Monooxygenation of an Aromatic Ring by F43W/H64D/V68I Myoglobin Mutant and Hydrogen Peroxide

Myoglobin (Mb) is used as a model system for other heme proteins and the reactions they catalyze. The latest novel function to be proposed for myoglobin is a P450 type hydroxylation activity of aromatic carbons (Watanabe, Y., and Ueno, T. (2003) Bull. Chem. Soc. Jpn. 76, 1309–1322). Because Mb does not contain a specific substrate binding site for aromatic compounds near the heme, an engineered tryptophan in the heme pocket was used to model P450 hydroxylation of aromatic compounds. The monooxygenation product was not previously isolated because of rapid subsequent oxidation steps (Hara, I., Ueno, T., Ozaki, S., Itoh, S., Lee, K., Ueyama, N., and Watanabe, Y. (2001) J. Biol. Chem. 276, 36067–36070). In this work, a Mb variant (F43W/H64D/V68I) is used to characterize the monooxygenated intermediate. A modified (+16 Da) species forms upon the addition of 1 eq of H2O2. This product was digested with chymotrypsin, and the modified peptide fragments were isolated and characterized as 6-hydroxytryptophan using matrix-assisted laser desorption ionization time-of-flight tandem mass spectroscopy and 1H NMR. This engineered Mb variant represents the first enzyme to preferentially hydroxylate the indole side chain of Trp at the C6 position. Finally, heme extraction was used to demonstrate that both the formation of the 6-hydroxytryptophan intermediate (+16 Da) and subsequent oxidation to form the +30 Da final product are catalyzed by the heme cofactor, most probably via the compound I intermediate. These results provide insight into the mechanism of hydroxylation of aromatic carbons by heme proteins, demonstrating that non-thiolate-ligated heme enzymes can perform this function. This establishes Mb compound I as a model for P450 type aromatic hydroxylation chemistry.

Myoglobin (Mb) 1 is a small (17 kDa), well characterized heme protein that is often used as a model system for other heme proteins and the reactions they catalyze. In addition to its native function as an oxygen carrier, Mb has been engineered to efficiently perform peroxidase, catalase, and peroxygenase (sulfoxidation and epoxidation) activities (1)(2)(3)(4). Scheme 1 shows three major alternate pathways for the reaction with peroxide. The latest novel function to be proposed for Mb is cytochrome P450-type aromatic carbon hydroxylation. Although hydroxylation by P450s has been extensively studied (5)(6)(7), the mechanism is still not fully understood (6,8). Input from new model systems could provide additional insight. P450 enzymes capture their substrates near the heme via specific interactions (9 -11), which Mb cannot do. A Trp was, thus, engineered in the heme pocket of Mb to model P450 hydroxylation of aromatic compounds ( Fig. 1) (12,13). The monooxygenation product has not previously been observed or isolated due to rapid subsequent oxidation steps.
The mechanism of hydroxylation by P450s is proving to be quite complex and may even vary among various P450s (7). Some of the proposed mechanisms involve 1) an epoxide intermediate for the aromatic hydroxylation, 2) a concerted direct insertion of oxygen, 3) a non-concerted sequential insertion involving hydrogen abstraction and oxygen rebound, or 4) a more simple non-concerted radical mechanism (6 -8, 14). Recently it has been found that these mechanisms may not be entirely valid, and the actual mechanism may involve two electrophilic oxidants and/or two spin states of the iron oxo species (6,8,14). Although P450cam has been shown to lose its hydroxylation activity when an imidizole is substituted for the axial thiolate ligand (15,16), in contrast, chloroperoxidase retains its chlorination, peroxidation, epoxidation, and catalase activities when its thiolate ligand is mutated to histidine (17). Also, studies with iron porphyrin compounds have demonstrated that imidazole ligation can replace the thiolate in P450 type reactions under certain conditions (18). Therefore, because the study of P450s requires specialized low temperature techniques to trap its intermediates (5), other systems such as myoglobin could provide useful insight once suitable substrate binding sites are engineered.
When Mb reacts with peroxide it forms a two-electron oxidized species called compound I that consists of an iron ferryl species and a porphyrin cation radical (Scheme 1). Compound I can perform either two-electron oxidation chemistry (catalase and peroxygenase activity), regenerating the ferric resting state, or one-electron oxidation chemistry (peroxidase activity), generating compound II, which consists of a ferryl heme species. Compound II can then perform a second one-electron oxidation which returns the enzyme to its ferric resting state.
