Contrasting Mechanisms of Aromatic and Aryl-Methyl Substituent Hydroxylation by the Rieske Monooxygenase Salicylate 5-Hydroxylase

Abstract

The hydroxylase component (S5HH) of salicylate-5-hydroxylase catalyzes C5 ring hydroxylation of salicylate but switches to methyl hydroxylation when a C5 methyl substituent is present. Use of ¹⁸O₂ reveals that both aromatic and aryl-methyl hydroxylations result from monooxygenase chemistry. The functional unit of S5HH comprises a nonheme Fe(II) site located 12 Å across a subunit boundary from a one-electron reduced Rieske-type iron-sulfur cluster. Past studies determined that substrates bind near the Fe(II), followed by O₂ binding to the iron to initiate catalysis. Stopped-flow-single-turnover reactions (STO) demonstrated that the Rieske cluster transfers an electron to the iron site during catalysis. It is shown here that fluorine ring substituents decrease the rate constant for Rieske electron transfer, implying a prior reaction of an Fe(III)-superoxo intermediate with substrate. We propose that the iron becomes fully oxidized in the resulting Fe(III)-peroxo-substrate-radical intermediate, allowing Rieske electron transfer to occur. STO using 5-CD₃-salicylate-d₈ occurs with an inverse KIE. In contrast, STO of a 1:1 mixture of unlabeled and 5-CD₃-salicylate-d₈ yields a normal product isotope effect. It is proposed that aromatic and aryl-methyl hydroxylation reactions both begin with Fe(III)-superoxo reaction with a ring carbon, yielding the inverse KIE due to sp²→sp³ carbon hybridization. After Rieske electron transfer, the resulting Fe(III)-peroxo-salicylate intermediate can continue to aromatic hydroxylation, whereas the equivalent aryl-methyl intermediate formation must be reversible to allow the substrate exchange necessary to yield a normal product isotope effect. The resulting Fe(III)-(hydro)peroxo intermediate may be reactive or evolve through a high-valent iron intermediate to complete the aryl-methyl hydroxylation.

Graphical Abstract

INTRODUCTION

Rieske oxygenases (ROs) are nonheme iron-containing enzymes that catalyze a remarkably expansive range of reactions and act upon a diverse group of substrate chemical types.^1–10 Found widely across nature,^11–29 the ROs are increasingly appreciated for their roles and applications in human health,^{28, 30–33} environmental biotechnology,^34–41 agriculture,⁴² and chemoenzymatic synthesis.^43–48 Collectively, the catalytic repertoire of the RO family exceeds even those of other well-studied oxygenases, including cytochrome P450, dinuclear iron-containing hydrocarbon monooxygenases, flavin oxygenases, and α-ketoacid-linked monooxygenases.^{7, 49–55}

The RO family has a highly conserved quaternary structure (generally α₃β₃ or α₃) with a Rieske-type [2Fe-2S] iron-sulfur cluster and a 2-His-1-carboxylate type mononuclear Fe(II) site in each α-subunit. The head-to-tail organization of the α-subunits places the Rieske cluster of one subunit adjacent to the mononuclear iron site of another, and this cross-subunit boundary pair of metal centers is considered to be the reactive unit.⁵⁶ Oxygen activation and substrate oxidation occur at the mononuclear iron site with the reduced Rieske cluster [Fe(II)-Fe(III)] providing an essential electron for the process.^57–59 Despite the many fundamental similarities of the RO enzymes, considerable specificity in substrate selection, reaction type, and regiospecificity is observed for individual family members. X-ray crystal structures of ROs that catalyze a range of reaction types fail to distinguish a basis for this specificity in the structural regions encompassing the Rieske cluster, the residues between the metal centers, and mononuclear iron ligation, which are remarkably conserved.^{31, 38, 39, 56, 59–73} Progress has been made in understanding the key features of the active site surrounding the mononuclear iron that orient the substrate and provide a level of regiospecificity.^{27, 59, 72, 74–77} However, there is currently little insight into whether the various reactivities require different types of activated oxygen species or how such divergent species might evolve from a highly conserved metal center.

The aromatic cis-dihydrodiol-forming subclass of Rieske dioxygenases have been the most intensively studied in terms of delineating the substrate range, mechanism, and biophysical properties.^{49, 58, 76, 78–86} Transient kinetic and computational studies coupled with diagnostic substrates and spectroscopic techniques have led to the proposal that substrate binding near the mononuclear Fe(II) site causes loss of a water ligand, permitting subsequent O₂ binding to the metal to yield an Fe(III)-superoxo intermediate (Scheme 1).^{82, 84, 87} This intermediate is proposed to react at an electron rich carbon of the aromatic substrate ring based on two key observations: (i) the rate constant for electron transfer from the Rieske cluster was shown to depend on the specific substrate present in the active site, implying reaction of the one electron reduced oxygen species with substrate before transfer of the second electron required for product formation,^{78, 84} and (ii) using a set of ring fluorinated substrate analogs, the log of the rate constant for Rieske electron transfer was shown to correlate linearly with the net charge at the position of attack for cis-dihydroxylation in benzoate 1,2 dioxygenase (BZDO).⁸⁴ During reaction with the Fe(III)-superoxo intermediate, one-electron oxidation of the aromatic ring would yield an Fe(III)-peroxo-aryl radical intermediate. The now fully oxidized mononuclear Fe(III) site would allow electron transfer from the reduced Rieske cluster to yield a bridging Fe(III)-aryl-peroxo intermediate that can undergo O-O bond cleavage to yield an epoxide intermediate by a mechanism akin to one proposed for electrophilic aromatic substitution.^88–91 The cis-dihydrodiol product could be formed by reaction of the Fe(III)-OH with the carbocation that results from ring-opening of the epoxide.⁸² Several other mechanistic proposals have been advanced⁷ based on formation of a dioxetane,^{56, 92, 93} hydroperoxo,^{81, 94} or HO-Fe(V)=O intermediates.^{58, 86, 95} However, formation of these initially reactive species requires transfer of the Rieske electron prior to reaction of the activated oxygen species with the substrate. The strong dependence of the rate constant for electron transfer from the Rieske cluster on the type and charge distribution of the substrate suggests that electron transfer before a reaction with the substrate is unlikely.⁸⁴

Scheme 1.

Mechanism Proposed for Rieske Dioxygenases

More recently, attention has turned to the Rieske monooxygenase class, members of which fall into several subclasses that catalyze hydroxylation of aromatic,^{68, 96, 97} cyclic,^{26, 70} and aliphatic^{12, 19, 98, 99} substrates. Additionally, hydroxylation of heteroatoms,¹⁰⁰ dehydrogenation,¹⁰¹ desaturation,²¹ and O- and N- demethylation^{42, 71, 73, 102–105} of numerous environmentally relevant or economically valuable substrates have been reported.^{5, 106, 107} Our studies have focused on salicylate 5-hydroxylase (S5H) which catalyzes the formation of gentisate (2,5-dihydroxybenzoate) as part of one of the naphthalene biodegradation pathways in Ralstonia.^{108, 109} The hydroxylase component (S5HH) of the three component S5H system (reductase, ferredoxin, and hydroxylase) appears to catalyze strictly monooxygenase chemistry, while conserving both the α₃β₃ quaternary structure and metal cofactors of Rieske dioxygenases.^{59, 108–110} The initial studies of the S5HH mechanism surprisingly showed that the rate constant for electron transfer from the Rieske cluster when benzoate was used as the substrate in place of salicylate was slowed approximately 20-fold. Thus, the reaction exhibited the same dependence on the specific structure and electronic characteristics of the substrate as described for the Rieske dioxygenases.^{84, 108} A very similar mechanism was proposed based on initial attack by an Fe(III)-superoxo species (Scheme 2). The two mechanisms were proposed to differ in either the mechanism of O-O bond cleavage or, if an epoxide is formed in both cases, the steps subsequent to its opening. The monooxygenase reaction would involve internal hydride transfer rather than hydroxyl transfer from Fe(III)-OH. Ligand-to-metal charge transfer from the aromatic phenolate product from salicylate during the S5HH catalyzed reaction allowed direct demonstration of the formation of a chromophoric Fe(III)-product complex as the penultimate step in the reaction cycle.

