oxidation, an ~1:1 mol/mol stoichiometry was apparent for the consumption of H2O2 and l-Trp (Fig. 4E). With further incubation, the reaction stoichiometry increasingly favored H2O2 consumption such that ~2 molecules of H2O2 were consumed per molecule of l-Trp consumed (Fig. 4E).
When acting as a dioxygenase, IDO catalyzed the stoichiometric conversion (i.e. 1:1) of l-Trp into N-formylkynurenine, which hydrolyzes into its more stable product, kynurenine. However, although treatment of rIDO with H2O2 and l-Trp resulted in the formation of N-formylkynurenine (measured as its more stable hydrolysis product, kynurenine), it accounted for ≤30% of the total amount of l-Trp oxidized (Fig. 4, A–C), indicating that other oxidation products are formed. HPLC analysis of the reaction mixture of rIDO, l-Trp, and H2O2 at 256 nm revealed the formation of new peaks (refer to peaks I and III; Fig. 5C) that co-eluted with an oxindolylalanine standard (Fig. 5B), which upon formation readily yields diastereoisomers upon equilibration (46). Fig. 5C, peak II, was identified as kynurenine, which also absorbs at 256 nm (Fig. 5A). Consistent with peaks I and III representing diastereoisomers of oxindolylalanine, when either peak was collected and individually re-analyzed by HPLC, they provided the same pair of peaks, I and III (data not shown). Moreover, collection of the HPLC peak I and subsequent LC/ESI-MS analysis of the collected peak indicated the formation of a product with m/z ratio of ~221, a finding indicative of the incorporation of one oxygen atom into the indole ring of l-Trp to form oxindolylalanine (Mr = 220.2) (data not shown). Quantitation of oxindolylalanine levels showed that, depending on the reaction conditions, its formation accounted for 11–17% of the total amount of l-Trp oxidized by rIDO treated with H2O2, compared with yields of 14–30% for kynurenine (Fig. 4, A–C). Oxindolylalanine was not detected when rIDO was incubated with l-Trp in the absence of H2O2 or when H2O2 and l-Trp were incubated in the absence of rIDO (data not shown). Comparison of H2O2-induced l-Trp oxidation by rIDO under normal atmosphere versus O2-depleted reaction conditions revealed that although the levels of oxindolylalanine formed did not differ, kynurenine levels were inhibited by ~60% under O2-depleted conditions (Fig. 4F). Together, these findings support that l-Trp is oxidized by IDO compound I to form oxindolylalanine and kynurenine, involving O2-independent and O2-dependent reactions.
Heme is essential for IDO dioxygenase activity with its reduction from the FeIII to the FeII form necessary for O2 and l-Trp binding to form the active ternary species (14). Also, H2O2 converts the FeIII form of peroxidases into the short-lived compound I and longer lived compound II species, both characterized by a FeIV=O heme (Scheme 1). Employing optical absorption and RR spectroscopy, we monitored changes to the rIDO heme upon exposure to H2O2 at concentrations that effectively inhibit dioxygenase activity (i.e. 1:1–25:1 mol/mol), in the absence or presence of l-Trp.
Addition of H2O2 to FeIII-rIDO in the absence of l-Trp induced within the first 0.5 min a dose-dependent decrease in intensity and red shift of the γ-Soret band from 404 to 407 nm (Fig. 6A). For the α and β bands, H2O2 induced a dose-dependent increase in peak intensities at ~538 and ~576 nm and a decrease in intensities at 500 and 630 nm (Fig. 6B). In the presence of l-Trp, addition of H2O2 to rIDO also resulted in a dose-dependent decrease in intensity and red shift of the γ-Soret band; this time from 404.5 nm to a maximum of 412 nm (Fig. 6C). For the α and β bands, the presence of l-Trp afforded increased band intensities at ~541, ~576, and ~590 nm and reduced intensities at 500 and 630 nm (Fig. 6D). These spectral changes and parameters in the presence of l-Trp are indicative of formation of a FeIV=O, compound II-type species, which was recently documented for human IDO (26).
The RR spectrum of FeIII-rIDO heme indicates a 6-coordinate, mixed high and low spin iron center, as shown by ν2 and ν3 doublets at 1564/1582 and 1482/1505 cm−1, respectively (Fig. 7A) (17). A time-dependent study of the spin-state changes to the FeIII species induced by H2O2 revealed that 10 mol eq of the oxidant caused an almost complete conversion of the heme iron to the low spin state within the first 20 s of treatment, as indicated by the increased intensity of the 1582 and 1505 cm−1 peaks corresponding to the low spin heme iron component, which then rapidly decayed back to the high spin state, although not completely back to the level observed for the untreated FeIII form within the 2-min time frame of the measurements (Fig. 7A). In the presence of l-Trp, H2O2 treatment afforded a similar immediate shift to low spin in the RR spectrum of rIDO heme, although this time the low spin species appeared not to decay as rapidly (Fig. 7B). The appearance of a transient, 6-coordinate low spin pattern in the RR spectrum upon H2O2 treatment is consistent with the formation of a FeIV=O species, which is more stable in the presence of bound l-Trp. RR spectroscopy provided further direct evidence for the formation of a FeIV=O species upon treatment of rIDO with H2O2 in the presence of l-Trp. Measuring the RR spectra in the low wavenumber region for H216O2- and H218O2-treated rIDO in the presence of l-Trp, we noted that the difference between the two spectra revealed an oxygen isotope-sensitive band peak at ~799 cm−1 (Fig. 7C), a band recently assigned as the FeIV=O2− stretching mode of a compound II-type species in rIDO (18, 26).
