Isotope effects and alternative substrate reactivities for tryptophan 2,3-dioxygenase.

Tryptophan 2,3-dioxygenase (EC 1.13.1.12) is a hemoprotein which catalyzes the first step in the oxidative degradation of tryptophan. The reaction is believed to proceed by addition of O2 across the 2,3-bond of the indole ring, followed by decomposition of the resultant dioxetane to give N-formylkynurenine. A primary D2O isotope effect of 4.4 on Vmax/Km was observed at the pH optimum, pH 7.0. This implies that abstraction of the indole proton is at least partially rate-determining. An inverse secondary isotope effect of 0.96 was observed for L-[2-3H]tryptophan at this pH. The secondary isotope effect signals the formation of the C-O bond at C-2. As the rate of proton abstraction increased with increasing pH, the D2O isotope effect decreased to 1.2 at pH 8.5 and the secondary isotope effect increased to 0.92. The rate-determining steps therefore change with increasing pH, and bond formation at C-2 becomes more rate-limiting. The secondary isotope effect did not change significantly with varying O2 concentration so that substrate binding is primarily ordered with O2 binding first. The specificity of the enzyme towards substituted tryptophans shows that substitution of the phenyl ring of the indole is sterically unfavorable. Steric hindrance is highest at the 4- and 7-positions, while the 5- and 6-positions are less sensitive. 6-Fluoro-L-tryptophan was more reactive than tryptophan, and the increased reactivity can be explained by an electronic effect that enhances of the rate of C-O bond formation at C-2.


Isotope Effects and Alternative Substrate Reactivities for
Tryptophan 2,S-Dioxygenase" (Received for publication, October 21, 1992, and  Tryptophan 2,3-dioxygenase (EC 1.13.1.12) is a hemoprotein which catalyzes the first step in the oxidative degradation of tryptophan. The reaction is believed to proceed by addition of 0 2 across the 2,3-bond of the indole ring, followed by decomposition of the resultant dioxetane to give N-formylkynurenine.
A primary DzO isotope effect of 4.4 on V,aJKm was observed at the pH optimum, pH 7.0. This implies that abstraction of the indole proton is at least partially rate-determining. An inverse secondary isotope effect of 0.96 was observed for ~-[2-'H]tryptophan at this pH. The secondary isotope effect signals the formation of the C -0 bond at C-2. As the rate of proton abstraction increased with increasing pH, the DzO isotope effect decreased to 1.2 at pH 8.5 and the secondary isotope effect increased to 0.92. The rate-determining steps therefore change with increasing pH, and bond formation at C-2 becomes more rate-limiting. The secondary isotope effect did not change significantly with varying Oa concentration so that substrate binding is primarily ordered with O2 binding first. The specificity of the enzyme towards substituted tryptophans shows that substitution of the phenyl ring of the indole is sterically unfavorable. Steric hindrance is highest at the 4-and 7-positions, while the 5-and 6-positions are less sensitive. 6-Fluoro-~-tryptophan was more reactive than tryptophan, and the increased reactivity can be explained by an electronic effect that enhances of the rate of C -0 bond formation at C-2. Tryptophan 2,3-dioxygenase (EC 1.13.1.12) catalyzes the addition of oxygen across the 2,3-double bond of the indole ring, oxidatively cleaving it to N-formylkynurenine. As an iron-dependent dioxygenase, it is a member of the small class of enzymes that includes lipoxygenase and cyclooxygenase. Most studies of tryptophan 2,3-deoxygenase have concentrated on the enzyme from rat liver and to a greater extent on two related enzymes, Pseudomonas acidovorans tryptophan dioxygenase and indoleamine dioxygenase, a less specific mammalian enzyme that is induced by interferon-y .
Rat liver tryptophan dioxygenase is tetrameric with 4 identical subunits (Maezono et al., 1990). Purified enzyme retains 2 mol of noncovalently bound heme per tetramer, and exogenous heme stimulates activity, suggesting a stoichiometry of one heme/subunit . Tryptophan dioxygenase has allosteric as well as catalytic binding sites * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "duertkement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 6137; Fax: 919-990-6147.
3 TO whom correspondence should be addressed. Tel.: 919-990-for tryptophan at neutral pH . a-Methyl-DL-tryptophan is not a substrate but is an allosteric effector. When the pH is increased to 8.0, a conformational change is observed by ultracentrifugation, and allosteric kinetics are no longer observed.
