Heme-thiolate ferryl of aromatic peroxygenase is basic and reactive

Significance

The heme-thiolate peroxygenase of Agrocybe aegerita is a remarkably capable biocatalyst and a mechanistic analog of cytochrome P450. The stability of this fungal protein has provided a rare opportunity to study P450-like C−H hydroxylation in a novel and unrelated enzyme. Both APO-I and APO-II have been generated, and their redox potentials have been determined. The ferryl species Cys−S−Fe^IV−OH (APO-II) has been generated cleanly via reduction of the corresponding APO-I and a basic pK_a revealed for the Cys−S−Fe^IV−OH ⇄ Cys−S−Fe^IV=O equilibrium. Most significantly, APO-II displays surprisingly high reactivity toward benzylic C−H (bond-dissociation energy 80−86 kcal/mol) and phenolic substrates with rate constants orders of magnitude larger than those of typical peroxidases or model compounds due to the basic ferryl.

Abstract

A kinetic and spectroscopic characterization of the ferryl intermediate (APO-II) from APO, the heme-thiolate peroxygenase from Agrocybe aegerita, is described. APO-II was generated by reaction of the ferric enzyme with metachloroperoxybenzoic acid in the presence of nitroxyl radicals and detected with the use of rapid-mixing stopped-flow UV-visible (UV-vis) spectroscopy. The nitroxyl radicals served as selective reductants of APO-I, reacting only slowly with APO-II. APO-II displayed a split Soret UV-vis spectrum (370 nm and 428 nm) characteristic of thiolate ligation. Rapid-mixing, pH-jump spectrophotometry revealed a basic pK_a of 10.0 for the Fe^IV−O−H of APO-II, indicating that APO-II is protonated under typical turnover conditions. Kinetic characterization showed that APO-II is unusually reactive toward a panel of benzylic C−H and phenolic substrates, with second-order rate constants for C−H and O−H bond scission in the range of 10–10⁷ M⁻¹⋅s⁻¹. Our results demonstrate the important role of the axial cysteine ligand in increasing the proton affinity of the ferryl oxygen of APO intermediates, thus providing additional driving force for C−H and O−H bond scission.

Nature has evolved various and remarkably efficient strategies for the selective oxygenation of even the most unreactive C−H bonds in hydrocarbons and other small molecules (1, 2). Copper-containing methane monooxygenases (MMO) (3), nonheme, diiron hydroxylases such as soluble MMO (4, 5), and the alkane ω-hydroxylases, AlkB (6, 7), allow microorganisms to grow on petroleum and natural gas as their sole sources of carbon. The widely distributed heme-thiolate monooxygenases, such as cytochrome P450 (CYP), serve similar roles in catalyzing C−H hydroxylation reaction. CYP enzymes also participate in the primary pathways for oxidative steroid and prostaglandin biosynthesis as well as phase I drug metabolism (8–12). For example, 20 isoforms of CYP with significant physiological functions exist in Mycobacterium tuberculosis, making them potential drug targets (13). Additionally, C−H bond oxidation processes are environmentally significant, particularly after oil spills such as the Deepwater Horizon event in 2010 (1).

CYP has been called the Rosetta Stone of iron-containing oxygenases (8). Decades of research have created a rich tapestry that has intertwined structural, spectroscopic, mechanistic, computational, genetic, metabolic, and chemical modeling approaches toward a deep understanding of such an important oxygenation system (9, 14–17). The central paradigm of CYP oxygen activation is recognized to involve the formation of ferryl intermediates, Fe^IV=O. Oxoiron(IV) porphyrin cation radicals (compound I) and oxoiron(IV) porphyrins (compound II) have been observed in CYP (18) as well as classic peroxidases (19), chloroperoxidase (CPO) (20), and catalases (21) and in a variety of model porphyrin systems (22, 23). Variations in the reactivities of these ferryl species and the roles of axial ligation and heme environment continue to be vigorously studied and debated.