Scheme 2.

Mechanism Proposed for Rieske Monooxygenase S5HH

S5HH is a particularly relevant target for detailed mechanistic studies because: (i) it has been structurally characterized,⁵⁹ (ii) it reacts with a wide variety of potentially diagnostic salicylate analogs including benzoate, the substrate of cis-diol-forming BZDO,^{108, 109, 111} and (iii) it reportedly catalyzes aryl-methyl substituent hydroxylation in addition to aromatic hydroxylation.¹⁰⁹ The latter two reactions are mechanistically distinct where encountered in other well-studied types of oxygenases,^112–114 and thus they offer the opportunity to explore catalytic diversity in a single, native active site of a RO. In the current study, the reactivity and the mechanism of O₂ activation by S5HH are explored in more detail using alternative substrates, product identification, computational approaches, and kinetic and product isotope effects. The results support a common mechanism for all RO reaction types through the point of electron transfer from the Rieske cluster. Thereafter, the pathways appear to diverge in response to the electronic properties and positioning of the substrate in the active site as manifested in the kinetics of subsequent reaction steps.

EXPERIMENTAL PROCEDURES

General Methods.

Standard procedures and chemical suppliers are described in the Supporting Information.

Production of S5H Proteins.

S5H reductase (S5HR), S5H ferredoxin (S5HF), and His-tagged S5HH were purified as previously described.¹⁰⁸ His-tagged S5HH destined for tag cleavage was purified as previously described for His-tagged S5HH up to the enzyme bound column wash step.¹⁰⁸ At this point, the column (25 ml Roche cOmplete His-Tag purification column) was washed with 50 mM HEPES, 300 mM NaCl, 10 % glycerol, 20 mM imidazole, pH 8 buffer without dithiothreitol (wash buffer). Thrombin cleavage of the His-tag was based on the method of Hefti.¹¹⁵ Thrombin (1000 U, Sigma-Aldrich) was dissolved in 20 ml wash buffer and applied to the column. The thrombin-containing buffer was recirculated overnight using a peristaltic pump. The cleaved protein was eluted with wash buffer and concentrated. The concentrated protein was applied to a Sephacryl-300 HR column (2.5 × 120 cm) equilibrated in 50 mM HEPES, 200 mM NaCl, pH 8 for final purification and buffer exchange. Fractions containing S5HH were pooled, concentrated, and glycerol was added to 5 %. The enzyme was flash-frozen in liquid nitrogen and stored at −80 °C. Reduced S5HH used for single turnover experiments was prepared as described.¹⁰⁸

Preparation of Substrates and Oxygen-containing Solutions.

Substrate solutions were prepared in 50 mM HEPES, 100 mM NaCl, 5 % glycerol, pH 8 buffer (standard buffer). Substrates with low aqueous solubility were prepared at high concentration in methanol and then diluted into the standard buffer to give a final methanol concentration of < 2 %. Oxygen-saturated solutions (1.8 mM) were prepared by bubbling O₂ gas through buffer solutions in capped Wheaton serum vials with needle venting at 4 °C. ¹⁸O₂ solutions were prepared by preparing a capped Wheaton serum vial completely filled with anaerobic standard buffer (14 ml). Then ¹⁸O₂ gas was flowed into the inverted vial equipped with a vent needle to allow 2 ml of buffer to be displaced. The resulting solution was vortexed, and then vigorously stirred on ice for 30 m (1.8 mM ¹⁸O₂ at 4 °C).

Stopped-flow Measurements and Analysis of Single Turnover Reactions.

The pre-steady state kinetics of reduced S5HH with substrates were measured and the time courses fit using nonlinear regression analysis to a multi-summed-exponential expression as previously described.¹⁰⁸ All enzyme concentrations in this study refer to active site concentrations rather than total protein concentration in the trimeric enzyme. Standard reaction conditions after mixing: 30 μM reduced S5HH active sites (His-tag removed), 5 mM substrate, 900 μM O₂, standard buffer, 4 °C.

Kinetic Isotope Effects.

The kinetic isotope effect (KIE) for the Rieske oxidation reaction was measured for salicylate and 5-CH₃-salicylate via stopped-flow single turnover experiments. Substrate solutions (10 mM) were prepared and then the UV-visible spectrum of each solution was measured. Appropriate dilutions were made so that the solutions gave the same absorbance at λ_max. Reduced S5HH (30 μM) was reacted with salicylate, salicylate-d₆, 5-methyl-salicylate or 5-methyl-salicylate-d₈ (5 mM) in oxygen-saturated (900 μM) standard buffer using an Applied Photophysics SX.18MV stopped-flow spectrophotometer at 4 °C and monitored at 453 nm (all concentrations after mixing). The KIE was determined from the ratio of the Rieske oxidation first phase reciprocal relaxation times (1/τ₁) for unlabeled substrate and fully deuterated substrate.

Rapid Chemical Quench.

Anaerobic procedures were performed in a Coy anaerobic chamber. Deoxygenated, stoichiometrically reduced, His-tagged S5HH (230 μM active sites) plus the internal standard (ISTD) 3,5-dihydroxy-benzoic acid (30 μM) was reacted with 5-methyl-salicylate (5 mM) in O₂ saturated standard buffer (900 μM O₂) using an Update Instruments 715 rapid mixing ram syringe system (all concentrations after mixing). Control experiments showed that the ISTD is not an S5HH substrate or inhibitor. The single turnover reaction with 5-methyl-salicylate was chemically quenched at discrete timepoints by rapid dispensation into an equal volume of rapidly stirring 10 % trifluoroacetic acid (TFA) in a Reacti-vial. After vortexing for 30 s, an equal volume of 1 M HEPES, pH 8 was added. The sample was centrifuged to remove precipitated protein. Product formation at each quench time point was analyzed using high-pressure liquid chromatography (HPLC) (See Supporting Information).

Reactions with ¹⁸O₂ and H₂¹⁸O.

Incorporation of oxygen from O₂ into salicylate substrates was assessed via single turnover reactions of stoichiometrically reduced S5HH (25 μM) with salicylate (5.37 mM), 4-fluoro-salicylate (56.7 mM), 5-fluoro-salicylate (56.6 mM), or 5-methyl-salicylate (5.66 mM) in the presence of ¹⁸O₂ (1.55 mM) at 23 °C. Anaerobic substrate solutions were added to reduced S5HH prior to addition of ¹⁸O₂ saturated buffer (total volume was 2.32 ml). After 1 h, the reactions were processed as described in Supporting Information for analysis using gas-chromatography-mass spectrometry (GCMS).