The porphyrin radical species of compound I in WT Mb is very unstable and cannot be detected. It can, however, be detected in certain variants where His-64 has been mutated, for example in the Mb F43H/H64L variant (19). His-64 has been proposed to donate an electron to the porphyrin cation radical, destabilizing compound I. Subsequently Trp and Tyr residues donate an electron forming more stable Trp or Tyr radicals (20 -28). Mutation of this distal His-64 prevents leakage of the oxidizing equivalents vital for two-electron oxidation chemistry away from the catalytic heme center. Wong and co-workers (29) have engineered a Tyr at position 43 of Mb to retain the oxidizing equivalents close to the heme catalytic center. This increased the two-electron, epoxidation rate 25fold over WT Mb. Ozaki et al. (13) also increased the twoelectron sulfoxidation of Mb 15-fold by engineering a Trp, another electron-rich residue at this position 43. In the process of studying Mb with an engineered Trp at the position 43, it was discovered that the protein was oxidatively modified, resulting in an increase in molecular mass of 30 Da (13). Proteolytic digestion with Lys-C indicated the modification was on the peptide fragment consisting of residues 43-47, which includes the engineered Trp-43 (12). It was speculated that the modification could be due to the insertion of two oxygen atoms into Trp-43. Based on mass spectrometry and NMR of the modified peptide fragment, Hara et al. (12) concluded that the modification was indeed due to the insertion of two oxygen atoms into Trp-43, which led to the formation of 2,6-dioxoindole and 2-imino-6-oxoindole derivatives. These products were proposed to be formed via a 6-hydroxytryptophan (6-OH-Trp) intermediate (Scheme 2); however, direct evidence of its formation was not observed.
The previously studied Mb F43W/H64L was modified only by mCPBA and not H 2 O 2 . Only the final product could be detected. The monoxygenation product could not be isolated, pos-sibly due to rapid, subsequent oxidation steps. In this work we have engineered Mb F43W/H64D/V68I, called Mb WDI, which was oxidatively modified by H 2 O 2 and allowed for the characterization of the intermediate(s) formed in the oxidation of the engineered Trp. The proposed 6-OH-Trp intermediate was isolated and characterized by mass spectrometry and NMR, which provided data in support of the mechanism proposed by Hara et al. (12). Its formation was determined to be heme-dependent, which establishes Mb compound I as a model for P450 aromatic carbon hydroxylation chemistry.

EXPERIMENTAL PROCEDURES
Materials-H 2 18 O 2 was prepared from 18 O 2 as described by Sawaki and Foote (30), and the 18 O content of the peroxide was determined to be 90% by oxidation of triphenylphosphine and gas chromatography/MS analysis (31). All other chemicals were obtained from Wako, Nakalai Tesque, and Sigma-Aldrich and used without further purification.
UV-Visible Spectroscopy-The UV-visible spectra of ferric Mb WDI was recorded on a Shimadzu UV-2400 spectrophotometer. The measurements were carried out with 10 M protein in 50 mM potassium phosphate buffer (pH 7.0).
Preparation of Myoglobin Variants-The H64D and V68I mutations were introduced by cassette mutagenesis. The cassette including the desired His-64 and Val-68 substitutions and a new silent HpaI restriction site was inserted between the BglII and HpaI sites. The F43W mutation was introduced by the QuikChange mutagenesis kit (Stratagene). The Mb WL variant was previously prepared (13). Protein was expressed in Escherichia coli strain TB-1, and purification of the mutants was performed as previously reported (32,33). Protein concentration was determined by the pyridine hemochromogen method (34). The purified protein was stored at Ϫ80°C until used.
Reaction with Hydrogen Peroxide or mCPBA-Reactions of the Mb WDI with peroxides were carried out with 20 M protein in 50 mM potassium phosphate buffer (pH 7.4) or 5 mM ammonium acetate buffer (pH 6.9) and analyzed by ESI-TOF MS. Reactions were started by the addition H 2 O 2 or mCPBA (1-20 eq) and incubated at 4°C for 30 min, then immediately set on ice. Analysis was performed using an LCT ESI-TOF MS (Micromass).