Aromatic peroxygenases (APO), also referred to as unspecific peroxygenases (UPO, EC 1.11.2.1), are a newly discovered and large superfamily of heme-thiolate proteins from fungal sources (24). These proteins are unrelated to CYP enzymes, according to their amino acid sequences, and are only distantly related to CPO, with about 30% sequence similarity and a similar tertiary structure (25). APO proteins have shown high activity for the oxygenation of aliphatic and aromatic hydrocarbons and a variety of other organic substrates, in sharp contrast to CPO (26). These APO proteins are unusually stable, heavily glycosylated, extracellular proteins that apparently serve to mobilize food sources for the growing organisms through oxidative degradation and to detoxify harmful compounds in their microenvironment. As such, they are proving to be unusually efficient biocatalysts (27). Numerous drug compounds are efficiently oxidized to products that are often the same as those from mammalian CYP transformations (28). These properties have also afforded a rare and revealing opportunity to study the reaction steps in C−H bond hydroxylation events with a heme-thiolate protein that is not a CYP, but which may function in analogous ways.

We have recently shown that the APO from Agrocybe aegerita (AaeAPO) forms an oxoiron(IV) porphyrin radical cation (APO-I) that can be detected by UV-visible (UV-vis) spectroscopy (29). This intermediate was shown to be highly competent for the hydroxylation of even strong C−H bonds. Further, APO-I reacts with chloride and bromide ions rapidly and reversibly, allowing a direct determination of the thermodynamics of oxygen transfer by APO-I (E’ = 1.2 V) (30). Here, we report a previously unidentified method of generating the elusive ferryl state, APO-II, directly from APO-I over a wide range of pH. We are able to determine that APO-II is protonated at physiological pH, Cys−S−Fe^IV−OH, and has a basic pK_a of 10.0. Further, kinetic studies of the reduction of APO-II by a panel of benzylic C−H and phenolic substrates has revealed surprisingly high chemical reactivity for this intermediate.

Direct Reductive Generation of APO-II from APO-I with Nitroxyl Radicals

APO-II is the key intermediate in the reaction pathway for C−H hydroxylation by APO-I. Since APO-I can be generated in good yield via metachloroperoxybenzoic acid (mCPBA) oxidation of the resting ferric protein (29), we sought a method to produce APO-II via one-electron reduction. We considered several nitroxyl radicals as chemical reagents to reduce APO-I by one electron to APO-II (Fig. 1A). The oxidation of nitroxyl radicals, which have potentials in the range of 0.8–1.0 V, is known to produce oxoammonium species [O=N⁺(R)₂] (31). The oxidation potentials of 3-carboxy-PROXYL and 4-carboxy-TEMPO are close to this range (Fig. S1), and similar to our estimate of the APO-II/ferric APO couple (∼0.8 V vs. NHE at pH 7) (30). Accordingly, we anticipated that these water-soluble nitroxyl radicals might be able to reduce APO-I readily but would not reduce APO-II efficiently. Furthermore, since cyclohexane carboxylic acid was found to be an excellent substrate for APO (29), the similar molecular topographies of 4-carboxy-TEMPO and 3-carboxy-PROXYL suggested that they would fit into the relatively small, conical active site of APO (25). We tested the reaction of 3-carboxy-PROXYL with the well-studied enzyme CPO in a single-turnover experiment. When 4 µM CPO was mixed with a solution containing 2 eq of mCPBA and 10 mM of 3-carboxy-PROXYL, CPO-II was readily detected, displaying a UV-vis spectrum that was the same as those generated by previous methods (absorptions at 371 nm, 438 nm, 541 nm, and 571 nm; see Fig. S2) (20, 32). Notably, CPO-II generated by this nitroxyl method persisted in solution for more than a minute. One explanation of this persistence is the fast conversion of ferric CPO to CPO-I by excess mCPBA and the fast conversion from CPO-I to CPO-II by excess nitroxyl radical. The CPO-II spectrum returned to that of ferric CPO after the depletion of mCPBA.