A consequence of compound I-mediated protein oxidation in the pseudo-peroxidases myoglobin and hemoglobin is irreversible damage or loss of protein heme (37). We therefore investigated the extent to which H2O2-treated rIDO can be reverted back to the original native, FeIII-IDO form using UV-visible absorption spectrophotometry. In the absence of l-Trp, the initial spectral changes induced by H2O2 treatment (20:1 mol/mol) partially reverted to the FeIII-rIDO spectrum with further incubation time, i.e. the γ-Soret band maximum of 407 nm measured within 1 min of H2O2 addition reverted to 406.5 nm over the next 10 min, albeit with reduced intensity (Fig. 8A). Subsequent addition of catalase (to remove unreacted H2O2) and ascorbate (to reduce residual compound II) resulted in the γ-Soret band maximum shifting further to 405 nm but still with reduced intensity (Fig. 8A). The α/β band region showed a similar partial restoration of the FeIII-rIDO heme spectrum (Fig. 8B). These findings are consistent with H2O2 promoting some irreversible heme change and/or loss in rIDO. In contrast to treatment of rIDO with H2O2 alone, addition of catalase and ascorbate to rIDO pretreated with H2O2 for 10 min in the presence of l-Trp resulted in a significantly better recovery of the native FeIII-rIDO heme spectrum (Fig. 8, C and D), indicating the protective action of l-Trp against irreversible heme changes. Prior addition of ascorbate, an efficient substrate for compounds I and II (31), completely blocked the permanent changes to the optical absorption spectrum of rIDO heme induced by H2O2 (Fig. 8, E and F). These observations with ascorbate are indicative of the rapid turnover of the enzyme via its peroxidase cycle back to its native FeIII form and the prevention of irreversible heme changes.
The rate of reduction of the FeIV=O center in compound II by H2O2 or reducing substrates is considered the rate-limiting step for peroxidase turnover (29). Notably, the optical spectroscopy data showed that the IDO compound II species formed in the presence of l-Trp is detectable over the ensuing 10 min after H2O2 addition (Fig. 8, C and D), consistent with it representing the predominant heme species present during the steady-state turnover of IDO peroxidase activity in the presence of l-Trp (Figs. 2C and and44E). This observation, together with our data showing that l-Trp inhibits rIDO-mediated H2O2 consumption (Fig. 2C), indicates that l-Trp is a poor reductant of IDO compound II, thereby inhibiting peroxidase turnover. In contrast, ascorbate promoted rIDO-mediated H2O2 consumption (Fig. 2D), indicating that ascorbate reduces IDO compound II to enhance turnover of the IDO peroxidase cycle. Consistent with this, in rIDO pretreated with H2O2 in the presence of l-Trp to first form IDO compound II, co-addition of catalase (to remove unreacted H2O2) and ascorbate (Fig. 9, C and D) enhanced the rate of conversion of this heme species back to FeIII-IDO in the presence of l-Trp, compared with removal of H2O2 with catalase alone (Fig. 9, A and B).
Our data implicate compound I-mediated IDO protein oxidation as a key mechanism underlying the oxidant's ability to inhibit dioxygenase activity. We next performed CD spectroscopy on rIDO with and without H2O2 treatment to determine the implications of protein oxidation for rIDO protein secondary structure. The α-helix content of human rIDO is ~60% (48, 57). Treatment of FeIII-rIDO with H2O2 dose-dependently decreased the α-helix content of the protein (Fig. 10A). Quantitative modeling of the CD spectrum indicated that the α-helix content of the protein decreased to ~40% or less following treatment with 5 or more mole excess of H2O2 (Fig. 10B). Protein structural changes induced by H2O2 were inhibited by l-Trp (Fig. 10A, inset) or cyanide (data not shown). Together, the immuno-spin trap and CD data indicate that the reaction of FeIII-IDO with H2O2 in the absence of substrates leads to compound I-mediated protein oxidation and alteration of IDO protein secondary structure.