Purification of Peptide Fragments Bearing Trp-43-Protein digestion was performed with Lys-C achromobacter (1/100 w/w) in 100 mM Tris-HCl buffer (pH 9.0) containing 2 M urea incubated at 37°C for 12 h or chymotrypsin (1/50 w/w) in 50 mM potassium phosphate buffer (pH 8.0) incubated at 37°C for 7 h. The digestion was stopped by flash-freezing in liquid N 2 . The digested products were analyzed on a Q-STAR hybrid LC/MS system (PerkinElmer Life Sciences). Flow rates were 0.7 ml/min for analytical scale with a Vydac C-18 reverse phase column (0.46 ϫ 25 cm) or 2.0 ml/min on a LC-10ADVP HPLC system (Shimadzu) with a Vydac C-18 reverse phase column (1.0 ϫ 25 cm) for preparative scale. Peptides were eluted with a gradient of solvent A (0.1% trifluoroacetic acid in water) to solvent B (60% acetonitrile and 0.1% trifluoroacetic acid in water) over 55 min. The eluent was monitored either at 215 nm for amide bonds or 280 nm for aromatic residues in peptide fragments. The isolated fragments containing modified Trp-43 (ϩ16 Da) were further purified with a gradient over 120 min. Isolated fragments containing modified Trp-43 (mass ϩ30 Da) were not further purified.
MS/MS Analysis of the Peptide Fragments-The fragments, containing modified Trp-43, were analyzed on a REFLEX III (Bruker Daltonics) or Voyager DE-PRO (PerSeptive Biosystems) for MALDI-TOF MS. The spectrometer was calibrated with angiotensin I (molecular mass 1296.48 Da). 1 l of concentrated sample was mixed with 1 l of saturated ␣ -cyano-4-hydroxycinnamic acid in water/acetonitrile (7:3) and applied to the sample plate by the dried-droplet method. Product ions were obtained using a post-source-decay method. Theoretical ionization products were predicted using Sherpa-lite 4.0.
NMR Spectroscopy-1 H NMR spectra of the fragments containing modified Trp were obtained either on an ECA 800 MHz (JEOL) spectrometer or on an INOVA 500 MHz (Varian) spectrometer. 1 H NMR spectra of 5-OH-L-Trp (Wako) was measured on a 270 MHz GSX-270 NMR (JEOL). 1 H NMR measurements were undertaken in D 2 O at 30°C, and chemical shifts were referenced to HDO. Complete proton resonance assignments were made using double quantum filtered-COSY, two-dimensional total correlation spectroscopy (TOCSY), and ROESY experiments.
Preparation of Apo-unmodified and -modified Mb WDI and Their Reactions with Peroxides-ApoMb WDI was prepared by heme extraction with butanone as previously described (35). To investigate whether SCHEME 1. General reaction cycles of heme proteins with peroxide.
heme was involved in the Trp-43 oxidation, 1-20 eq of H 2 O 2 or mCPBA were added to apoMb WDI as described above. To determine whether the insertion of the second oxygen atom was heme-dependent, 1 eq of H 2 O 2 was added to holo-Mb WDI to generate the ϩ16-Da intermediate. Heme was then removed, and oxidation reactions were carried out. ESI-TOF MS of the reaction mixtures were measured before and after the reaction to determine the mass change.
Stopped-flow UV-Visible Spectroscopy-The spectral changes were monitored on a Hi-Tech SF-43 stopped-flow apparatus equipped with a MG6000 diode array spectrophotometer for a multiwavelength scan or on a Unisoku RSP-601 stopped-flow apparatus for a single-wavelength scan. The reaction with H 2 O 2 was carried out in the presence of 5 M protein in potassium phosphate buffer (pH 7. Determination of Oxygen Source in the Modified Trp-43-One equivalent of H 2 18 O 2 was added to Mb WDI as described above. The reaction mixture was digested with chymotrypsin as well, and the peptide containing the modified Trp-43 was isolated with HPLC. Subsequently they were analyzed by MALDI-TOF MS, and the percentage of 18 O incorporation was determined from the relative MS intensity.