Fig. 1.

(A) Reaction scheme for in situ reduction of APO-I to APO-II by the nitroxyl radical 3-carboxy-PROXYL, and subsequent reaction with substrates. (B) UV-vis spectra of 10 μM ferric APO (black), APO-II (red) obtained from mCPBA oxidation of APO in the presence of 20 mM 3-carboxy-PROXYL at pH 7, and, for comparison, APO-I (green) obtained as previously described (29). (C) UV-vis transients of APO-II (at 0.45 s) and its decay process (at 0.65 s, 0.75 s, 0.85 s, 0.95 s, 1.15 s and 1.35 s) to ferric state (at 2 s) observed upon 1:1 mixing of 10 μM of ferric APO with a solution containing 50 μM of mCPBA and 10 mM of 3-carboxy-PROXYL at pH 7.0, 4 °C. (Inset) The time course for APO-II formation and decay under these conditions.

The reaction of ferric APO (10 µM) with mCPBA (50 µM) and excess 3-carboxy-PROXYL (10 mM), in a single-mixing stopped-flow experiment, produced a new and distinct UV-vis spectrum (Fig. 1B) similar to that of CPO-II. We assign this spectrum to APO-II based on the characteristic split Soret band (370 nm and 428 nm) and two Q bands centered at 535 nm and 567 nm, slightly blue shifted from those of the ferric state (18). APO-I is the dominant intermediate by reacting ferric APO with an oxidant only, such as mCPBA (29, 30). However, with the presence of nitroxyl radical, the accumulation of APO-I was not apparent. These results indicate that the one-electron reduction of APO-I to APO-II by nitroxyl radicals is fast and efficient. Additionally, APO-II was also generated with a double-mixing experiment when APO-I was initially accumulated at the first mixing step and mixed with nitroxyl radicals at the second step, as shown in Fig. S3. This experiment confirmed that APO-II was formed by a direct reduction of APO-I. The kinetic traces showed that APO-II reached its maximal conversion after 0.2 s, persisted for about 0.5 s, and then decayed back to the ferric state. (Fig. 1C) The accumulation of APO-II was optimal at pH 7.0 and was nearly fully formed. The spontaneous decay rate of APO-II (k₀) under these conditions could be estimated by kinetic modeling to be ∼1 s⁻¹. This brief steady-state appearance of APO-II is likely to be due to a few turnovers, with the final turnover showing the appearance of ferric enzyme. The simulation also indicated that all of the mCPBA would have reacted at the end of the plateau region (∼0.75 s after mixing).

Determination of the Ferryl OH pK_a for APO-II

Because APO is a highly robust, extracellular protein, its UV-vis spectrum could be obtained over a wide range of pH. The far-UV CD of ferric APO was found to be invariant from pH 7 to pH 12, indicative of minimal conformational changes (see Fig. S4). In marked contrast, CPO is known to have an alkaline transition at pH > 8 (33). This pH tolerance allowed the generation and spectral observation of APO-II under basic conditions. The effect of pH on APO-II was examined using a pH-jump, double-mixing stopped-flow technique (see Fig. S5) (34). APO-II was produced in the first push with mCPBA in the presence of 3-carboxy-PROXYL as described above using 25 mM NaP buffer at pH 7.0. After a 200-ms aging time, APO-II was mixed 1:1 with a 200-mM buffer solution with a pH in the range of 7.0–12.

Combining and plotting the family of APO-II spectra obtained via pH jump to various pH values revealed a clear spectral transition (Fig. 2B). The intensities of the split Soret bands (428 nm/370 nm) changed, the Q bands were slightly red shifted at higher pH, and a single set of multiple isosbestic points was observed, suggesting a single protonation event. All of the APO-II spectra have split Soret bands, indicating that the cysteine sulfur remained ligated to the heme iron throughout. Since we have shown that APO does not appreciably change conformation in this pH range, and showed no heme bleaching under the working conditions, we assign this spectral change to the deprotonation of Fe^IV−O−H of APO-II at high pH to form a normal ferryl species (Fig. 2A). Titration curves were obtained by plotting absorbances at 455 nm and 569 nm vs. pH as shown in Fig. 2C to afford a pK_a of 10.0 ± 0.1 for the hydroxyl group in APO-II.