Next, we measured the extent to which H2O2-induced changes in IDO secondary structure also involved an alteration to the distal heme pocket environment where l-Trp binds. For this, we measured the RR spectra of CO complexes of rIDO- and H2O2-treated rIDO. Previous studies of a wide variety of CO-heme complexes established a firm correlation between the wavenumber of the Fe-CO stretch (νFe-CO) and the polarity/stericity of the distal heme pocket (17, 58). A decrease in νFe-CO is associated with a less polar and/or more open distal pocket. The CO ligand is therefore a sensitive “probe” of alterations to the distal heme pocket environment. We therefore prepared CO complexes of nontreated and H2O2-treated rIDO samples and measured νFe-CO to reveal changes to the distal heme pocket resulting from H2O2 treatment. In the absence of l-Trp, rIDO-CO exhibited a νFe-CO at 511 cm−1 and a weak Fe-CO bend (δFe-CO) at 580 cm−1 (Fig. 11A, top spectrum) (48, 59). Pretreatment of rIDO with a 10 mol excess of H2O2 for 10 min before forming the FeII-CO complex produced a significant downshift of νFe-CO to 498 cm−1 (Fig. 11A). In the presence of l-Trp, νFe-CO and δFe-CO for native rIDO appear at 538 and 592 cm−1, respectively (Fig. 11B, top spectrum) (48), which are significantly higher than the corresponding vibrations in the absence of l-Trp (Fig. 11A, top spectrum). The l-Trp-induced upshift is indicative of a strong interaction between the CO ligand and l-Trp in the distal pocket (17, 48, 59). When FeIII-rIDO in the presence of l-Trp was treated with H2O2 and the FeII-CO complex formed, there was no detectable difference for the Fe-CO vibrations compared with the native protein (Fig. 11B, center spectrum). However, when FeIII-rIDO was first treated with H2O2, l-Trp then added, and the CO complex subsequently formed, νFe-CO was significantly downshifted to 499 cm−1 (Fig. 11B, bottom spectrum). Together, these observations support that H2O2 treatment of rIDO causes the distal heme pocket to open up, which draws l-Trp and CO apart and weakens their mutual interaction.
Having noted significant changes to the active-site heme upon H2O2 addition, we next determined the extent to which oxidant pretreatment affected l-Trp binding affinity. Native rIDO exhibits two l-Trp-binding sites as follows: the first exhibits a binding constant in the low micromolar range and the second in the millimolar range (26, 48). In this study, the first binding constant was measured to be 4.17 ± 0.95 μm (mean ± S.E., n = 3) and the second binding constant 1.44 ± 0.13 mm (mean ± S.E., n = 3) (Fig. 12). Following rIDO treatment with 10:1 mol excess of H2O2, the l-Trp binding constants were 1.43 ± 0.21 μm (mean ± S.E., n = 3) and 1.44 ± 0.42 mm (mean ± S.E., n = 3) (Fig. 12). Thus, compared with the native enzyme, pretreatment with H2O2 slightly enhanced the subsequent binding of l-Trp to the first binding site of rIDO by ~3-fold with no change in binding to the second lower affinity site. This shows that the ability of H2O2 to inhibit IDO dioxygenase activity does not involve a decrease in the binding affinity of l-Trp.
The metabolism of NO and its oxidation product nitrite (NO2−) is increasingly recognized as key reactions catalyzed by heme peroxidases within inflammatory environments, e.g. compounds I and II of heme peroxidases react with and consume NO (33, 60) and convert NO2− into the nitrating species nitrogen dioxide radical (•NO2) that mediates the formation of 3-nitrotyrosine on proteins (61, 62). Using a NO-specific electrode, we show that rIDO dose-dependently accelerated the rate of NO consumption in the presence H2O2 (Fig. 13, A and B). The total amount of NO consumed over time was supra-stoichiometric with respect to the concentration of rIDO. Also, the rate of NO consumption was significantly inhibited by the addition of l-Trp (Fig. 13C). Addition of NO2− to rIDO in the presence H2O2 accelerated the consumption of the oxidant by the enzyme (Fig. 14A), and this correlated with the dose-dependent formation of 3-nitrotyrosine on rIDO (Fig. 14B), in a manner inhibited by cyanide or l-Trp (Fig. 14C). These data show that IDO peroxidase utilizes NO2− to catalyze •NO2 formation and subsequent nitration of its protein tyrosine residues and reveal a previously unrecognized NO oxidase activity of IDO, with both of these peroxidatic functions inhibited by l-Trp.
Previous studies report that H2O2 reacts with the heme of rIDO to form a FeIV=O species (26) or inhibit rIDO dioxygenase activity (63). However, the molecular mechanism(s) by which H2O2 inhibits dioxygenase activity and the potential physiological relevance of IDO's peroxidase function are unknown. This study shows for the first time that H2O2 inhibits cellular IDO dioxygenase activity at the post-translational level and that this process requires peroxidase-dependent oxidative changes to IDO heme and protein structure. Furthermore, we identify that H2O2 activates peroxidase-dependent NO consumption and IDO protein nitration, with these events also likely to be of physiological significance. Notably, we show that l-Trp plays a critical role in suppressing peroxidase-mediated dioxygenase inhibition, NO consumption, and protein nitration.