Equilibrium binding studies with Pseudomonas tryptophan dioxygenase and indoleamine dioxygenase (Sono, 1986;Sono et al., 1980;Makino et al., 1980a) have provided evidence that the ferrous form of these enzymes is the active form, and the ternary complex of enzyme, Oz, and tryptophan is catalytically competent (Sono, 1989;Taniguchi et al., 1979;Ishimura et al., 1970). The reduction potential of the heme of rat liver tryptophan dioxygenase is relatively low, -0.11 V at pH 7 and -0.28 V at pH 8.4 (Makino et al., 1980b). The present investigation serves to extend characterizations of the enzymesubstrate complex by using alternative substrate reactivities and isotope effects to provide information about intermediates along the reaction pathway.

Materials-~-[2-'~C]Tryptophan was purchased from Research
Products Int. ~-[3'-'~C]Tryptophan was purchased from New England Nuclear Research Products. [2-3H]Tryptophan was synthesized by catalytic tritiation of the N-1 trifluoroacetyl methyl ester derivative of 2-bromotryptophan (Du Pont-New England Nuclear). The labeled protected tryptophan was resolved and deprotected with chymotrypsin and carboxypeptidase A. Synthesis of the protected 2bromotryptophan and deprotection were according to Phillips and Cohen (1986). Fluorotryptophan analogues were resolved by the same derivatization and enzymatic deprotection scheme. The resulting Lisomers were purified by HPLC' on a C18 column in 5 mM potassium phosphate, pH 4, using a gradient from 0 to 50% acetonitrile. Enantiomeric purity was confirmed by derivatization with Marfey's reagent (Marfey, 1984). Specificity of tritium labeling was demonstrated by preparing [2-'H]tryptophan in parallel and characterizing the product by deuterium NMR. Additionally, tritium was shown to be released quantitatively as formate after incubation of labeled substrate with tryptophan dioxygenase and hydrolysis of the N-formylkynurenine product (see below). 7-Fluoro-~-tryptophan was a generous gift from Dr. Robert Phillips, Athens, GA. Other substituted tryptophans and L-tryptophan were purchased from Sigma. N-(-9-Fluorenyl)methoxycarbonyl-alanine (Fmoc-Ala) was from Applied Biosystems, San Jose, CA. Sprague-Dawley rats were purchased from Charles River.
Purification of Tryptophan 2,3-Diorygenase"The procedure of  was adapted for the induction and purification of tryptophan dioxygenase from rat liver with minor modifications. Prednisolone sodium phosphate was used to induce the enzyme at a single dose of 50 mg/kg body weight, which was given intraperitoneally 5.5 h before excising livers. DEAE-cellulose was substituted for DEAE-Sephadex and the final gel electrophoresis step was omitted. Instead, DEAE-cellulose fractions were concentrated using Centriprep-10 protein concentrators (Amicon) and desalted on a G-25 column. The DEAE-cellulose-purified tryptophan dioxygenase was loaded onto a Mono Q column (Pharmacia LKB Biotechnology) The abbreviations used are: HPLC, high performance liquid chromatography; Fmoc, N-(9-fluorenyl)methoxycarbonyl. equilibrated with 20 mM HEPES, pH 7.0, containing 10 mM Ltryptophan. Activity was eluted with a linear gradient from O to 1.0 M KC1 in the equilibrating buffer. Purified tryptophan dioxygenase was stored at -80 "C in the presence of 10 mM L-tryptophan. This preparation had no contaminating kynurenine formamidase activity.
Routine radiometric assays were based on the fact that N-formylkynurenine is readily hydrolyzed to kynurenine and formic acid in the presence of 3% perchloric acid (Ozaki et al., 1986). The assay solution contained 4.9 D M L-tryptophan, 20 mM sodium formate, 2 mM sodium ascorbate, 0.46 p~ hematin, and ~-[2-"C]tryptophan (typically 400 dpm/nmol) in 0.1 M KHzP04, pH 7.0. The reaction was initiated by addition of enzyme. After 30 min at 37 'C, 100 pl of 6% perchloric acid was added and the samples were incubated an additional 30 min. 1.0 ml of 10% charcoal (w/v) suspended in 3% perchloric acid containing 1.0 M formic acid was added, and the samples were vortexed and then centrifuged for 5 min at 15,000 rpm to remove unreacted tryptophan. A 0.6-ml aliquot from each tube was counted to determine the extent of labeled formate formation.