Fig. 2.

(A) The equilibrium between protonated APO-II and deprotonated APO-II. (B) UV-vis spectra of APO-II at pH 7–12. (Inset) 4× enlargement of the same UV-vis spectra from 480 nm to 640 nm. (C) The pH titration curves for APO-II derived from the spectral data in B at 455 nm and 569 nm. The plots are fitted to the Henderson–Hasselbalch equation to calculate the pK_a = 10.0 ± 0.1.

Green and coworkers have shown that the compound II intermediates of CYP and CPO are protonated (18, 35). CPO-II was estimated to have a pK_a ≥ 8.2 based on the fact that the UV-vis spectrum of CPO-II showed no changes from pH 3 to pH 7. The pK_a of CYP-II was measured to be 11.9. High electron density on the ferryl oxygen is thought to arise from the push effect of the trans-thiolate ligation (14, 36). By contrast, HRP-II, cytochrome c peroxidase (37, 38), and myoglobin (39), which have neutral histidines ligated to the heme, retain the ferryl oxo form. Resonance Raman measurements, X-ray absorption spectroscopy, Mössbauer spectroscopy, and X-ray crystallographic results suggest that the pK_a is ≤ 4 for these ferryl states.

The basic pK_a indicates that APO-II is in the ferryl hydroxide form under normal catalytic conditions. Thus, C−H hydrogen abstraction by APO-I will proceed to a protonated APO-II in a proton-coupled electron transfer event in which one electron will be transferred to the porphyrin radical cation of APO-I (29). It is at this stage that the incipient substrate radical can either rebound to form product alcohols or rearrange in the enzyme active site as has been observed (26).

The basic pK_a of APO-II has important implications in C−H activation chemistry by providing an additional driving force for hydrogen atom abstraction [hydrogen atom transfer (HAT)] by compound I. As shown in Eq. 1, pK_a(II) and the one-electron reduction potential of compound I, E^o(I), determine the dissociation energy, D (O−H), of the newly formed FeO−H bond of compound II (40–42). Thus, both pK_a(II) and E^o(I) are important in determining the C−H bond activation capability of APO-I. With the same E^o(I), a higher ferryl pK_a would make D (O−H) stronger. Accordingly, hydrogen atom abstraction by Fe=O would become more energetically favored without increasing the redox potential, E^o(I) (43). Such delicate redox potential and pK_a tuning may be essential to allow such difficult C−H scission reactions to proceed within a protein scaffold (18, 36).

$$D\left(\text{O}-\text{H}\right)\mathrm{=}nF{E}^{\text{o}}\left(\text{I}\right)\mathrm{+}\mathit{2}\mathit{.3}RT\text{p}{K}_{\text{a}}\left(\text{II}\right)\mathrm{+}57$$ [1]

Kinetic Characterization of APO-II Toward C−H and O−H Substrates

The accumulation and slow decay of APO-II made it possible to investigate its reactions with a panel of substrates. High reactivity was observed both for phenols and substrates with relatively weak, benzylic C−H bonds (80–86 kcal/mol). Data were collected for a double-mixing, stopped-flow experiment wherein APO-II was prepared in the first push with a solution containing mCPBA and 3-carboxy-PROXYL as described above. Various concentrations of substrate were then added in the second push. The observed APO-II decay rates (k_obs) were obtained by fitting the disappearance of APO-II, measured at 441 nm, according to the reaction scheme in Fig. 3A. Plots of k_obs vs. substrate concentrations gave linear relationships. The results are summarized in Table S1. We also examined substituted phenols, which are typical substrates for peroxidases. Data are summarized in Table S2. The second-order rate constants, k₂, for APO-I were obtained by monitoring the Soret peak at 361 nm. APO-II was monitored as above. The kinetics were measured at pH 7.0, below the pK_a of all of the phenolic substrates.