The ¹⁸O-buffer was prepared by lyophilizing standard buffer, and then adding H₂¹⁸O to the initial buffer volume. For single turnover reactions in H₂¹⁸O, S5HH was exchanged into ¹⁸O-buffer via concentration and dilution (yielding 96 % H₂¹⁸O). To ensure full exchange of S5HH active site water molecules, the enzyme was incubated in ¹⁸O-buffer overnight at room temperature. S5HH was deoxygenated on ice by flowing high-purity argon gas over the surface with stirring and then stoichiometrically reduced with sodium dithionite in a Coy anaerobic chamber. The enzyme (280 μM) was reacted with salicylate (13.5 mM) or 5-methyl-salicylate (13.9 mM) as described above for the ¹⁸O₂ incorporation reaction except that unlabeled oxygen gas was utilized (total solution volume 0.49 ml). The reactions were then processed for GCMS analysis as described in Supporting Information.

Product Isotope Effect.

Single turnover reactions of stoichiometrically reduced S5HH (28 μM) with a solution of salicylate (7.46 mM) and salicylate-d₆ (7.24 mM) or a solution of 5-methyl-salicylate (7.33 mM) and 5-methyl-salicylate-d₈ (7.5 mM) in oxygen-saturated (1.55 mM after component additions) standard buffer at 23 °C were conducted. After 1 h, the reactions were processed for GCMS analysis as described in Supporting Information. Protiated and deuterated products were determined from the mass spectra of undiluted samples and protiated and deuterated substrates were determined from the mass spectra of the samples diluted 200-fold with dichloromethane. The product isotope effect was computed after corrections for the small differences in substrate concentrations as described in Supporting Information.

Product Identification by NMR Spectroscopy.

Steady-state reactions were conducted to accumulate sufficient quantities of S5HH reaction products for NMR spectroscopy. The reactions contained NADH (5 mM), S5HR (0.4 μM), S5HF (3.2 μM), S5HH (14 μM), and substrate (4.2 mM) in O₂-saturated (1.55 mM after component additions) standard buffer (5.9 ml, 23 °C). Substrates analyzed were: 4-fluoro-salicylate, 5-fluoro-salicylate, 5-methyl-salicylate, or 3-methyl-benzoate. Reactions were conducted in a Reacti-vial with stirring.

S5HH reactions were quenched with the addition of an equal volume of trifluoroacetic acid (10 %) while stirring. The solution was vortexed and then centrifuged. An equal volume of 1 M HEPES, pH 8.0 was added to the supernatant containing the reaction product. The reaction product was purified via HPLC using multiple separate 1 ml injections (see Supporting Information).¹⁰⁸ Fractions were collected and assessed via UV-visible spectroscopy. Fractions corresponding to the elution time of the product chromatogram peak (detected at 238 nm) were pooled and lyophilized. The lyophilized product was dissolved in CD₃OD prior to NMR spectroscopy.

¹H NMR data for the purified S5HH reaction product with substrates 5-methyl-salicylate or 3-methyl-benzoate were collected at the UMN Chemistry NMR Center using a Bruker Avance III HD Nanobay AX-400. Spectra of authentic 5-hydroxymethyl-salicylate and 3-hydroxymethyl-benzoate standards were also recorded.

NMR data for the purified S5HH reaction product with substrates 4-fluoro-salicylate and 5-fluoro-salicylate were collected at the Minnesota NMR Center using a 600-MHz Bruker Avance NEO spectrometer with a 5-mm CPTCI ¹H(¹⁹F)/¹³C/¹⁵N/²H Z-gradient cryoprobe controlled by Topspin 4.0 software. Spectra were acquired at 25 °C in CD₃OD and referenced through the solvent lock (²H) signal according to the IUPAC recommendations for secondary referencing and to the manufacturer’s protocols.

Product Identification by HPLC.

5-Hydroxymethyl-salicylate and 3-hydroxymethyl-benzoate produced from steady state reactions of S5HH with 5-methyl-salicylate or 3-methyl-benzoate, respectively, under the conditions described above were identified by comparison of their HPLC retention time to those of authentic product standards. HPLC was performed as previously described.¹⁰⁸

Computational Methods.

All calculations were performed using the M06–2X functional¹¹⁶ and the 6–311+(2df,p) basis set¹¹⁷ within the electronic structure modeling program Gaussian 16 version c01.¹¹⁸ Geometries were optimized for salicylates and benzoates in their deprotonated conjugate base forms. The SMD (solvation model based on density) continuum solvation model using a dielectric constant of 5.6968, appropriate for the hydrophobic enzyme active site, was used.¹¹⁹ CM5 (charge model 5) atomic charges were calculated with charges of hydrogens summed into those of heavy atoms.¹²⁰ Vibrational frequency calculations were performed for each optimized structure and thermal and entropic contributions added to the self-consistent field energy. Bond dissociation free energies were calculated for bonds broken and formed in the enzyme catalyzed hydroxylation of the substrate.

Bond dissociation free energy (BDFE) of the bond R-X was calculated using Equation 1 where R-X represents the salicylate or benzoate molecule, R^• represents the molecule after bond dissociation, and X^• represents the dissociated atom (either H or F).

$$\text{BDFE}\left(\text{R-X}\right)=\Delta \text{G}({\text{R}}^{•})+\Delta \text{G}({\text{X}}^{•})\text{ - }\Delta \text{G}\left(\text{R-X}\right)$$ Equation 1

Here, we only report calculations using the M06–2X functional; however, additional computations were also performed using the ωB97XD functional,¹²¹ and these computations showed that identical conclusions would be reached.

Electrostatic potential surface maps were produced on the ρ = 0.02 au contour of the molecular electronic density, from the optimized structures computed with the parameters described above. Electrostatic potential surface map images were made in GaussView 6.1.¹²²

RESULTS

S5HH Reactions with Substituted Salicylates and Benzoates.

Early studies used whole cell biotransformations or cell-free extracts to investigate the substrate range of S5HH.^{109, 110} Later, these studies were extended to the use of purified enzyme without direct product identification.¹¹¹ In the current study, we use purified enzyme together with product characterization and various single and multiple turnover kinetic techniques to examine the mechanism(s) of O₂ activation and substrate oxygenation by this Rieske monooxygenase. The gentisate product of the reaction of S5HH with salicylate has been previously identified^{108, 110} The products from reactions of purified S5HH with fluorine- and methyl-substituted salicylates and benzoates are reported in Table 1 and the methods used for product identification are detailed in Experimental Procedures and Supporting Information Figures S1–S13.

Table 1.