RESULTS AND DISCUSSION
It had been proposed that many heme proteins other than P450s (and chloroperoxidase) could perform the hydroxylation of aromatic carbons (4,12,13), including those lacking thiolate ligation, which was long thought to be the key factor in conferring hydroxylation activity to P450 enzymes (7). A primary factor regulating the hydroxylation is likely the presence of an appropriate binding site that positions the substrate at the correct distance and orientation from the reactive compound I intermediate. We have, thus, used a covalently attached Trp (at position 43) in the distal heme pocket of a myoglobin variant to test this hypothesis (Fig. 1). In previous studies, Trp-43 was oxidized to 2,6-dioxoindole and 2-imino-6-oxoindole derivatives in a 6-electron oxidation process (12). The use of a strong oxidant (mCPBA) precluded the detection of two-electron-oxidized intermediate species. We hypothesized that a different mutant that could be modified by a milder oxidant such as H 2 O 2 may allow the reaction to stop at the two-electron-oxidized intermediate. Substitution of the Val-68 in H64D variants of Mb has been shown to be a key in regulating peroxidase and peroxygenase rates as well as stereoselectivity of the products (36,37). The combination of H64D/V68I mutations in Mb increased peroxygenase activity 1600-fold over WT Mb and altered the dominant product from (R) to the (S) isomer (36). The addition of H64D/V68I mutations to Mb F43W allowed for the oxidation of Trp by a single equivalent of H 2 O 2 and characterization of this key intermediate.
UV-Visible Spectroscopy of Mb WDI-The UV-visible spectra of ferric Mb WDI in 50 mM potassium phosphate buffer (pH 7.0) is shown in Fig. 2. Ferric Mb WDI displays a Soret band at 408 nm and a visible band at 505 nm with a shoulder at 540 nm assignable to ␣ and ␤ bands, respectively, and a band at 636 nm assignable to a charge transfer band. The ratio of A 408 /A 280 (RZ) is 3.2, which is lower than WT Mb but consistent with the addition of an extra Trp residue at position 43 in close proximity to the heme. This spectrum contains a broad shoulder at 380 nm, typical of penta-coordinated high spin heme found in WT Mb. The UV-visible data suggest no major structural changes affecting the heme of the Mb WDI variant have occurred.
Oxidation of Mb F43W/H64D/V68I with Hydrogen Peroxide and mCPBA-In contrast to the previously studied Mb F43W and Mb F43W/H64L (called Mb WL) variants with an engineered Trp at position 43 (12,13), which were not modified by H 2 O 2 , Mb WDI was modified by H 2 O 2 . ESI-TOF MS data shows that the addition of 1 eq of H 2 O 2 to Mb WDI resulted in an increase in mass of 16 Da from 17,360 to 17,376 Da (Fig. 3). In the previous studies the strong oxidant (mCPBA) precluded the detection of this ϩ16-Da species. Although it is not possible to determine the nature of the modification from the MS data, these results are consistent with the formation of a 6-OH-Trp intermediate formed with 1 eq of H 2 O 2 . SCHEME 2. 1, proposed intermediates in tryptophan oxidation. 2, possible rearrangement of 2,6-dioxoindole to 2-imino-6-oxoindole.
The addition of 3 eq of H 2 O 2 (Fig. 3) resulted in the almost complete conversion to a species with an increase in mass of 30 Da. The ESI-TOF MS spectrum after the addition of 5 eq of H 2 O 2 showed a clear conversion to the ϩ30-Da species had occurred (Fig. 3). Formation of 2,6-dioxoindole and 2-imino-6oxoindole, tryptophan derivatives identified by Hara et al. (12) for Mb F43W/H64L, is consistent with the increase of 30 Da.
The reaction of Mb WDI with mCPBA revealed that more equivalents of mCPBA than of H 2 O 2 were required to increase the mass of the enzyme (Fig. 3). These results are interesting because mCPBA is a stronger oxidant than H 2 O 2 ; therefore, less should be required. Because mCPBA is bulkier, its access to the heme iron may be more restricted than H 2 O 2 . A crystal structure may aid to confirm this hypothesis. The possibility that mCPBA oxidizes the easily oxidizable amino acid residues such as Met on the surface of the protein is consistent with the formation of higher molecular weight species observed in the ESI-TOF MS spectrum (38).