Fig. 3.

(A) The generation of APO-II and its reaction with substrates (S). (B) Plot of log k₂’ vs. substrate C–H BDE for APO-II at pH 7.0 and APO-I at pH 5.0. (C) Plot of log k₂ vs. phenolic σ⁺ for both APO-I (R² = 0.6) and APO-II (R² = 0.93) at pH 7.0.

It is interesting to compare the reactivity of APO-II with those of APO-I (Fig. 3B). For benzylic C−H substrates, apparent second-order rate constants, k₂’, were corrected based on the number of equivalent protons in the substrates in these hydrogen atom transfer reactions. Overall, the reaction rates for APO-II are remarkably fast, although much slower than those of APO-I (29). For example, k₂’ for APO-II is 5 and 3.5 orders of magnitude smaller than those of APO-I for 4-ethylbenzoic acid and 4-isoprophylbenzoic acid, respectively. The observed differences in reactivity indicate that APO-II is a milder oxidant than APO-I, consistent with our estimation of the redox potentials of these intermediates, E_cpdI/cpdII = 1.4 V, E_cpdII/ferric = 0.8 V at pH 7.0 (30), and our estimate that the O−H bond-dissociation energy (BDE) of Cys−S−Fe^IV−O−H is ∼100 kcal/mol (29).

The high reactivity of APO-II observed here is very surprising. Chloroperoxidase and HRP compound II are known to be unreactive toward C−H bonds. It is revealing to compare the reactivity of APO-II to synthetic Fe^IV oxo porphyrin model complexes. For example, the rate constant for the reaction of isopropylbenzoic acid with APO-II is 1,800-fold larger than that for oxoFe^IV-4-TMPyP (44), and that for the pentafluorophenyl porphyrin, (TPFPP)Fe^IV=O, reacting with triphenylmethane (C−H BDE of 81 kcal/mol) has been reported to be only 0.85 M⁻¹⋅s⁻¹ (45). This rate, as well as that of a recently reported Cu(III)-hydroxide at 0.3 M⁻¹⋅s⁻¹ (46), are three to four orders of magnitude slower than that of APO-II reacting with fluorene-4-carboxylic acid (C−H BDE of 80 kcal/mol). APO-II reacts extraordinarily fast with phenols in comparison with typical peroxidases. For example, myeloperoxidase compound II is known to react with phenol with a rate constant of 1.6 × 10⁴ M⁻¹⋅s⁻¹ (21), while, for APO-II, it is nearly 10⁸ M⁻¹⋅s⁻¹. We attribute the high reactivity of APO-II with these substrates to the high basicity of Cys−S−Fe^III−O−H despite the relatively modest Fe^IV/Fe^III redox potential of only 0.8 V. The pK_a of aqua ferric APO must be very high since we were unable to see any spectroscopic changes in the heme-thiolate spectrum over the entire accessible range of pH.

A correlation of log k₂’ with the substrate C−H BDE shows a linear Brønsted–Evans–Polanyi relationship with a slope (−0.4) that is much steeper than that of APO-I. For APO-I, the reaction rates are nearly insensitive to the C−H BDE below 90 kcal/mol (Fig. 3B). The sensitivity of the observed rates to the substrate C−H BDE for APO-II is similar to that of APO-I reacting with strong C−H bond substrates (29), suggesting that APO-II has a nearly symmetrical transition state for hydrogen atom transfer for these weaker C−H bonds (47).