Summary of Reactions of Fluoro- and Methyl- Substituted Salicylates and Benzoates with Purified S5HH

Substrate	Product	Rieske oxidation kineticsa 1/τ1 (s−1)	Abs700 increase kineticsb 1/τ (s−1)	Relative ΔAbs700 in first phasec	Computed Substrate Atomic Charge
					C4	C5	C6
None	None	0.14 ± 0.01	0.18 ± 0.01	0.05	-	-	-
Salicylate (Sal)	5-OH-Sald,e	119 ± 10	23.0 ± 1.9	1.0	−0.0169	−0.0451	−0.0217
3-F-Sal	3-F,5-OH-Salf	36.7 ± 2.6	7.9 ± 0.6	0.99	−0.0052	−0.0269	−0.0234
4-F-Sal	4-F,5-OH-Salg.h	1.18 ± 0.09	1.3 ± 0.1	0.83	+0.0823	−0.0310	+0.0016
5-F-Sal	5-F,6-OH-Salg.h	55.9 ± 3.9	4.8 ± 0.4	0.22	−0.0136	+0.0700	−0.0041
6-F-Sal	5-OH,6-F-Salf	26.4 ± 2	13.6 ± 1	0.99	−0.0088	−0.0313	+0.0882
Benzoate (Ben)	3-OH-Bend	3.9 ± 0.3	2.3 ± 0.2	0.91	−0.0222	−0.0227	−0.0216
2-F-Ben	2-F,5-OH-Benf	2.5 ± 0.2	2.2 ± 0.2	0.65	−0.0043	−0.0245	−0.0098
3-CH3-Ben	3-CH2OH-Beng,h,i	0.24 ± 0.05	0.21 ± 0.05	0.06	−0.0281	−0.0303	−0.0242
5-CH3-Sal	5-CH2OH-Salg,h,i,j	3.3 ± 0.3	1.43 ± 0.1	0.07	−0.0317	−0.0436	−0.0163
5-CF3-Sal	None detected	0.42 ± 0.05	0.51 ± 0.05	0.05	+0.0102	−0.0648	+0.0091

^aReaction conditions: 30 μM reduced S5HH (without His-tag), 5 mM substrate, 900 μM O₂, 50 mM HEPES, 100 mM NaCl, 5 % glycerol, pH 8, 4 °C. The reaction time courses are best fit with a 3-summed exponential time course. The fastest and largest amplitude phase (1/τ₁) has been shown to correlate with product formation.¹⁰⁸

^bThe absorbance at 700 nm (Abs₇₀₀) has been shown to arise from both a small increase due to Rieske cluster oxidation and, in some cases, a larger increase from formation of a ferric-phenolate product complex that subsequently dissociates to release the hydroxylated product and complete the catalytic cycle.¹⁰⁸

^cThe largest absorbance increase at 700 nm appears to occur when a ferric-phenolate product complex is formed and when the aromatic hydroxylation is in the salicylate C5 (or benzoate C3) position.¹⁰⁸

^dBy HPLC with an authentic product standard after a single turnover and by LCMS.¹⁰⁸

^eBy NMR after multiple turnover in a whole cell system.¹¹⁰

^fPresumed product based on increase in Abs₇₀₀ after a single turnover reaction monitored by stopped-flow spectroscopy.

^gBy NMR after steady state turnover with purified enzyme components with subsequent HPLC product purification.

^hBy GCMS of the TMS product derivative after a single turnover (4-F-salicylate reaction with ¹⁸O₂).

ⁱBy HPLC using an authentic product standard after steady state turnover.

^jBy NMR after multiple turnover in a whole cell system.¹⁰⁹

Single Turnover Transient Kinetics of S5HH Reactions.

The time course for reaction of S5HH with substrates and O₂ can be monitored by the change in Rieske cluster absorbance at 453 nm as an electron is transferred from the Rieske cluster to the mononuclear iron site (Figure 1). Past studies of the Rieske oxygenases have shown that the time course of Rieske oxidation is best fit by nonlinear regression using a 3-summed exponential equation, indicating 3 kinetic phases.^{78, 84, 108, 123} The unexpected triphasic behavior for transfer of a single electron was attributed to 3 parallel reactions. This finding, along with the assumption of quasi-irreversibility of electron transfer against a substantial redox potential barrier,^{124, 125} allows correlation of the reciprocal relaxation times from computed fits with rate constants for the physical reactions. Accordingly, the fastest and largest amplitude phase (1/τ₁, Table 1) has been shown to correlate with the rate constant for product formation.^{84, 108} The origin of the other phases remains undetermined, but they presumably result from nonproductive oxidation pathways for some of the Rieske clusters. As we have observed previously for Rieske dioxygenases,^{78, 84} the Rieske oxidation kinetics of S5HH are sensitive to the chemical structure of the substrates tested here (Figure 1), indicating a direct role of the substrate in catalytic step(s) that allow the electron transfer to occur.

Figure 1.

Single turnover time courses monitored at 453 nm for reaction of reduced S5HH (30 μM active sites) with O₂ (900 μM) and substrates (5 mM). All concentrations are after mixing. The substrates utilized are identified on the figure (abbreviations as in Table 1). The time courses are offset for clarity. Conditions: 50 mM HEPES buffer, 100 mM NaCl, 5 % glycerol, pH 8.0, 4 °C.

Fluorine substituents slow Rieske oxidation (and product formation) when placed anywhere in the ring, but in most cases, ring hydroxylation still occurs at C5 (Table 1, Figures S1–S4). An exception occurs when fluorine is placed at C5, in which case ring hydroxylation appears to occur at C6. This conclusion is supported by the direct identification of C6 hydroxylation when chlorine is placed at C5 in whole cell experiments¹⁰⁹ and by the NMR characterization presented here in Figures S5–S9. Also, our past studies of S5HH have shown that when products are formed with a OH at C5 (gentisate from salicylate or the equivalent 3-OH-benzoate from benzoate), a complex with a long-lived charge-transfer chromophore at ~700 nm is formed.¹⁰⁸ This species arises as the mononuclear Fe(III) formed during the reaction cycle shifts in the active site during the last step of the reaction to form a complex with the hydroxylated product. As shown in Figure 2, the 6-F-salicylate reaction product forms a similar intense chromophore to that seen during the reaction with salicylate, but no such chromophore is formed as 5-F-salicylate reacts. Indeed, 3- and 4-F salicylates, as well as 2-F-benzoate, form similar ~700 nm chromophores, consistent with C5 (or C3 for benzoate) hydroxylation (Table 1). The lack of a chromophore despite rapid ring hydroxylation for 5-F-salicylate is consistent with hydroxylation at a different ring position.

Figure 2.

Single turnover time courses monitored at 700 nm for reaction of stoichiometrically reduced S5HH (30 μM) with O₂ (900 μM) and substrates (5 mM). All concentrations are after mixing. The substrates utilized are identified on the figure (See Table 1). Conditions: 50 mM HEPES buffer, 100 mM NaCl, 5 % glycerol, pH 8.0, 4 °C.

5-Methyl-Salicylate Oxygenation Occurs on Methyl Rather than the Aromatic Ring.

Product identification by NMR and GCMS after either single or multiple turnover as well as HPLC versus an authentic standard shows that 5-methyl-salicylate (5-CH₃-Sal) is oxygenated to form 5-hydroxymethyl-salicylate (5-CH₂OH-Sal, Table 1, Figures S10–S11).¹⁰⁹ Methyl substituent hydroxylation has also been observed in reactions of other ROs.^{46, 97, 126–128} For S5HH, no evidence of ring hydroxylation was detected and no increase in absorbance at 700 nm occurred beyond that expected for oxidation of the Rieske cluster (Table 1). Each of these findings is suggestive of a distinct type of oxygenation chemistry. The single turnover reaction can again be fit by nonlinear regression using a 3-summed exponential expression indicative of 3 kinetic phases (Figure 3A). The reciprocal relaxation time for phase 1 of this reaction is decreased 35-fold from that of the salicylate reaction. Importantly, as shown in Figure 3B, the rate constant for product formation assessed by rapid chemical quench and HPLC analysis is similarly slowed (RFQ k₁ = 4.9 ± 1.5 s⁻¹ versus k₁ = 4.5 ± 0.5 s⁻¹, rate constants increase slightly when using His-tagged S5HH).