Proteolytic Digestion and Purification of Modified Products-To determine the location of modification, the protein was digested with chymotrypsin or Lys-C after reaction with 1 or 5 eq of H 2 O 2 , and the resulting peptides were separated by reverse phase HPLC as described under "Experimental Procedures." LC/MS and MALDI-TOF MS analysis were used to identify the peptides in each peak and/or fraction. Peaks corresponding to the unmodified fragment WDRFK (751 Da) of residues 43-47 containing the engineered Trp-43, as predicted from a theoretical digest with Sherpa Lite 4.0 and a ϩ30-Da product (781 Da), could be detected in the Lys C-digested sample. This corresponds to the insertion of two oxygens and the loss of two protons as shown by Hara et al. (12); however, a ϩ16-Da product from single oxygen insertion was not detected. Either the ϩ16-Da product is unstable under the digestion conditions (pH 9.0 at 37°C for 12 h), or it does not ionize well, or it is destroyed upon ionization. The inability to detect a Lys-C-digested fragment with an increase in mass of 16 Da (767 Da) proposed to be due to a 6-OH-Trp intermediate prompted a switch to chymotrypsin. Chymotrypsin has a milder optimum for digestion (pH 8.0 for 6 h at 37°C) than Lys-C (pH 9.0 for 12 h at 37°C). Analysis of the sample digested with chymotrypsin after the reaction with 1 eq of H 2 O 2 by LC/MS resulted in the detection of a peak at 896 Da. This peak was consistent with the molecular mass of the modified (ϩ16 Da) EKWDRF fragment if chymotrypsin did not cleave at the modified tryptophan. The missed cleavage provides support for the modification occurring on Trp-43 since modification with oxygen(s) would make this residue more hydrophilic and bulkier to the extent that it could possibly no longer bind in the hydrophobic active site of chymotrypsin. This 896 Da (ϩ16 Da) NMR Characterization of Digested Modified Samples-HPLC purified ϩ16-Da (see Fig. 6) and ϩ30-Da peptide fragments were characterized by one-dimensional and two-dimensional (COSY and ROESY) 1 H NMR. First, it was important to confirm the nature of the ϩ30-Da product. The ϩ30-Da product was expected to be identical to the product identified by Hara et al. (12), however it was not infeasible that the product could be different since a different Mb variant and different oxidant were used. Peaks at 6.61 (s), 6.65 (d), and 7.21 (d) ( Table I) can be assigned to C7, C5, and C4 protons, respectively. Compari-son of the 781 Da (ϩ30 Da) fragment NMR spectra with data from Hara et al. (12) clearly indicates this product is identical to the 2-imino-6-oxoindole obtained by Hara et al. (12) (Table I).
Coupling between C4 and C5 protons in the COSY spectrum also supports this assignment. That the final products obtained in the Mb WL and the Mb WDI variants are identical provides strong evidence that the variants share a common mechanism.
In the 896-Da (ϩ16 Da) fragment spectrum the aromatic region between 6.5 and 7.5 ppm contains 7 peaks (Fig. 5). Table  I lists the chemical shifts and coupling constants. They have been assigned to a modified 6-OH-Trp-43 and Phe-46. Coupling between the C4 and ␣ and ␤ protons in the ROESY spectra (Fig.  5B) assigns the C4 proton to the peak at 7.43 ppm. Based on this assignment, the resonances at 6.70 (dd), 6.86 (d), 7.07 (s), and 7.43 (d) ppm are unambiguously assigned to C5, C7, C2, and C4 protons of 6-OH-Trp-43, respectively. Furthermore, these shifts (and coupling constants) are consistent with those previously reported for 6-OH-D,L-Trp (39). Coupling between the C4 and C5 protons in COSY and ROESY spectra also supports this assignment. Comparison with authentic 5-OH-L-Trp (Wako) spectra further rules out 5-OH-Trp. The NMR data for 4-OH-Trp and 7-OH-Trp as obtained by Van Wickern et al. (39) are also included in Table I   Heme Dependence of Mb F43W/H64D/V68I Oxidation-Heme was extracted from Mb WDI to determine whether apoMb WDI could be modified by H 2 O 2 . No change in mass occurs with up to 20 eq of H 2 O 2 or 5 eq of mCPBA. Heme is, thus, required for the formation of the ϩ16-Da oxidized product, as depicted in Scheme 3. Heme was also extracted after formation of the ϩ16-Da modified species. The resulting apo ϩ16-Da Mb WDI also did not show any mass increase when reacted with up to 20 eq of H 2 O 2 . These studies suggest that both the formation of the ϩ16-Da intermediate and subsequent oxidation to form the final ϩ30-Da product are catalyzed by the heme cofactor, most probably via the compound I intermediate.