Studies with purified human rIDO establish that H2O2 inhibits dioxygenase activity by activating IDO's peroxidase activity that mediates self-inflicted oxidative events (Scheme 2). Thus, reaction of FeIII-rIDO with H2O2, at low oxidant to enzyme ratios (1:1–25:1 mol/mol), forms compound I (Scheme 2, reaction A) that in the absence of small molecule peroxidase substrates initiates IDO protein oxidation, which alters the protein secondary structure and consequently the distal heme pocket environment (Scheme 2, reaction B). These H2O2-induced structural changes lead to the nonproductive binding of l-Trp to the distal active-site pocket, resulting in the inhibition of IDO dioxygenase activity (Scheme 2, reaction B). When l-Trp is present at the time of H2O2 treatment, it is oxidized by IDO compound I to yield oxindolylalanine (Scheme 2, reaction C) and kynurenine (Scheme 2, reaction D) via O2-independent and O2-dependent reactions, respectively. The two-electron oxidation of l-Trp by compound I results in the formation of oxindolylalanine and FeIII-IDO (Scheme 2, reaction C), whereas the one-electron oxidation yields a tryptophanyl radical and IDO compound II (Scheme 2, reaction D). The tryptophanyl radical may then react with O2 to form a peroxyl radical and, eventually, kynurenine (Scheme 2, reaction D). The reduction of compound II is suppressed by l-Trp, resulting in the attenuation of turnover of IDO's peroxidase cycle (Scheme 2, reactions E and F). These events correlate with protection of IDO against protein oxidation, prevention of perturbations in IDO protein secondary structure, and preservation of dioxygenase activity by l-Trp.
Our immuno-spin trapping (Fig. 3) and CD spectroscopy (Fig. 10) studies show that H2O2 treatment induces compound I-mediated IDO protein oxidation and a significant loss of the enzyme's α-helix content in a manner inhibited by a heme ligand (cyanide), a compound I substrate and radical scavenger (tempol) or l-Trp. Substrate binding data (Fig. 12) further indicate that the perturbation of IDO protein secondary structure does not limit l-Trp binding, verifying that dioxygenase inhibition does not involve a decrease in substrate binding. As dioxygenase activity is inhibited under these conditions, the maintenance of l-Trp binding is interpreted as being nonproductive. Consistent with this, our data show that the H2O2-induced protein structural changes also alter the substrate binding pocket such that l-Trp binds nonproductively. Thus, the RR spectroscopy data for FeII-CO complexes (Fig. 11) show that νFe-CO for H2O2-treated rIDO has almost the same wavenumber value with or without bound l-Trp (i.e. 498–499 cm−1). This contrasts the situation for native rIDO where νFe-CO is significantly higher when l-Trp is bound (538 cm−1) compared with l-Trp's absence (511 cm−1), indicating a strong interaction between l-Trp and CO within the distal pocket (17). These findings show that the reaction of rIDO with H2O2 prior to l-Trp addition causes a widening or opening of the distal pocket that draws l-Trp and CO apart and weakens their mutual interaction. A more open, less sterically hindered distal pocket is envisioned to allow l-Trp to bind less specifically and hence nonproductively with respect to dioxygenase activity.
The mechanism by which H2O2 inhibits IDO dioxygenase activity is similar to the inhibitory action we recently described for the selenazole ebselen (48). That study showed that covalent modification of the IDO protein via attachment of ebselen to IDO cysteine residues perturbs the protein secondary structure and the distal heme pocket leading to nonproductive l-Trp binding. The previous study using ebselen (48) and this study on H2O2 support a novel paradigm for the inhibition of IDO dioxygenase activity whereby covalent or oxidative modification of specific redox-active amino acid residues perturbs IDO's protein conformation that in turn promotes nonproductive substrate binding within the more open distal pocket of the active site.
Previous crystallography and site-directed mutagenesis studies of human IDO (8, 64) showed that no distal amino acid group in IDO is catalytically essential for interacting with and activating O2 or l-Trp in the enzyme's distal heme pocket. Instead, reactant proximity is of central importance for the dioxygenase activity of IDO, whereby l-Trp and O2 are locked into a specific orientation within the distal pocket, poised for reaction with the heme-iron (17). In particular, three distal residues, Phe-226, Phe-227, and Arg-231, appear to govern l-Trp binding and orientation within the distal pocket (8). This study using H2O2, together with our previous study using ebselen (48), suggests that protein conformational changes brought about by oxidative or covalent modification of other amino acids in IDO affect the spatial orientation of one or more of these critical distal pocket residues. This allows l-Trp to bind in an orientation that is not conducive to reaction with the heme-bound O2 necessary for dioxygenase activity.