The observed rates of O−H bond scission gave a good linear correlation with Hammett σ⁺ values for these substituted phenols (48). Plots of log k₂ vs. σ⁺ afforded ρ values of −0.67 and −3.3 for APO-I and APO-II, respectively (Fig. 3C). This large difference of ρ values suggests that the degree of charge separation in the transition states for these HAT reactions is different for APO-I and APO-II. (Fig. 4) APO-I reacted very rapidly with the entire range of phenols with little sensitivity to either the O−H BDE or the Hammett σ⁺. The relatively small negative ρ value for APO-I agrees well with the mechanism of hydrogen atom abstraction (49). However, for APO-II, the large negative ρ value implies that APO-II is sensitive to the electron deficiency on the oxygen and there is a significant positive charge developed on the substrate in the transition state.

Fig. 4.

Depiction of C−H hydrogen atom abstraction by APO-I and APO-II.

In conclusion, APO-II has been generated by a one-electron reduction of APO-I with nitroxyl radicals. Changes in the UV-vis spectra of APO-II under basic conditions have revealed a basic pK_a of 10.0 for APO-II. Kinetic characterization of APO-II showed that it has surprisingly high reactivity toward substrates with benzylic C−H bonds with BDE up to 85 kcal/mol, representing the first example of the C−H activation by an enzymatic compound II intermediate. The high reactivity of APO-II compared with other ferryl species such as CPO-II, HRP-II, and synthetic Fe^IV oxo porphyrin complexes probably arises from the effect of the axial thiolate ligand on the pK_a of the resting aqua ferric enzyme. These results also shed light on the analogous C−H bond scission reactions of CYPs and set the stage for the further investigations of these APO proteins toward the design of new biocatalysts.

Methods

Reagents.

Wild-type peroxygenase from A. aegerita (isoform II, pI 5.6, 46 kDa) was produced in bioreactors with a soybean flour suspension as growth substrate, and purified as described previously (24). The enzyme preparation was homogeneous by SDS/PAGE, and exhibited an A_{418 nm}/A_{280 nm} ratio of 1.7. The specific activity of the peroxygenase was 59 U⋅mg⁻¹, where 1 U represents the oxidation of 1 μmol of 3,4-dimethoxybenzyl alcohol to 3,4-dimethoxybenzaldehyde in 1 min at RT. Chloroperoxidase was purchased from Bio-Research Products, Inc., with A_{400 nm}/A_{280 nm} > 1.3, 1,296 U⋅mg⁻¹, where 1 unit represents the amount of CPO catalyzing the formation of 1 μmol of dichlorodimedone from monochlorodimedone in 1 min at 25 °C, pH 2.75, 0.02 M KCl, 2 mM H₂O₂. All chemicals were of the best available purity from Aldrich. Chemicals with purity lower than 99% were recrystallized before use.

Instruments.

UV-Vis spectral measurements were performed on a Hewlett-Packard 8453 diode array spectrophotometer at room temperature. Stopped-flow experiments were carried out on a Hi-Tech SF-61 DX2 double-mixing instrument with a 1-cm path length equipped with an ISOTEMP 3013 D thermostat bath. All concentrations reported are the final concentrations. Diode array experiments were blanked with the same concentrations of nitroxyl radicals used in experiments. All kinetics experiments were carried out at 4 °C. The data were analyzed using Kinetic Studio from Hi-Tech. Cyclic voltammetry was done with 100 mM potassium phosphate buffer, Pt electrode, Ag/AgCl KCl 3M as reference with a sweep rate of 0.1 V⋅s⁻¹. E^o = (E_pc + E_pa)/2. The ¹H NMR spectra were recorded on a 500-MHz Bruker Avance II spectrometer. CD spectra were recorded using an Applied Photophysics Chirascan circular dichroism spectrometer. The molar ellipticity, [θ] (millidegrees per square centimeter per decimole) is expressed. Cuvette with path length of 1 mm was used to measure far-UV (190–250 nm) CD spectra. GC-MS analyses were run using an Agilent 7890A GC coupled to a 5975 Inert mass selective detector with an Rtx-5Sil MS column. Kinetic simulations were performed using the Berkeley Madonna kinetics package.