HPLC analysis of reaction mixtures was performed on a C18 column using gradient elution with acetonitrile/water buffered with either 5 mM KH~POI, pH 4.0 , or 10 mM ammonium acetate, pH 7.0 (Takikawa et al., 1988). At pH 4, three cartridge C18 columns (4.6 X 33-mm each, 3-rm particles, Perkin-Elmer) were used in series at a flow rate of 1.5 ml/min. The columns were eluted for 5 min at 9% acetonitrile followed by a 10-min linear gradient to 32% acetonitrile. At pH 7, a 4.6 X 100-mm Brownlee ODS column (5-p particles, SCI-CON, Winter Park, FL) was eluted with a 10-min linear gradient from 0 to 5% acetonitrile, a 5-min linear gradient to 10% acetonitrile, and a 5-min linear gradient to 80% acetonitrile followed by 5 min at 80% acetonitrile. Quantitation was accomplished with an LC-235 diode array detector (Perkin-Elmer) connected to a model 171 radioisotope flow-through detector (Beckman Instruments, Inc., San Ramon, CA).

3H/14C
Isotope Effect Experiment-Tryptophan dioxygenase was desalted by centrifugation through a G-25 Sephadex column equilibrated in assay buffer containing 3 mM a-methyl-DL-tryptophan. The assay mixture contained 0.1 mM L-tryptophan, 3 mM a-methyl-DLtryptophan, 50 mM sodium formate, 4 mM sodium ascorbate, 0.92 g M hematin, 0.08 pCi of ~-[2-"C]tryptophan plus 0.25 pCi of L -[~-~H ] tryptophan per 100 pl. For assays at pH 7.0, a single reaction mixture was prepared in 0.1 M potassium phosphate. Aliquots were quenched by addition to 1% perchloric acid, and incubated for 1 h to hydrolyze formylkynurenine. Measurements at pH 8.4 were performed in 0.1 M Tris.HC1, at both 250 and 5 p~ Oz. 0, concentrations were maintained by saturation with air or 0.4% 0, in Nz, respectively. The reaction mixtures, including enzymes, were made up without ascorbate in glass vials sealed with two rubber septa. After the samples had been equilibrated to the desired 0, concentration, the reactions were started by addition of a small volume of ascorbate through a syringe. Reactions were quenched by freezing on dry ice. Samples for all conditions were analyzed by HPLC, and 1-min fractions were collected and counted for 20 min.
Tryptophan Analogues as Tryptophan Dioxygenase Substrates-Tryptophan analogues were assayed at a concentration of 1.8 mM in the presence of [2-"C]tryptophan, 0.92 p~ hematin, 2.0 mM sodium ascorbate, and 0.17 mg/ml Fmoc-alanine (the internal standard), with or without 3 mM a-methyl-DL-tryptophan. After incubation at 37 "C, the samples were frozen on dry ice for later analysis by HPLC.
Compounds were incubated with enzyme in triplicate and the amount of products formed was compared to a control without enzyme. HPLC analysis for experiments without a-methyl-DL-tryptophan were performed using a single Perkin-Elmer C18 cartridge with the pH 4 elution buffer and a modified elution profile: 8 min at 0% acetonitrile followed by a 10-min gradient to 24% acetonitrile, a 4-min gradient to 64% acetonitrile, and a 5-min gradient to 80% acetonitrile. When a-methyl-DL-tryptophan was included in the reaction mixture, a Brownlee C18 column was used at pH 4 with the following elution profile: 20 min at 0% acetonitrile followed by a 10-min linear gradient to 80% acetonitrile.
Reaction Rates in DZO-Buffers contained 100 mM potassium phosphate and 100 mM Tris in either HzO or D20 at pH 6.5, 7.0, 7.5, 8.0, and 8.4. pH meter readings in DZO were corrected for the known shift of 0.4 units from the true pD (Schowen and Schowen, 1982). 3 mM a-Methyl-DL-tryptophan was included in all the reaction mixtures in order to saturate allosteric interactions with the enzyme . The enzyme was concentrated and exchanged into buffers of the appropriate pH containing 50% HzO and 50% DzO by desalting on a P-10 column. Either 5 or 10 gl were added to reaction mixtures to give a final volume of 100 pl so that the isomeric purity of the water was >95% in the reaction mixtures. The rate of oxidation of L-tryptophan was determined using the charcoal precipitation assay described above. The rates of 5-fluoro-~-tryptophan and 6-fluoro-~-tryptophan oxidation relative to L-tryptophan were determined by HPLC analysis as described above.

RESULTS
Substrate Specificity-Reactivities of 4-, 5-, 6-, and 7-substituted tryptophans relative to tryptophan are given in Table   I. A number of other tryptophan analogues were tested but turned over at (1% of the rate for tryptophan: 7-aza-DLtryptophan, 5-hydroxy-cu-methyl-~~-tryptophan, 1-thio-DLtryptophan, 5-hydroxy-~~-tryptophan, and N-methyl-DLtryptophan. Relative V / K values for tryptophan analogues were measured directly by comparing the rate of turnover of each analogue relative to tryptophan when the two substrates were incubated together in the same reaction mixture (Abeles et al., 1960). Note that rates for L-isomers were determined in the presence of a-methyltryptophan, an allosteric effector. In all cases, substrates with methyl or bromo substituents were poorer substrates than tryptophan. This agrees with previous studies (Civen and Knox, 1960) and most probably reflects steric hindrance, since these substituents have only a small electronic effect on the pK. of tryptophan (Yagil, 1967). This effect is greatest at positions 4 and 7 and smallest at positions 5 and 6.
The 5 -, 6-, and 7-fluorotryptophans also showed a wide range of reactivities. In parallel with the methyl-substituted analogues, 7-fluorotryptophan was least reactive, and the 6fluoro analogue was most reactive. The general trend (7 < 5 < 6) follows that predicted by steric effects, even for a substituent as small as fluorine. 6-Fluoro-~-tryptophan was actually a better substrate than tryptophan, however. This may well represent an electronic effect of fluorine that is large enough to dominate steric effects at this position.
When comparing the reactivity of substrate analogues in a two-substrate reaction, the relative V,,,/K,,, values for one substrate versus its analogues can change when the concen-  tration of the second substrate is varied. Variations in relative V,,,,,/K, values depend on the order of substrate binding ("Appendix"). Accordingly, the K, for O2 was calculated by fitting oxygen electrode data for 0, consumption to the integrated Michaelis-Menten equation in order to establish a meaningful range for varying 0, concentration. The K, was 7.2 ~L M at pH 7.0 and 6.9 mM tryptophan. The relative V,,,/ K, values for 6-fluoro-~~-tryptophan compared to tryptophan were measured at 0, concentrations above and below the K, and were 2.14 +_ 0.06 at 5 FM 02, compared to 2.13 +-0.06 at 250 ~L M 02. These results are consistent with ordered binding with O2 binding first.
DZO Isotope Effects-The initial velocities for tryptophan oxidation by tryptophan dioxygenase are presented in Fig. 1 as a function of pH in H20 and DzO. Velocities were measured below the K, for tryptophan and reflect V,,,/K, values.
These reactions were all performed in the presence of 3 mM a-methyl-DL-tryptophan to minimize ambiguities due to allosteric effects . Rates of tryptophan oxidation in H20 were optimum at pH 7.0. The pH rate profile in D 2 0 differed substantially, however. The reaction in D20 was 5.25-fold slower than the reaction in H,O at low pH, pH 6.5. When the pH was increased, the D20 isotope effect decreased to only 1.19 at pH 8.4. The pH optimum in D20 was pH 8.4 or higher. The relative rates of turnover of 6fluoro-L-tryptophan versus \2-'4C]tryptophan were also determined over a range of pH values in both H,O and DzO (Fig.  2). The 6-fluor0 substituent effect did not change with pH in D20 and was small, approximately 1.5. In HzO, the substituent effect was similarly small at low pH but increased to 2.8 at pH 8. Changes in the substituent effect will reflect changes in rate-limiting steps with pH or solvent, and steps sensitive to 6-fluor0 substitution apparently predominate at high pH in H20. The data for 6-fluorotryptophan are also plotted as absolute rates in the inset for Fig. 2. The substrate analogue shows a solvent isotope that is comparable in magnitude to that for tryptophan and is similarly sensitive to pH, although the maximum isotope effect persists up to pH 7.5 for this substrate.
3H Isotope Effects-The 3H isotope effect on V,,,,,/K, for   3(V/K) is the rate of turnover of [2-14C]tryptophan relative to 12-"Hltryptophan and is the isotope effect on V,,,aJK,,,. Standard errors are shown and the number of determinations is given in parentheses. The isotope effects were measured in the range from 9 to 52% conversion and were independent of the extent of reaction in this range.
[2-3H]tryptophan was determined in competition with ~-[ 2 -'%]tryptophan (Table 11). In the initial experiment at pH 7.0, the observed isotope effect was 0.96 & 0.01. At pH 8.4, the isotope effect was significantly larger, 0.92 & 0.01. The variation of isotope effects with substrate concentration can provide information about order of substrate binding if isotope effects below and above the substrate K, are compared (Klinman et al., 1980). The 3H isotope effect for tryptophan dioxygenase did not change significantly when measured at 5 PM O2 compared to a normal 0, concentration of 250 pM, again indicating ordered addition with 0, binding first.

DISCUSSION
Based on solvent and substrate isotope effects, alternative substrate reactivities, and pH rate profiles, the mechanism of Fig. 3 is proposed for tryptophan 2,3-dioxygenase. It is assumed that oxidative cleavage of the indole ring procedes via formation and decomposition of a dioxetane intermediate. Our results provide information regarding the steps leading to dioxetane formation as outlined below.
Large D20 isotope effects were observed for tryptophan at low pH: 5.2 at pH 6.5 and 4.4 at pH 7.0 (Fig. 1). A similarly large isotope effect of 6.1 was observed for 6-fluorotryptophan at pH 7.5 (Fig. 2, inset). There are numerous possible sources of solvent isotope effects (Klinman, 1978;Schowen and FIG. 3. The proposed reaction mechanism for the oxidation of tryptophan by rat liver tryptophan 2,3-dioxygenase. The reaction is shown to proceed by abstraction of the indole proton, addition of oxygen to carbon 3 of the indole ring, and cyclization of oxygen to the second carbon of the indole ring to form the key dioxetane intermediate. Species enclosed in brackets are thermodynamically allowed but may not be discrete intermediates. The numbering convention for the indole ring is provided for convenience. Schowen, 1982). It is common, for example, that pK,, values will shift to approximately 0.5 units higher in D20, and this may be observed as a shift by as much as 0.5 units to higher pH in enzyme pH rate profiles. The solvent isotope effect for tryptophan dioxygenase, however, does not appear to be related to a simple shift of pH rate profiles. Thus, a shift of the curve for tryptophan turnover in D20 by 0.5 units to lower pH would have no effect on the observed isotope effect at pH 7.0. In addition, the pH rate profile for 6-fluorotryptophan in D20 is very similar to that for tryptophan, while the pH optimium in H 2 0 is pH 7.0 for tryptophan and pH 8.0 for the fluoro derivative. The DzO pH rate profiles for the two substrates, therefore, do not appear to represent simple and equal shifts of the profiles in H20. In the absence of a major contribution by a simple shift in pK,, the size of the observed isotope effects is large enough to be considered a primary kinetic isotope effect for transfer of an exchangeable proton. Considering the nature of the tryptophan dioxygenase reaction, the most straightforward interpretation of the effect would be that it represents deprotonation of the indole nitrogen.
A secondary isotope effect was observed for ~-[Z-~H]-tryptophan. This isotope effect increased from 0.96 at pH 7.0 to 0.92 at pH 8.4. The secondary isotope effect is inverse and is indicative of a change in hybridization from sp2 to sp3 at position 2, i.e. carbon-oxygen bond formation at C-2. The magnitude of the secondary isotope effect at pH 8.4 is sufficiently large to indicate that bond formation is essentially complete in the rate-limiting step at high pH (Klinman, 1978). Since the 3H isotope effect increased with pH as the solvent isotope effect decreased, proton transfer and bond formation at C-2 must represent different steps in the reaction mechanism. Furthermore, the proton transfer step must precede C -0 bond formation. This is because secondary isotope effects will generally be largest if they precede the rate-determining step and will be masked only if they follow the slow step.
The fact that the secondary isotope effect decreased by only a factor of two between pH 8.4 and 7 indicates that C-0 bond formation at C-2 is still partially rate-limiting at the pH optimum. This in turn implies that proton abstraction from the indole nitrogen is only partially rate-limiting at pH 7, and the intrinsic solvent isotope effect may therefore be larger than that observed by as much as a factor of two.