Our studies show for the first time that IDO is akin to the pseudo-peroxidases hemoglobin and myoglobin in that exposure of FeIII-IDO to H2O2 in the absence of substrates leads to the formation of protein-centered radicals. A feature of peroxidase-mediated protein oxidation is intra- and inter-molecular long range electron transfer reactions via electron tunneling in the amino acid framework (65). Our immuno-spin trapping studies indicate that H2O2 induced a dose-dependent increase in formation of DMPO-adducts, suggesting that multiple amino acids in IDO form protein-centered radicals (Fig. 3). Protein radical formation also correlated with the appearance of higher molecular weight IDO oligomers indicative of inter-molecular transfer reactions between amino acids of different protein molecules (Fig. 3A). We propose that peroxidase activation and intra- and inter-molecular electron transfer reactions are therefore a feature of IDO and that these reactions have significant implications for H2O2-induced changes to IDO protein conformation, distal heme pocket environment, and dioxygenase activity. Protein-centered radicals in heme proteins primarily form on tyrosine, cysteine, and tryptophan residues. IDO is rich in these reactive amino acids as it contains 12 tyrosine, 8 cysteine, and 6 tryptophan residues. Detailed studies are underway to identify the amino acids within IDO that form protein-centered radicals and to define the intra- and inter-molecular electron transfer pathway from the heme to the different reactive amino acids in IDO.
Although our immuno-spin trapping data show the formation of protein-centered radicals on IDO, the spin trap did not prevent IDO cross-linking that yields high molecular weight oligomers (Fig. 3A). This suggests that DMPO does not scavenge the amino acid radicals directly responsible for IDO cross-linking and/or the yield of DMPO-adducts is low. This low yield does not necessarily mean, however, that few residues in IDO form radicals upon peroxidase activation. It may instead reflect the inefficiency of DMPO to trap radicals formed on certain IDO amino acids before they rapidly react with adjacent residues within the IDO protein framework.
Although previous studies suggest that IDO can exhibit a peroxidase activity (24, 25) and a recent study employed H2O2 as a surrogate to form a compound II species (26), the possibility that IDO forms compound I and its implications for dioxygenase activity have not previously been addressed. This is likely due to the high reactivity of compound I making direct detection difficult, particularly in pseudo-peroxidases where the porphyrin radical rapidly transfers to adjacent amino acids to form protein-centered radicals, e.g. direct detection of compound I in myoglobin requires replacement of a reactive distal His with Asp, which permits accumulation of compound I upon H2O2 addition (66). Although IDO compound I has yet to be directly detected, our immuno-spin trapping studies provide strong evidence that IDO is similar to myoglobin in that reaction of the FeIII enzyme with H2O2 results in compound I-mediated formation of protein-centered radicals (Fig. 3). Thus, H2O2 induces protein-centered radical formation on IDO in a manner sensitive to cyanide, which inhibits compound I formation, or tempol, which efficiently inhibits compound I-mediated oxidative reactions by acting as a competitive substrate for this reactive heme species (67, 68). Similarly, l-Trp, ascorbate, and tyrosine are all excellent compound I substrates (31, 32, 53, 69), which correlate with their ability to inhibit H2O2-induced IDO protein oxidation and protect against inactivation of IDO dioxygenase by H2O2. Although these agents can protect IDO by scavenging compound I, it is also plausible that they scavenge and repair protein radicals formed subsequent to the reaction of IDO amino acids with compound I. A recent study by Kuo and Mauk (70) describing an indole peroxygenase activity of IDO is further evidence for the formation of IDO compound I upon addition of H2O2.
A key finding of our study is that low physiological levels of l-Trp (50–100 μm) protect both cellular and purified IDO against H2O2-induced dioxygenase inactivation and that this protective action of l-Trp relates to its ability to react with compound I and inhibit IDO protein oxidation (Scheme 2, reactions C and D). Consistent with this, we show that l-Trp is oxidized by rIDO upon H2O2 addition in a manner prevented by cyanide and inhibited by the radical scavengers and compound I substrates, tempol or ascorbate (Fig. 4D).
Further evidence for the reaction of l-Trp with compound I is that l-Trp oxidation coincides with the formation of the FeIV=O species of compound II. Thus, exposure of rIDO to low H2O2 levels (5–25:1 mol eq) in the presence of l-Trp formed an IDO species with a γ-Soret peak at 412 nm and visible region peaks at ~542 and ~576 nm (Figs. 6, ,8,8, and and9).9). These wavelength changes are analogous to those recently reported by Lu and Yeh (26) after exposure of rIDO to high H2O2 concentrations (~800–900:1 mol eq) in the presence of l-Trp to form the FeIV=O species of IDO compound II. The detection of an FeIV=O stretch at ~800 cm−1 in the RR spectrum of H2O2-treated rIDO in the presence of l-Trp in this study (Fig. 7) and previously (26) is further evidence for formation of an FeIV=O heme in IDO. Together, our data show that as for other heme enzymes (32, 52), l-Trp is a reactive substrate for IDO compound I but not compound II.