Figure 3.

Time course for product formation from 5-CH₃-Sal by His-tagged S5HH. (A) Nonlinear regression 3-exponential fit (orange dashed line) is shown superimposed on the single turnover reaction time course (black). The fit residual is shown in the inset. The standard reaction conditions given in Experimental Procedures were utilized. (B) Comparison of the time course for product formation with that for phase 1 of Rieske cluster oxidation. The product formed was analyzed using RFQ and the reaction conditions are given in Experimental Procedures. HPLC analysis of the product yield from RCQ samples (⬛) are shown. A non-linear regression single exponential fit is superimposed on the data (red dashed line). The first phase of the single turnover time course monitored at 453 nm was extracted from the data and is shown superimposed (solid blue line).

A 5-Trifluoro-Methyl Substituent Blocks Oxygenation but not Binding.

Single turnover reaction of reduced S5HH with 5-trifluoro-methyl-salicylate (5-CF₃-Sal) and O₂ was found to mimic the autooxidation of the Rieske cluster in the absence of a substrate (Figure 4). Accordingly, no product was detected from the reaction at long times (Table 1). The binding of 5-CF₃-Sal in the S5HH active site was examined by reacting S5HH with an equimolar concentration of salicylate and 5-CF₃-Sal (Figure 4, red trace). The rapid first phase of Rieske oxidation was slowed by 2.55-fold indicating that 5-CF₃-Sal binds in the active site with a similar affinity to salicylate but cannot undergo a monooxygenase reaction (Figure 4).

Figure 4.

Stopped-flow time courses for Rieske oxidation showing the competition between salicylate and 5-CF₃-Sal for binding in the active site. Reduced S5HH (30 μM) was reacted with O₂ (900 μM) and salicylate (364 μM), 5-trifluoro-methyl-salicylate (5 mM), salicylate (364 μM) together with 5-CF₃-Sal (364 μM), or no substrate. All concentrations are after mixing. The buffer was 50 mM HEPES, 100 mM NaCl, 5 % glycerol, pH 8.0, 4 °C.

¹⁸O Incorporation During S5HH Ring Hydroxylation and Aryl-Methyl Hydroxylation Reactions.

Single turnover reactions of S5HH with substrates were conducted in the presence of ¹⁸O₂ to confirm that the origin of the oxygen in the product was O₂ and then to determine the percentage incorporation of ¹⁸O (Figure 5 and Table 2). For both ring hydroxylation and methyl substituent hydroxylation reactions, a single ¹⁸O was incorporated at >85 % occupancy, demonstrating that S5HH is a monooxygenase for both types of reaction. When the potential ¹⁸O source was water, no incorporation was observed (Figure S14 and Table 2). Taken together, both data indicate that the conditions for the experiment with ¹⁸O₂ resulted in a ~15 % contamination with atmospheric O₂ rather than incorporation due to an exchange of oxygen from water into a potential reactive intermediate.

Figure 5.

The oxygen atom incorporated into the product originates from O₂. Mass spectra of the TMS-derivatives of products from single turnover reaction of S5HH using ¹⁸O₂ with (A) salicylate, (B) 5-CH₃-Sal, (C) 4-F-salicylate and (D) 5-F-salicylate. The mass species corresponds to the parent TMS derivative minus a methyl group. RT = GC retention time. Reaction conditions are given in Experimental Procedures.

Table 2.

Incorporation of ¹⁸O into the Reaction Products from Single Turnover Reactions of S5HH

Substratea	Product 18O (%)b
	18O2 / H216O	16O2 / H218O
salicylate	86.35	0
5-CH3-salicylate	85.15	0
4-F-salicylate	86.99	not tested
5-F-salicylate	86.89	not tested

^aReaction conditions are given in Experimental Procedures

^bGCMS detection

Kinetic Isotope Effects Differ for Aromatic Hydroxylation Versus Aryl-Methyl Substituent Hydroxylation.

Stopped-flow experiments were performed using deuterated substrates as mechanistic probes to determine whether any individual catalytic step up to and including electron transfer from the Rieske cluster exhibits a rate-limiting KIE (Figure 6). The reciprocal relaxation times were compared for the first (product forming) phases of single turnover reactions of unlabeled and deuterium labeled substrates. The salicylate reaction (1/τ₁ salicylate / 1/τ₁ salicylate-d₆) exhibited no KIE within error (KIE = 1.02 ± 0.04). In contrast, the 5-CH₃-Sal reaction (1/τ₁ 5-CH₃-Sal / 1/τ₁ 5-CD₃-Sal-d₈) monitored by oxidation of the Rieske cluster exhibited an inverse KIE (KIE = 0.86 ± 0.05). Contamination of 5-CD₃-salicylate-d₈ with salicylate would be one means to account for this unexpected observation, but the method used to synthesize the deuterated substrate by the supplier obviates this possibility and no salicylate was found experimentally in HPLC samples.

Figure 6.

Deuterium kinetic isotope effects observed when monitoring the Rieske electron transfer step in a single turnover reaction. (A) salicylate or salicylate-d₆ as the substrate and (B) 5-CH₃-salicylate or 5-CD₃-salicylate-d₈ as the substrate. Reduced S5HH (30 μM) was reacted with individual substrates (5 mM) and O₂ (900 μM) (Concentrations after mixing) at 4 °C in 50 mM HEPES, 100 mM NaCl, 5 % glycerol, pH 8.0. Time courses were fitted by nonlinear regression using a summed three exponential expression. The fastest kinetic phase in which product is formed was extracted from the data and plotted in each case.

Product Isotope Effects Differ for Aromatic Hydroxylation versus Ring Methyl Substituent Hydroxylation.

The reaction cycle of S5H is strongly rate limited by product release, thus determination of a product KIE by sampling during steady state turnover could not be employed. Single turnover experiments were conducted using equimolar mixtures of salicylate and salicylate-d₆ or 5-CH₃-Sal and 5-CD₃-Sal-d₈ as the substrate, and then the products extracted and analyzed by GCMS. While no product IE within error was observed for salicylate hydroxylation (see Discussion), methyl hydroxylation of 5-CH₃-Sal exhibited a value of 1.55 ± 0.04 (Figure 7 and Table 3).

Figure 7.

Product isotope effect determined from mass spectrometry analysis of products from single turnover reaction of S5HH with (A) equimolar salicylate and salicylate-d₆ or (B) equimolar 5-CH₃-salicylate and 5-CD₃-salicylate-d₈. RT= GC retention times. Reaction conditions are given in Experimental Procedures.

Table 3.

Kinetic and Product Isotope Effects for Single Turnover Reaction of S5HH with Salicylate and 5-Methyl-Salicylate.^a

Substrate	Isotope Effect
	Kinetic IE	Product IE
salicylate	1.02 ± 0.04	1.03 ± 0.03
5-methyl-salicylate	0.86 ± 0.05	1.55 ± 0.04

^aReaction conditions are given in Experimental Procedures

Computed Substrate Atomic Charge.