Varying the 0, concentration from 5 to 250 PM did not significantly affect the secondary isotope effect, and this can be taken as indication of a predominantly ordered binding pattern with 0, binding first (Klinman et aL, 1980). We also found that the rate of turnover of 6-fluorotryptophan relative to tryptophan did not change with varying O2 concentration. The interpretation of the effect of O2 concentration on the relative rates of alternative substrate turnover is similar but not identical to that for competitive isotope effects. An analysis is provided in the "Appendix." The fact that O2 binds first indicates that 0, binds to heme and implies that the function of the heme is to localize and activate O2 rather than tryptophan. This order of binding is contrary to that proposed for the closely related enzymes indoleamine dioxygenase (Ishimura et al., 1970) and tryptophan dioxygenase of P. acidouoram (Koike and Feigelson, 1971) based on equilibrium binding studies. The reason for the discrepancy may be either that the related enzymes have different kinetic mechanisms or that the equilibrium results are misleading in their prediction of kinetics.
Free energy changes for selected steps of the tryptophan dioxygenase reaction are provided in Table I11 and allow us to both strengthen and extend the above conclusions. Tryptophan is a weak acid (Equation 1, Table 111) and we must consider whether it is thermodynamically feasible that the tryptophanyl anion is an intermediate in the reaction. The accessibility of high energy intermediates in enzyme reactions has been discussed by Jencks (1980) and recently by Gerlt and Gassman (1992). The barrier to proton abstraction from tryptophan by a base on the enzyme with pK. = 7 would be 13.6 kcal/mol. To this thermodynamic barrier must be added a kinetic barrier which should be no more than the 3 kcal/ mol barrier for diffusion-controlled transfer of a proton from a nitrogen base in solution. The total barrier of 16.6 kcal/mol is very close to the predicted allowable limit of 16.7 kcal/mol (Gerlt and Gassman, 1992) for the tryptophan dioxygenase reaction with kc, = 10 s-l . We  O2 is the corresponding nitrogen anion. The free energy is based on pK. = 24.5, calculated from the pK. of aniline (pK, = 27; Dolman and Stewart, 1967) with a correction for the electron-withdrawing effect of the dioxetane oxygens (Fox and Jencks, 1974). dDi~~ociation of oxygen from the heme of TDO calculated from Kd = 2 ~I M for the Trp-Pseudomonas TDO complex, consistent with the K,,, for oxygen for rat liver TDO (Ishimura et al., 1970;Sono et al., 1980). e The free energy of oxidation of Trp to the dioxetane, Trp. 02, in the gas phase was calculated from heats and entropies of formation (Benson, 1976;Richardson, 1989). The gas-phase free energies were corrected to aqueous solution using AM1-SM2 calculations (Cramer and Truhlar, 1992).
/The free energy for 1-electron oxidation of Trp was calculated from a reduction potential of 1.05 V versus NHE for Trp at pH 7 (DeFelippis et al., 1989). 1-Electron reduction of the heme-oxygen complex. Complexation of O2 to an iron-porphyrin complex raises the reduction potential of the 0 2 / O ; couple by 0.52 V (Sawyer and Valentine, 1981;Tsang and Sawyer, 1990).
Concerted addition of 0 2 to Trp to form the anion of the dioxetane (Equations 2 + 3 + 4). conclude that the tryptophanyl anion is a viable intermediate and note that it is probably not fortuitous that the observed rate of reaction is equal to that predicted from the thermodynamics. Iron-catalyzed formation of dioxetanes has been proposed to be a concerted singlet biradical reaction (Sheu e t aL, 1990), and we can address the question whether formation of the dioxetane in the tryptophan dioxygenase reaction proceeds by concerted addition of 0 2 across the 2,3-bond. Formation of the C -0 bond at C-2 occurs in a step subsequent to and discrete from proton transfer. Furthermore, if deprotonation of nitrogen is catalyzing addition of 02, then reprotonation must occur in a step subsequent to addition of O2 (Fig. 3).