Employing stop-flow spectroscopy, Lu and Yeh (26) also studied the rapid (ms) formation of compound II when rIDO was exposed to high H2O2 (~800:1 mol eq) in the absence of l-Trp. This species exhibited a γ-Soret peak at 415 nm, visible region peaks at 547 and 587 nm, and decreases at 499 and 633 nm. We show that exposure of FeIII-rIDO to low H2O2 (1–25:1 mol eq) induced a dose-dependent red shift of the γ-Soret band from 404 to 407 nm, visible region peaks at ~538 and ~576 nm, and decreases at 500 and 632 nm (Fig. 6). We envisage this spectrum measured ~30 s after H2O2 reflects a lower steady-state level of IDO compound II as it is less stable in the absence of l-Trp.
Reduction of compound II is considered the rate-limiting step in the catalytic cycle of heme peroxidases (29). We show this is also the case for IDO. Thus, ascorbate, an efficient compound II reductant (31, 53), enhanced rIDO-mediated H2O2 consumption (Fig. 2D) and promoted the rate of reduction of the FeIV=O heme (formed in rIDO pre-treated with H2O2 in the presence of l-Trp) back to FeIII-IDO (Fig. 9). In contrast, l-Trp, as a poor compound II substrate (32, 52), inhibited H2O2 consumption by rIDO (Fig. 2) that coincided with the detection of IDO compound II as the major steady-state heme species detected in the presence of l-Trp and H2O2 (Figs. 6, ,8,8, and and9).9). The stabilizing action of l-Trp on IDO compound II likely reflects its poor reactivity with this species and the ability of l-Trp to bind to the distal pocket and retard the reaction of H2O2 or ascorbate (Fig. 2D) and tyrosine (Fig. 2E) with the FeIV=O center (Scheme 2, reaction E). This inhibitory action of l-Trp toward peroxidase substrates appears unique for IDO in keeping with the ability of l-Trp to bind in close proximity to the heme iron of IDO, e.g. l-Trp rapidly reacts with compound I of myeloperoxidase to form compound II, but it does not retard the reduction of compound II of myeloperoxidase by tyrosine or ascorbate (32).
Although l-Trp retards the turnover of the IDO peroxidase cycle, the enzyme still consumes H2O2 in the presence of the amino acid (Figs. 2C and and44E). The oxidation of l-Trp exhibited a reaction stoichiometry of ~1:1 between H2O2 and l-Trp within the initial 2 min of reaction. Further reaction favored the consumption of H2O2 over l-Trp by a ratio of ~2:1 (Fig. 4E). These findings show that sufficient oxidizing equivalents are available to account for the l-Trp oxidation observed and support that ongoing l-Trp oxidation requires a second molecule of H2O2 to reduce the FeIV=O species of IDO compound II to form the native FeIII enzyme (Scheme 2, reaction F), which is then available to react with H2O2 to form compound I capable of oxidizing another molecule of l-Trp. Although O2 is formed upon reduction of compound II by H2O2, addition of superoxide dismutase did not affect l-Trp oxidation by rIDO ruling out a role for this reactive oxygen species.
The optical absorption spectroscopy data raise the possibility that treatment of rIDO with H2O2 at >10:1 mol excess may also induce irreversible heme changes and/or loss. Thus, although addition of catalase and ascorbate to rIDO previously exposed to H2O2 at 20:1 mol eq for 10 min recovered a γ-Soret band with a maximum at ~405 nm, the band showed a reduction in intensity of ~25% (Fig. 8A). At this concentration, H2O2 inhibited rIDO activity by ~90% suggesting that a putative loss of heme may account for up to 25% of the inhibition of dioxygenase activity afforded by this dose of H2O2. Loss of heme was not apparent at lower H2O2 doses (≤10:1) as judged by RR spectroscopy, despite >75% loss of dioxygenase activity. Hence, although H2O2 treatment at higher doses may cause heme damage, the majority of loss of IDO dioxygenase activity relates to protein oxidation and alteration of the distal pocket. Notably, our data support that peroxidase-mediated protein oxidation is linked to heme loss. Thus, l-Trp or ascorbate both protect against IDO protein oxidation (Fig. 3) and irreversible heme changes (Fig. 8).
Our new finding that IDO oxidizes l-Trp in the presence of H2O2 is in apparent contrast with a study by Lu and Yeh (26) reporting no observable l-Trp oxidation when IDO was treated with H2O2. However, this study did not provide data or experimental protocols from which their conclusion was derived. We speculate this discrepancy may reflect differences in reaction conditions and/or relative extent of l-Trp oxidation, e.g. Lu and Yeh (26) frequently exposed rIDO to high levels of H2O2 (>600:1 mol/mol) and l-Trp (>1000:1 mol/mol), conditions that may favor compound II formation with which l-Trp does not react. This apparent discrepancy may also reflect that whereas l-Trp is oxidized by IDO in the presence of H2O2, it is only to a minor extent, especially compared with indole as a substrate for IDO in the presence of H2O2 (70). Thus, Kuo and Mauk (70) recently reported that 0.5 μm rIDO in the presence of excess indole and H2O2 oxidized ~222 μm indole within 5 min of reaction. In comparison, we find that with 10-fold more rIDO (i.e. 5 μm) that >10-fold less substrate is oxidized (i.e. ~15 μm of l-Trp) after 5 min (Fig. 4A). This indicates that compared with l-Trp, indole is oxidized more efficiently by rIDO in the presence of H2O2 by >100-fold. This marked difference may relate to the ability of l-Trp, but not indole, to bind to the FeIV=O species of IDO (26) to slow its reduction as verified by our H2O2 consumption data (Fig. 2C). Although l-Trp oxidation is minor, it is the amino acid's ability to react with compound I, together with its ability to retard IDO compound II reduction, that underlies the protective actions of l-Trp against compound I-mediated IDO protein oxidation.