Stopped Flow UV-Visible Studies of the Reaction of Mb WDI with Hydrogen Peroxide-The reaction of Mb WDI with H 2 O 2 was monitored by stopped flow UV-visible spectroscopy. The spectra and absorbance change at 408 nm are shown in Fig. 6. The Soret band in the initial spectra is decreased in intensity and slightly blue-shifted from the resting state (not captured on this time scale), which is consistent with a partial accumulation of compound I. The spectrum rapidly returned to the ferric resting state described above for the modified samples. No evidence of compound II formation as occurs in the reaction of WT Mb with excess H 2 O 2 can be detected. The lack of compound II formation could potentially be due to high catalase activity. This is, however, not the case (see below). The shoulder at 380 nm is less prevalent. This feature is consistent with a conversion a hexacoordinated high spin heme state. A water or hydroxyl group of 6-OH-Trp could provide the sixth ligand and may result in a less reactive ferric state. A modification of the Trp-43 that hydrogen bonds to an axially bound water molecule as shown in Scheme 4 could result in such spectral features. Additional H-bonding may also be provided by Asp-64 (Scheme 4).
Catalase Activity Assay of Mb Variants-To investigate if the inability to detect compound II was due to high catalase activity, oxygen evolution by Mb WDI in the reaction with excess H 2 O 2 was monitored. The results are shown in Table II. The initial rate was found to be only four times higher than that of WT Mb (Table II), which should not account for the lack of compound II detection. Interestingly, only about four turnovers  occur in the catalase reaction under the conditions of the assay. Catalase activity of the ϩ30-Da modified enzyme was too low to be detected. This may indicate that the final product spectra might be of an inhibited form of the enzyme, possibly an unreactive ferric state with a strong hydrogen-bonding interaction with water as shown in Scheme 4.
Determination Although direct opening of the epoxide to produce only 6-OH-Trp (Scheme 5) should be more likely than hydrolysis followed by elimination of a hydroxyl group, this would result in 100% incorporation of labeled oxygen, contradicting the observed results. Monooxygenation by P450 enzymes has been proposed to occur via a number of mechanisms (6 -8). In addition to the epoxidation mechanism, concerted direct insertion of oxygen or non-concerted stepwise insertion involving hydrogen abstraction and oxygen rebound could afford 6-OH-Trp (6 -8). The direct hydroxylation mechanism is also expected to afford 6-OH-Trp with 100% incorporation of labeled oxygen. If we assume the oxygen exchange between compound I and water molecule, ϳ50% incorporation of labeled oxygen could be observed even though the oxidation proceeds without forming the diol intermediate (40 -42). Although we have observed more than 90% incorporation of 18 O in the products of sulfoxidation and epoxidation (4,43,44), we are not able to eliminate the possible involvement of oxygen exchange.
To our knowledge Mb WDI is the first enzyme to preferentially form 6-OH-Trp. P450 oxidation of indole (Trp side chain) has been shown to yield the pigments indigo and indirubin via 3-OH-indole (indoxyl) (45). In this reaction 6-OH-indole is detected as a side product. Although the mechanism of its formation was not discussed, the possibility that it is identical to that used by Mb WDI cannot be ruled out. Oxindole, which has an oxygen inserted at the C2 position, was also detected in that study. P450 BM-3 has also been engineered to form indigo and indirubin (46). It does so by forming a mixture of 2-OH-indole and 3-OH indole, which is then further oxidized. Naphthalene dioxygenase also preferentially oxidizes the C2 and C3 positions of indole to form 2,3-dihydroxyindolin (47). Styrene monooxygenase also shows a preference for epoxidation of the double bond between the C2 and C3 positions of indole forming the 2,3-epoxyindole (48). In contrast to these enzymes, tryptophan hydroxylase hydroxylates Trp at the C5 position (49). The preference for oxidation at the C2 and C3 positions by the abovementioned enzymes and position C5 by tryptophan hydroxylase or C6 by Mb WDI cannot be due solely to the fact that the C2 position of Trp is substituted (␤-carbon). Indole 2,3dioxygenase, another heme enzyme which binds Trp as a substrate, specifically oxidizes the C2 and C3 positions to degrade Trp to N-formylkynurenine (50).
The specificity of these enzymes for yielding different products further highlights the importance of substrate binding orientation and position in the active site. Crystallographic studies of our F43W variants may yield further insight into the structural factors controlling the oxidation mechanism for selective hydroxylation at the C6 position.
Recently two Trp residues in cytochrome c oxidase were reported to be oxidatively modified, resulting in an increase of 16 Da (51). These two Trp residues are 44 and 62 Å from the Cu B binuclear heme site. In this case the modification has been proposed to occur via the migration of radicals formed at the heme though the protein. A recent crystal structure of lignin peroxidase shows that the ␤-carbon of the catalytically active Trp at position 171 is hydroxylated (52). This is also proposed to occur via radical transfer from the heme. As discussed above, a radical mechanism would not likely account for ϳ50% incorporation of labeled 18 O into 6-OH-Trp-43 and is, thus, not likely occurring in Mb WDI. The nature of the ϩ16-Da products in cytochrome c oxidase may provide further insight into the difference between their respective mechanisms. Alternatively, a number of mechanisms may occur simultaneously to yield 46% incorporation of labeled 18 O, but this warrants further investigation.
Implications for Understanding and Engineering Monoxygenation Activity in Heme Proteins-Monoxygenation by cytochrome P450 enzymes has been well characterized (5)(6)(7)(8)53). One factor preventing monoxygenation of various substrates by WT Mb and many other heme proteins is likely the lack of an appropriate substrate binding site in these proteins. Even though Mb does oxidize a number of substrates including thioanisole and styrene in an enantioselective manner, evidence for a specific high affinity binding site for these substrates has not been provided. The covalently attached indole of Trp-43 provides a model system that can be compared with other proteins. Although class I peroxidases (cytochrome c peroxidase, ascorbate peroxidase, and bacterial catalase peroxidases) also contain Trp in the distal heme pocket, this Trp is never reported to be hydroxylated. The reason for this is likely based on orientation, since the distance from the heme iron to Trp is 4.1 Å, similar to that of camphor to the heme iron of P450cam (4.2 Å) (54) and the modeled distance of Trp-43 to the heme iron in Mb WL (4.9 Å) (12). In Mb WDI the Trp is thought to be perpendicular to the heme, and the C6-C7 bond may be almost directly over the heme iron such that the Trp orbitals can interact with the lone pair electrons of the oxygen. Poulos and coworkers (55) have oxidized the distal Trp in cytochrome c peroxidase and cross-linked it to an engineered Tyr by replacing the distal His (55). A radical mechanism involving the SCHEME 4. Proposed coordination state of heme after modification. engineered Tyr is proposed in this case. Again, the engineered Tyr appears to be appropriately positioned for oxidation, forming a Tyr radical that then migrates to the adjacent Trp. We have demonstrated that variants of Mb (and possibly many other heme proteins lacking thiolate ligation) can perform stoichiometric monooxygenation of aromatic carbons under certain conditions if the substrate can be appropriately positioned. Efforts are currently under way to engineer specific, high affinity binding sites that position exogenous substrate at a suitable distance and orientation for monoxygenation. Although our current computer model of Mb WL and native P450 structures provides suitable starting points, we hope to gain greater insight from the crystal structures (work in progress) of our Trp-43 variants, ideally in both the native and modified forms.
In summary, Mb WDI was stoichiometrically oxidized by H 2 O 2 . The addition of 1 eq of H 2 O 2 results in a 16-Da mass increase. MALDI-TOF MS/MS shows the addition of 16 Da to be on Trp-43, and NMR data has been unambiguously assigned to 6-OH-Trp, confirming the proposal by Hara et al. (12). To our knowledge this is the first report of an enzyme that preferentially hydroxylates the indole ring of Trp at the C6 position. Upon the addition of 5 eq of H 2 O 2 , almost all of the protein is converted to a product with a 30-Da mass increase. NMR data showed this final product to be identical to that previously characterized by Hara et al. (12), suggesting a common mechanism for the different variants and oxidants. Heme extraction was used to demonstrate that both the formation of the (ϩ16 Da these results establish that thiolate ligation may not be required for hydroxylation by heme proteins and that Mb compound I could serve as a model for P450 type chemistry.