Previous studies of the mechanism of cis-diol forming benzoate 1,2-dioxygenase showed that the position of attack by the activated oxygen species occurred at the ring carbon with the largest atomic charge. The atomic charges at each carbon for the (unbound) substrates for S5H studied here were computed using the methods described in Experimental Procedures. Inspection of the active site of an S5HH-salicylate complex modeled from the crystal structure indicates that ring carbons 4 through 6 of the bound substrate will be closest to the iron (see below).⁵⁹ The atomic charges for the ring carbons 4 to 6 are listed in Table 1 and values for all carbons are shown in Figures S15–S17. The computed bond dissociation energies for specific bonds of interest are listed in Table S1 and the electrostatic potential surface maps are shown graphically in Figure 8. While the data supports reaction at the substrate carbon facing the mononuclear iron site with the largest atomic charge for S5H monooxygenase, the correlation with reaction rate is more complex than found for the cis-diol forming dioxygenase,⁸⁴ as discussed below.

Figure 8.

Electrostatic potential surface maps of (A) salicylate and (B) 4-F-salicylate, (C) 5-F-salicylate and (D) 6-F-salicylate (computed with GaussView 6.1).¹²²

DISCUSSION

The current study shows that a dedicated Rieske monooxygenase with no detectable dioxygenase activity exhibits dramatic sensitivity to the electronic structure of the substrate in the rate constant of electron transfer from the Rieske cluster to the reactive mononuclear iron site. In particular, the rate constant is greatly decreased by the inclusion of electron withdrawing fluorine substituents, mimicking findings for Rieske dioxygenases.^{78, 84} It is also confirmed by direct product analysis that the chemical reaction catalyzed by S5HH switches quantitatively from aromatic hydroxylation to methyl hydroxylation when a methyl substituent is placed in the normal position of hydroxyl incorporation on the aromatic ring. Unexpectedly, Rieske cluster oxidation kinetics and product formation exhibit opposite deuterium isotope effects during the methyl substituent hydroxylation reaction, offering new insight into the mechanism(s) of Rieske monooxygenation. These findings are discussed here in the context of structural, kinetic, and mechanistic studies of enzymes in the broad Rieske oxygenase family.

Correlation of Rieske Monooxygenase Activity with Substrate Atomic Charge in Aromatic Ring Hydroxylation.

The recent publication of the X-ray crystal structure of S5HH and computational docking of substrate in the active site by Li, Zhou and co-workers⁵⁹ (illustrated in Figure 9) suggested three important aspects of the active site: (i) substrate binds near, but not to, the mononuclear Fe(II), (ii) the substrate is fixed in place by hydrophobic interactions with the ring, electrostatic or hydrogen bonding of the carboxylate substituent with Arg323 and Gln316, and hydrogen bonding of salicylate C2-OH with Ser367, and (iii) the position of the ring places salicylate C5 closest to the iron (~ 4.0 Å) with C6 and C4 only slightly more distant (~ 4.1 Å). The S5HH crystal structure represents a static picture of the active site with the metals in a single oxidation state. However, during catalysis, both metal centers change their redox state,¹⁰⁸ and there are likely to be substantial changes to the relative positions of the metal centers and substrate as the reaction cycle progresses. The latter hypothesis is based on studies of three Rieske cis-diol forming dioxygenases with nearly identical structures to S5HH in terms of iron ligation and the cross-subunit boundary connection to the Rieske cluster. It was shown that the distance between the mononuclear iron and the closest substrate carbon is modulated by the oxidation state of the Rieske cluster.^{68, 80, 129, 130} The distance decreases as the mononuclear iron shifts toward the substrate when the cluster is in the oxidized state. The mechanism we proposed for S5HH (Scheme 2) suggests that reduction of the Rieske cluster shifts the mononuclear iron and the substrate apart, thereby opening sufficient space for O₂ to bind to the iron in an end-on conformation to form an Fe(III)-superoxo adduct. This binding mode allows electrophilic attack on the closest ring carbon, C5 in the case of S5HH, which is also the carbon with the most negative atomic charge of those facing the mononuclear iron (Table 1, Figures 8, 9 and S15–S17).

Figure 9.

Salicylate C4 is occluded from the iron in the active site model of an S5HH:salicylate (grey) substrate complex.⁵⁹ Ser367, which hydrogen bonds to the salicylate C2 hydroxyl, is behind salicylate in this orientation. His229 (magenta) lies between salicylate C4 and the iron. Salicylate C5 and C6 are not occluded. Distances between the selected salicylate carbons and the iron are show on the figure. Derived from PDB ID: 7C8Z.

The same S5H mechanism can be written for the salicylate and benzoate derivatives with fluorine substituents, as illustrated in Scheme 3. The fluorine withdraws electron density from the ring, slowing the attack by the metal-bound superoxo, but C5 still possesses the most negative charge of the ring carbons closest to the iron (C4, C5, and C6). Placement of the fluorine at C5 of salicylate is found to shift the position of hydroxylation to C6. Indeed, the atomic charge computations show that C5 is strongly deactivated while C4 and C6 now have the most negative charge (Table 1, Figures 8 and S16). In the model of the S5HH:salicylate substrate complex, C4 is adjacent to Phe232 and largely occluded from the oxygen binding site on the iron by the iron ligand His229 (Figure 9). This geometry may direct attack of the metal-superoxo to the unblocked C6. It is notable that aromatic ring hydroxylation also does not occur at C3 despite substantial negative charge, supportive of the docking model, where this position would also be distal from the mononuclear iron. The docking model is similarly supported by the observation that 3-CH₃-benzoate is hydroxylated only on the methyl substituent (Table 1) despite substantial negative charge on the C4-C6 ring carbons (Table 1, Figure S17). This finding suggests that the methyl can only be accommodated in the active site in the same orientation as the C5-methyl of 5-CH₃-Sal, and in this position, C6-C4 ring carbons are far from the iron.

Scheme 3. Proposed Mechanism of S5HH when a Fluorine Substituent is Present^a.

^aO-O bond cleavage may require intermediate arene oxide formation, although no fluorine NIH shift is observed.

Our current results support a model in which an Fe(III)-superoxo species attacks the ring at the closest carbon with the most negative charge as proposed for the Rieske dioxygenase class. However, we do not find the linear correlation between the log of the fastest reciprocal relaxation time and the atomic charge at the carbon where attack occurs as seen for the Rieske dioxygenase benzoate 1,2-dioxygenase using a range of fluorinated benzoates.⁸⁴ This finding presumably reflects effects of the active site to block access of the reactive species to specific ring carbons and/or control of bonding interactions with certain ring substituents. This hypothesis is consistent with the >20-fold difference in first phase 1/τ₁ values observed for 4-F- and 6-F-salicylate substrates without a change in atomic charge at C5 (Table 1). However, the true origin of the kinetic effects will require structural studies of the substrate complexes and more in-depth computations of the molecular mechanism.

Comparison of the Mechanisms of Rieske Mono- and Di- Oxygenases.

The question of where the proposed mechanisms for Rieske cis-dihydroxylation and aromatic hydroxylation diverge remains unresolved. Our current proposal is that both mechanisms follow the same course up through Rieske electron transfer. Computations for the dioxygenase mechanism in BZDO show that O-O bond cleavage in the resulting Fe(III)-peroxo-aryl intermediate would be very energetically favorable when electron and proton transfer are coupled.⁸² In the computed model, the cis-diol formation proceeds through an epoxide intermediate, which is commonly proposed for electrophilic aromatic hydroxylation (Scheme 4).^{88–90, 131} Such reactions sometimes involve an NIH shift, generally manifesting as an internal hydride transfer to an intermediate carbocation formed during epoxide ring opening. However, to yield cis-dihydrodiol products, this process would instead require transfer of the second oxygen from the O₂, nominally bound as Fe(III)-OH (Scheme 4). If the substrate orientation was unsuitable for this capture, then the intermediate might decay by the normal NIH shift hydride transfer to yield a monohydroxylation product. Such critical dependence on substrate positioning would account for the observation that some Rieske dioxygenases can also catalyze monooxygenase reactions with alternative substrates.^{5, 75, 97}

Scheme 4.

A Proposal for a Common Intermediate for Rieske Monooxygenase and Dioxygenase Reactions

It is also possible that ring hydroxylation in the case of the dedicated Rieske monooxygenase involves oxygen transfer without intermediate epoxide formation as proposed for some oxygenases,^{90, 132} and consequently, no NIH shift would be expected. While the presence or absence of a deuterium NIH shift has not been evaluated for a Rieske monooxygenase, the current results show that there is no fluorine shift or elimination in the S5H reaction. In contrast, in other heme and non-heme iron oxygenases, fluorine shift or defluorination is observed for aromatic substrates that are proposed to form epoxide intermediates including cytochrome P450, 2-halobenzoate-1,2-dioxygenase and tyrosine hydroxylase.^{90, 133–140}

Comparison of Isotope Effects for Ring and Methyl Substituent Hydroxylation.

The studies reported here demonstrate that O₂ is the only source of oxygen for both aromatic ring and methyl substituent hydroxylations. Thus, S5HH functions as an oxygenase in each case. In general, oxygenase reactions with aromatic rings do not involve direct hydrogen abstraction due to the strength of the C-H bonds involved (BDFE = 101–106 kcal/mol., Table S1),¹⁴¹ and thus, a large primary deuterium isotope is not expected. A smaller secondary isotope effect may be observed for a mechanism in which an electrophilic Fe(III)-superoxo intermediate directly attacks the aromatic ring (see, for example, Schemes 2 and 3) due to changes in zero-point energies for C-H versus C-D as the aromatic carbon changes hybridization. When observed, this isotope effect is inverse, indicating that the deuterated substrate reacts more rapidly.¹⁴² As shown in Figure 6, no KIE is observed when monitoring the time course of Rieske oxidation during the ring hydroxylation of salicylate. However, small KIE effects are difficult to observe in transient experiments of very fast reactions due to signal-to-noise limitations. In contrast, an inverse KIE value is observed when monitoring the time course of the relatively slow Rieske oxidation during the methyl substituent hydroxylation reaction when using 5-CH₃-Sal as the substrate. This result is unexpected if it is associated in any direct way with breaking the C-H bond of the methyl substituent. The strength of this bond is much less than that of the aromatic C-H bond (BDFE = 81.4 kcal/mol., Table S1), but its cleavage still requires a powerful oxidizing intermediate such as a Fe(IV)-oxo, Fe(V)-oxo, or perhaps Fe(III)-hydroperoxo.^{94, 141, 143} Direct cleavage of the methyl C-H bond by these species would yield a normal rather than inverse KIE.¹⁴⁴

One intriguing hypothesis for the origin of the inverse KIE on Rieske oxidation is illustrated in Scheme 5. It is proposed that methyl substituent hydroxylation reaction must begin in the same way as aromatic ring hydroxylation. The electron from the Rieske cluster is required to generate a reactive species, but it cannot be transferred to the already reduced Fe(II) site. Binding of O₂ to the iron begins its oxidation process, but the iron only becomes fully oxidized and able to accept the Rieske electron after attack on the aromatic ring by the Fe(III)-superoxo intermediate. Accordingly, the ring carbons at C4, C5, and C6 of 5-CH₃-Sal retain the net negative charge required for such an attack (Table 1, Figure S15). As in the case of aromatic ring hydroxylation, this reaction is rate-limiting in the sequence of reactions leading to electron transfer from the Rieske cluster. It is interesting to note that no Rieske oxidation above that expected for autooxidation occurs when 5-CF₃-Sal is bound in the active site (Figure 4). Accordingly, computation of the atomic charges for 5-CF₃-Sal shows that C4 and C6 are strongly deactivated for attack by the putative Fe(III)-superoxo intermediate, while the bulky methyl substituent prevents attack at C5 (Table 1, Figure S15).

Scheme 5. A Potential Mechanism for Methyl Substituent Hydroxylation by S5HH^a.

^aThe Fe(V)=O intermediate shown is one of several possible activated oxygen species at this stage of the reaction cycle.

Understanding the basis for the remaining steps in the mechanism proposed in Scheme 5 requires examination of the methods used to determine the product isotope effect. The overall reaction of S5H is strongly rate-limited by product dissociation.¹⁰⁸ Consequently, methods to determine isotope effects such as reaction sampling during steady-state turnover are complex. Similarly, the very rapid product formation after Rieske oxidation during ring hydroxylation does not permit sampling during the course of a single turnover. Such sampling might be possible during the single turnover methyl-substituent reaction (Figure 3B), but the close correlation between product formation and Rieske oxidation suggests a very short lifetime for any intermediate that might be definitive. Indeed, our initial attempts to find an intermediate using rapid trapping techniques have been unsuccessful. Unfortunately, the method employed here of sampling at the end of a single turnover reaction is not definitive for the salicylate aromatic ring hydroxylation reaction, because no product isotope effect is observed when equal concentrations of labeled and unlabeled substrate are present at the beginning of the reaction. Lack of an observed product isotope effect could mean that the presence of deuterium actually has no effect on the rate of product formation, as expected when the deuterium is bound to ring carbons. Alternatively, there may be an appreciable isotope effect from deuterium which is masked by a failure of the substrate to dissociate once bound. In a single turnover reaction, a non-dissociating substrate will result in an isotopic distribution in the products which mirrors that in the substrate mixture. In contrast, the observation of a deuterium isotope effect in the case of aromatic methyl-substituent hydroxylation in a single turnover reaction has substantial ramifications for the reaction chemistry. First, it indicates that there is a normal product isotope effect that is at least as large as that observed, supportive of direct hydrogen abstraction. Second, it requires that the substrates be able to exchange from solution into the active site up to the point of irreversible reaction, presumably hydrogen abstraction from the substituent methyl group. If this were not the case, then the isotope ratio in the product would be the same as that in the substrate in solution at the start of the reaction, i.e. 1:1. Third, the substrate exchange must be fast compared with product formation. Fourth, the apparent correlation between product formation and 1/τ₁ of the Rieske oxidation time course (when a two-electron reduced oxygen species is formed that is capable of substrate hydroxylation) suggests that product formation must be fast compared with the steps that rate limit the observed electron transfer (Figure 3B). Note that the rate constant for the electron transfer itself through bonds across 12 Å would exceed any of the rate constants reported here by many orders of magnitude.^{82, 145} This fact makes it unlikely that a simple change in redox potential of the mononuclear iron caused by an environmental change due to the substitution of a fluorine on the substrate ring would alter the observed electron transfer rate constant.

In the case of S5H, substrate binding at the beginning of the reaction cycle is very fast relative to observed Rieske oxidation,¹⁰⁸ supporting the notion that exchange in the short time period between electron transfer and irreversible hydrogen atom abstraction may be possible. Additionally, the hydroxyl radical rebound reaction in non-heme oxygenases is known to be very fast, typically occurring six or more orders of magnitude faster than the 1/τ₁ for Rieske cluster oxidation observed for S5HH.^{95, 146, 147}

The requirement for substrate exchange after O₂ activation, but before irreversible reaction with the substrate implies that the proposed preceding Fe(III)-peroxo-substrate intermediate must be reversibly formed, at least for slow substrates like 5-CH₃-Sal. For example, the stability of this intermediate might be affected due to perturbation of its position in the active site by the bulky methyl substituent. Similar shifts in precise alignment and reaction outcome are observed in other oxygenases after minor changes in substrate structure.¹⁴⁸ Reversal of the ring C-O bond formation without reverse of Rieske electron transfer against a steep redox potential gradient^{124, 125} would yield a comparatively stable Fe(III)-(hydro)peroxo intermediate. Moreover, the shift of the mononuclear iron toward the substrate that is associated with an oxidation of the Rieske cluster would be maintained.^{68, 80, 129, 130} In the case of BZDO, the H₂O₂ complex of the fully oxidized enzyme results in side-on rather than end-on H₂O₂ binding.⁸⁵ A shift from end-on to side-on bound (hydro)peroxo in S5HH might also be promoted by the bulk of the methyl substituent in the space near the iron (Scheme 5). This type of intermediate is predicted computationally to progress with almost no energy barrier to a reactive HO-Fe(V)=O species.⁸² The HO-Fe(V)=O species, now forced close to the aromatic methyl substituent by the shift in the mononuclear iron position, could readily abstract a methyl hydrogen atom from the substrate.¹¹³ Hydroxylation chemistry would then proceed via the well characterized hydroxyl radical rebound mechanism observed for P450 and methane monooxygenase oxygenases.^{146, 149, 150} Hydrogen atom abstraction by a high-valent iron-oxo species with substrate radical intermediate formation is consistent with radical clock experiments in which the Rieske dioxygenase naphthalene 1,2-dioxygenase catalyzes monooxygenation of probe molecules.⁹⁵ In this reaction, a radical intermediate with a lifetime of ~18 ns was formed prior to rearrangement of the probe molecule to diagnostic products.

Alternative Proposals for Rieske Oxygenase Mechanisms.

The proposal here for shared initial steps in the O₂ activation and substrate reactions of Rieske dioxygenases and monooxygenases (compare Schemes 3, 4 and 5) can be examined in the broader context of other mechanisms proposed for the Rieske oxygenase family. We have previously discussed many of the potential mechanisms involving high valent, peroxo, or dioxetane intermediates formed after the electron from Rieske cluster has been transferred to the mononuclear iron site.^{84, 108} These types of mechanisms are largely ruled out for the initial attack by an activated oxygen species on the substrate by the consistent observation that the type and electronic character of substrate dictate the rate constant for transfer of the Rieske electron.^{78, 84, 108} Nevertheless, a recent detailed study of the reaction of the Rieske dioxygenase nitrobenzene dioxygenase (NBDO) by Hofstetter and co-workers, used ¹⁸O₂ to show that the ¹⁶O/¹⁸O-kinetic isotope effects (¹⁸O-KIE) for a range of substrates were both similar to each other and approached the ¹⁸O-equilibrium isotope effect (¹⁸O-EIE) expected for an Fe(III)-(hydro)-peroxo species participating in the rate-determining step of the reaction with substrate.⁸¹ The ¹⁸O-KIE must be less than or equal to the ¹⁸O-EIE for any reaction,^151–153 and the observed ¹⁸O-KIE for the NBDO reactions significantly exceeds the literature value for the ¹⁸O-EIE value for the Fe(III)-superoxo species found in hemoglobin and myoglobin. While the ¹⁸O-KIE method is very sensitive, it is limited in the case of Rieske oxygenases by: (i) the presence of three parallel substrate-dependent Rieske cluster oxidation reactions, only one of which results in product formation,^{84, 108} and (ii) the unknown ¹⁸O-EIE for the type of Fe(III)-superoxo species proposed for Rieske oxygenases, which is coupled through bonds to the reduced Rieske cluster.^{59, 154} Also, the ¹⁸O-KIE is reflective of the rate determining step of the relative k_cat/K_m values for the reactions using ¹⁸O₂ and ¹⁶O₂. The k_cat/K_m value is composed of rate constants up to and including the first irreversible step in the reaction sequence, which may or may not be rate determining.¹⁵⁵ Our study suggests that the formation of the Fe(III)-peroxo-bridged aryl intermediate is reversible, at least for some slow substrates. Thus, the actual first irreversible step would occur later in the cycle, and the kinetics of the Fe(III)-peroxo-bridged aryl intermediate formation and decay would be reflected, at least in part, in the ¹⁸O-KIE. The ¹⁸O-EIE for such a complex has not been determined, but that for the similar Fe(III)-peroxo bridged-tert-butyl complex is known and has a value^{153, 156, 157} nearly identical to that of the Fe(III)-(hydro)-peroxo proposed as the alternative initial reactive species.^{94, 128, 143}

While attack on the substrate by an Fe(III)-superoxo intermediate prior to Rieske cluster oxidation appears to provide the most direct explanation for the observed single turnover kinetics, the downstream reactions are less well defined. Electron transfer to the proposed Fe(III)-peroxo-substrate radical intermediate would yield a species in which the O-O bond could be readily broken by any of several routes. Indeed, different routes may be employed by the Rieske oxygenases in response to the metabolic challenges presented.

CONCLUSIONS

The current results support a mechanism for Rieske monooxygenases that can proceed along more than one pathway depending upon the type and electronic properties of the substrate. For S5HH with both metal centers reduced, the natural substrate, salicylate, can bind in the proper alignment to allow rapid O₂ binding to the mononuclear iron to form an Fe(III)-superoxo intermediate. Salicylate also has the correct charge distribution to be rapidly attacked by this activated oxygen species, ultimately creating a fully oxidized mononuclear iron that can accept an electron from the Rieske cluster to complete the oxygen activation process. In the broader context of regulation in oxygenase reactions, these steps insure that O₂ is only activated: (i) after substrate is correctly bound, (ii) O₂ is bound to the mononuclear iron adjacent to the substrate, and (iii) both electrons required to complete the reaction are present in the enzyme. Once both electrons are transferred to the bridged Fe(III)-peroxo-substrate complex, the hydroxylation reaction could proceed by the well-established steps of electrophilic aromatic substitution. However, the current results introduce two potential variations. First, the lack of fluorine shift or elimination when present at or adjacent to the position of hydroxylation raises the possibility of a mechanism without formation of the arene-oxide common to electrophilic aromatic substitution. Second, the substrate exchange after oxygen activation required to explain the observed product isotope effect for substituent hydroxylation of 5-CH₃-Sal suggests that a Fe(III)-(hydro)peroxo intermediate, not bound to the substrate, may be formed after the initial Fe(III)-peroxo-aryl-radical intermediate. Such an intermediate might directly reattack the substrate for hydroxylation or dioxygenation akin to the mechanism favored in some past studies.^{94, 128, 143} Alternatively, it might convert to an HO-Fe(V)=O species capable of chemistry requiring a more energetic species. The route taken by any given Rieske oxygenase after the initial substrate-coupled activation steps is likely to be controlled by the precise electronic properties and positioning of the substrate, and ultimately, the kinetics of reaction with substrate. The reciprocal relaxation times of these reactions have been shown to differ by over 2 orders of magnitude even within a single enzyme. Given their conserved active site structure, it seems likely that all Rieske oxygenases have the potential for this multi-pathway reactivity, which may account for the remarkable diversity of chemistry that they exhibit.