The oxidation of l-Trp is complex with different one- and two-electron oxidants yielding an array of oxidation products, including the tryptophanyl radical, hydroxytryptophans, oxindolylalanine, dioxindolylalanine, N-formylkynurenine, kynurenine, and hydroxylated kynurenines, with these products commonly accounting for a portion of the total amount of l-Trp oxidized (71–75). Heme enzymes are catalytically diverse and depending on the nature and availability of co-factors and substrates can catalyze different oxidative reactions. Thus, in addition to its dioxygenase activity, IDO in the presence of H2O2 and relevant substrates can exhibit heme peroxidase (this study and see Ref. 26) or the recently described peroxygenase (70) activities, with both enzyme reactions characterized by the formation of compound I as an active intermediate. This study shows for the first time that in the presence of low mole equivalents of H2O2 (<100:1) and l-Trp (10:1–100:1), IDO oxidizes the amino acid in a manner that requires the formation of a reactive heme intermediate, namely compound I. It is conceivable, however, that the compound I-mediated l-Trp oxidation by IDO to form oxindolylalanine and kynurenine involves peroxidatic and peroxygenase mechanisms, which represent O2-dependent and O2-independent reactions, respectively, e.g. the formation of oxindoles and kynurenines during l-Trp oxidation by horseradish peroxidase is inhibited under O2-deprived conditions (71), whereas the ability of IDO peroxygenase to oxidize indole into oxindoles and analogues of N-formylkynurenine is O2-independent (70).
In this study, we observed that although the formation of kynurenine was partially inhibited (by ~60%) under O2-depleted conditions, formation of oxindolylalanine was unaffected (Fig. 4F), indicating that the reaction mechanisms by which compound I in IDO oxidizes l-Trp are complex. The observation that oxindolylalanine formation is O2-independent is consistent with a role for IDO's peroxygenase activity (70), in which the two-electron reduction of compound I is tightly coupled to the transfer of the ferryl oxygen into a substrate, in this case, the insertion of the oxygen atom into the indole ring of l-Trp to form oxindolylalanine (Scheme 2, reaction C). As indicated above, the efficiency of this putative peroxygenase activity with l-Trp as a substrate is markedly less when compared with indole as a substrate (70). Our spectroscopic data indicate that in the presence of H2O2 and l-Trp, the majority of IDO compound I undergoes a one-electron reduction to form compound II. However, it is plausible that a minor portion of IDO compound I rapidly oxidizes l-Trp via its peroxygenase activity to form oxindolylalanine and FeIII-IDO, the latter available to react with H2O2 to readily re-form compound I (Scheme 2, reaction A).
Although the oxidation of indole by IDO peroxygenase yields oxindoles as the major product, analogues of N-formylkynurenine are also formed (70). As such, it is possible that the O2-insensitive portion of kynurenine formed by IDO in the presence of H2O2 and l-Trp is derived, in part, from the putative peroxygenase activity. However, our finding that a majority of kynurenine formation is sensitive to O2 depletion is consistent with a peroxidatic mechanism of l-Trp oxidation (Scheme 2, reaction D). Based on previous studies, it is proposed that l-Trp initially reacts with the porphyrin radical of compound I (32, 69) to produce a neutral tryptophanyl radical, which may form after rapid deprotonation of a tryptophanyl radical cation (75, 76). It is also plausible that the initial reaction involves the reaction of l-Trp with an IDO protein radical (77). Upon formation, the neutral tryptophanyl radical is known to react with O2 to form a peroxyl radical (75, 77, 78), which upon forming a hydroperoxide can rearrange via yet to be characterized chemistry into N-formylkynurenine that hydrolyzes into kynurenine (79, 80). Further evidence for compound I-mediated oxidation of l-Trp is the observation that tempol and ascorbate both inhibited H2O2-induced oxidation of the amino acid by IDO (Fig. 4D), which may reflect their ability to scavenge compound I, an IDO protein radical and/or a free tryptophanyl radical.
Although we have proposed some plausible reaction pathways responsible for the oxidation of l-Trp into oxindolylalanine and kynurenine, these two products account for ~30–45% of the total amount of l-Trp oxidized by the enzyme in the presence of H2O2. More detailed studies are therefore required to unravel the potentially complex array of reactions by which IDO in the presence of H2O2 oxidizes l-Trp and to identify the oxidation products formed in addition to oxindolylalanine and kynurenine. Irrespective of this, our data show that the ability of the amino acid to act as a competitive substrate for IDO compound I to divert this species from mediating protein oxidation is a key mechanism by which it protects against H2O2-mediated dioxygenase inactivation.