1-Election oxidation of
The anilinium anion formed upon concerted addition of O2 would be a strong base with a pK, of approximately 24.5 (Equation 2, Table 111). The thermodynamic barrier to the overall reaction of deprotonation and addition of oxygen to form the anion of the dioxetane is calculated to be 27.7 kcal/ mol (Equation 7, Table 111). This is higher than the allowable limit of 16.7 kcal/mol, and the concerted pathway would be inaccessible by a barrier of at least 11 kcal/mol. In contrast, stepwise addition of O2 across the 2,3-bond would permit activation of tryptophan to oxidation by proton abstraction but would avoid formation of more unstable anionic intermediates. We therefore suggest that addition of O2 proceeds via the 3-peroxy intermediate (steps 1-3, Fig. 3).  in brackets to indicate that protonation of the imine may be a discrete step or may be concerted with C-0 bond formation. This mechanism parallels closely the chemistry of electrochemical and photosensitized oxidations of tryptophan (Nguyen et at., 1986;Nakagawa et al., 1977a, 197713) in terms of formation of a 3-peroxy intermediate and subsequent acidcatalyzed nucleophilic addition at (2-2. In the model reactions, however, the nucleophil that adds to C-2 is solvent or the aamino group of tryptophan, and the thermodynamically less favorable dioxetane formation is not observed. The reactivity of 6-fluorotryptophan is also consistent with the proposed mechanism. The 6-fluor0 analogue is a better substrate than tryptophan, and this is taken to represent an electronic effect. The effect of fluorine does not represent a simple effect on the pK, of tryptophan since the effect is suppressed in DzO, where proton transfer is most rate-limiting. The effect of fluorine is greatest at high pH where ring closure to form the dloxetane becomes more rate-limiting, and we propose that the effect of fluorine is to activate the intermediate imine to nucleophilic attack (step 5, Fig. 3).
Finally, we show C-0 bond formation at C-3 proceeding in steps of proton abstraction, 1-electron transfer, and bond formation. It is equally likely that C-0 bond formation is concerted with proton abstraction, and the proposed intermediates are enclosed in brackets to emphasize this possibility. Thermodynamic data, however, indicates that these intermediates would be allowed. The thermodynamic barrier calculated for the l-electron oxidation of tryptophan to the free radical by heme-bound oxygen is 15.9 kcal/mol (Equation 8, Table 111). It is not clear how large the additional kinetic barrier would be at the active site, but it is likely that it would be smaller than the barrier in solution of 2-3 kcal/mol (Perrin, 1984). The total calculated kinetic barrier is, therefore, 18-19 kcal/mol or less. Within the necessary uncertainties in predicting reduction potentials at the active site, this is very close to the maximum allowable barrier of approximately 17 kcal/mol, consistent with the formation of radical intermediates that would essentially instantaneously collapse by carbon-oxygen bond formation (step 3).

APPENDIX
Kinetic expressions for ordered binding mechanisms are provided in Table IV. Rate constant ratios defining k,,,/K, for tryptophan are given for the limiting cases of high and low 0 2 concentrations. This approach has been used to predict the behavior of competing isotopically labeled substrates (Klinman et al., 1980). The isotope effects are a simplified case in which the assumption can be made that any difference in rates of turnover must arise from steps associated with a chemical reaction (b in this case). The treatment can be generalized to any two competing substrates, in this case 6fluorotryptophan versus tryptophan, with one important limitation.
If O2 binds first, then it can be said without qualification that the relative rates of turnover of two competing substrates will not change as the O2 concentration is varied. This can be seen in the ratio of rate constants at low versus high O2 concentrations. This ratio depends only on the binding constant for binding of O2 to free enzyme and will therefore be independent of the identity of the substrate. If tryptophan binds first, then any differences in rate between alternative substrates that are reflected in rate constants k2, ks, k4, or kg will be observed only at low 0 2 concentrations. This is parallel to the situation for competing isotopically labeled substrates, for which the differences in rate would be associated only with kg. However, in the more general case of any two competing substrates, observed differences in rate can also be associated with kl, and this rate constant is a factor in the rate expression at all O2 concentrations. In the general case, therefore, is only possible to say that relative rates may change and that the direction of the change may be to increase or decrease with changing O2 concentrations, depending on the contribution of the rate of binding, kl, to the difference in rates.
For 6-fluorotryptophan versus tryptophan we would argue that the difference in kcJK,,, is not significantly determined by kl since the relative rates vary both with pH and DzO versus H20 as solvent. We argue that, in the present case, the increased reactivity of 6-fluorotryptophan is an effect on a chemical step of the reaction rather than simply on the rate of binding. The data for 6-fluorotryptophan at high and low 02, therefore, support a kinetic mechanism in which O2 binds first.