In addition to IDO, the oxidative cleavage of the pyrrole ring of l-Trp is catalyzed by TDO. Despite catalyzing identical dioxygenase reactions and showing similarities in their heme active sites (8–10), IDO and TDO differ in many ways, including the implications of H2O2 for their dioxygenase activity. Thus, although our study on IDO supports that compound I-mediated protein oxidation underscores the inhibitory action of H2O2 on dioxygenase activity, a recent study by Fu et al. (81) indicates that compound I-mediated formation of a protein-based radical is an intermediary step by which H2O2 reactivates TDO dioxygenase activity. Fu et al. (81) proposed that TDO reactivation by H2O2 involves a two-electron reduction of a FeIV=O intermediate by l-Trp to form a FeII-Trp adduct indicated by the formation of a distinct peak at 432 nm by optical absorption spectroscopy (81). Also, for TDO it was not possible to detect a FeIV=O species in the presence of H2O2 and l-Trp (81), which has been recently reported to exhibit a γ-Soret maxima at 414 nm and peaks in the visible region with reduced intensity at 504 and 630 nm, and increased peak intensities at 526 and 548 nm (82). In marked contrast, our optical absorption and RR spectroscopy data on IDO show that H2O2 in the presence of l-Trp forms an FeIV=O species as the major detectable IDO heme species, which exhibits a γ-Soret peak at 412 nm and visible region peaks at ~541 and ~576 nm (Figs. 6, ,8,8, and and9).9). We found no evidence for the formation of FeII-IDO that, if present, shows a γ-Soret peak at 426–428 nm and visible peaks at ~530 and 557/558 nm (83) and in the RR spectrum a ν4 peak at ~1355 cm−1 (17). Therefore, unlike TDO where l-Trp facilitates dioxygenase reactivation by H2O2 (81), l-Trp is oxidized by IDO compound I and retards the reduction of the FeIV=O heme of IDO compound II, resulting in protection of IDO from H2O2-induced protein oxidation and dioxygenase inactivation. Fu et al. (81) also reported that ~25% of the H2O2 metabolized by TDO is accounted for by the production of O2 and proposed that this reflects the enzyme's catalase activity, i.e. the two-electron reduction of TDO compound I to yield FeIII-TDO and O2 (81), a mechanism characteristic for “true” catalases (54). In contrast, we show that despite consuming H2O2 (Fig. 2), rIDO does not show a net increase in O2 levels (Fig. 2F). Our study indicates that H2O2 metabolism by IDO primarily proceeds via its peroxidase cycle, with the reaction of H2O2 with compound II the rate-limiting step in H2O2 consumption by IDO (Scheme 2, reaction F). Together, the above highlight distinct differences between IDO and TDO with respect to their reactions with H2O2.
Although the substrates for IDO as a dioxygenase are limited to indoleamines, we show that peroxidase activation broadens the substrates with which IDO reacts, including NO and NO2−. For NO2−, we show that IDO's peroxidase activity promotes self-nitration of protein tyrosines (Fig. 14). Thus, IDO is akin to other heme peroxidases (myeloperoxidase (84), horseradish peroxidase (85) and myoglobin (86)), in that reaction of NO2− with compound I forms •NO2, which reacts with protein-tyrosyl radicals to form 3-nitrotyrosine (34). IDO's peroxidase function also catalytically consumes NO (Fig. 13), similar to horseradish peroxidase (87), myeloperoxidase (33), or myoglobin (88), that scavenge NO via compounds I and II (60). Notably, l-Trp inhibits IDO tyrosine nitration and NO consumption supporting that l-Trp is a preferred substrate for IDO compound I compared with NO2− or NO and also retards the reaction of NO2− or NO with the FeIV=O center of IDO compound II.
It is likely that the findings of this study are of physiological relevance. IDO is a key immune control enzyme capable of promoting immune tolerance and suppression under various physiological and pathological conditions (3, 6, 7). Activation of IDO's peroxidase activity is likely to have important implications for the immune control and other biological functions (e.g. NO oxidase) when the enzyme is expressed in inflammatory cells and tissues where the local levels of IDO, NO, and H2O2 are elevated and levels of l-Trp and ascorbate can become limiting, e.g. local IDO-mediated depletion of l-Trp may increase the susceptibility of the enzyme to peroxidase-mediated self-inactivation. Such oxidative inactivation of a key immunosuppressive protein such as IDO may represent a pathological mechanism by which oxidative stress enhances T cell-mediated chronic inflammation.
*This work was supported in part by Australian National Health and Medical Research Council Project Grant 568774 (to S. R. T.), American Chemical Society Petroleum Research Fund Grant 46549-G4 (to A. C. T.), and Australian Research Council Discovery Grant DP0878559 (to P. K. W.).
4The abbreviations used are: