Characterization of a Selenocysteine-ligated P450 Compound I Reveals Direct Link Between Electron Donation and Reactivity Elizabeth Onderko†, Alexey Silakov†, Timothy H. Yosca‡, and Michael T. Green‡,* ‡Departments of Chemistry & Molecular Biology and Biochemistry, University of California, Irvine, CA 92697 †Department of Chemistry, Penn State University, University Park, PA 16802 Contact: [email protected] Abstract Strong electron-donation from the axial thiolate-ligand of cytochrome P450 has been proposed to increase the reactivity of compound I with respect to C–H bond activation. However, it has proven difficult to test this hypothesis, and a direct link between reactivity and electron donation has yet to be established. To make this connection, we have prepared a selenolate-ligated cytochrome P450 compound I intermediate. This isoelectronic perturbation allows for direct comparisons with the wild type enzyme. Selenium incorporation was obtained using a cysteine auxotrophic E. coli strain. The intermediate was prepared with meta-chloroperbenzoic acid and characterized by UV-visible, Mössbauer, and electron paramagnetic resonance spectroscopies. Measurements revealed increased asymmetry around the ferryl moiety, consistent with increased electron donation from the axial selenolate-ligand. In line with this observation, we find that the selenolate-ligated compound I cleaves C–H bonds more rapidly than the wild-type intermediate. Background Cytochrome P450s are a class of thiolate-ligated heme proteins that are known for their ability to functionalize unactivated C–H bonds. In efforts to understand reactivity in these and other thiolate-heme systems, comparisons are often drawn between P450s and the histidine-ligated heme peroxidases. Both classes of heme enzymes share a similar active intermediate: a ferryl (or iron(IV)oxo) radical species, called compound I1-3. But, only the thiolate-ligated compound I species are known to perform C–H bond activation4-7. It has been proposed that this difference in reactivity is due to strong electron donation from the axial thiolate ligand2,8-11. However, it has proven difficult to test this hypothesis. Although a number of experiments point to this role for the thiolate2,8,10,11, a direct link between reactivity and electron donation has yet to be established in P450s. To explore the effect that electron donation has on the reactivity and electronic structure of compound I, researchers have turned to synthetic model systems and site-directed mutagenesis. Work with model systems has demonstrated that donating ligands can increase reactivity with respect to oxygen-atom transfer and C–H bond cleavage12-15. These investigations have provided considerable insight, but synthetic model systems are typically many orders of magnitude less reactive than P450s. Efforts to directly probe the role of the axial ligand via heme protein modifications have proven more difficult. Attempts to convert histidine-ligated heme proteins into P450-like systems have resulted in enyzmes that lack P450-style reactivity16,17. Additionally, to prevent oxidation and subsequent dissociation of the cysteine residue introduced into these systems, other distal and proximal amino acid substitutions were required, underscoring the effect that the active-site environment can have on stability as well as reactivity18,19. Efforts to introduce less electron-donating ligands such as methionine, serine, and histidine into P450s have generally resulted in either apo- or misfolded protein20-22. Even when heme-containing protein was obtained, the resulting P450 variants were incapable of carrying out monooxygenase chemistry22-25. One axial ligand variant that has shown promise for the study of P450 chemistry is the substitution of selenocysteine for cysteine. The introduction of the selenolate ligand is attractive, because it represents a relatively conservative perturbation. The van der Waals radius of selenium is only 0.1 Å larger than that of sulfur26, and the formal charges of the selenolate and thiolate ligands are identical. However, the selenolate is more polarizable, has a lower redox potential, and is a better nucleophile than the thiolate27. These differences are expected to affect compound I reactivity. To date, selenocysteine has been successfully incorporated into the axial position of three different P450s28-32. Importantly, these SeP450s have been found to be catalytically competent28-30, and experiments have revealed an increase in electron donation from the axial ligand31,32. Efforts have been made to prepare compound I in these systems, but thus far only degradation products have been observed30. The analysis of P450-I reactivity with an alternative axial ligand could provide considerable insight into cytochrome P450 catalysis. Given the isoelectronic nature of the selenocysteine substitution, it would seem to be a prime candidate for these types of studies. Additionally, computations have predicted an increase in stability for the selenolate-ligated intermediate, suggesting it should be accessible via rapid mixing experiments33. Over the last few years, we have reported the preparation of compound I in several wild-type P450s1,2,34. These efforts have provided considerable insight into the conditions required for preparation of the reactive intermediate. Here, we present the capture and characterization of a selenolate- ligated cytochrome P450 compound I (Fig. 1). Results and Discussion Selenocysteine-substituted CYP119 (SeCYP119) was overproduced using a cysteine auxotrophic E. coli strain35-37 and purified as previously described1,28,30. CYP119 possesses only one cysteine residue (the axial thiolate ligand), therefore this method resulted in selenocysteine insertion solely at the axial site. Selenocysteine incorporation at this position was determined to be ~80% by EPR analysis of ferric SeCYP119. The UV-visible and EPR spectra of SeCYP119 were found to be in good agreement with those previously reported (Supplementary Figs. 1 and 2)28-31. Upon mixing the purified SeCYP119 with meta- chloroperbenzoic acid (m-CPBA), a substantial decrease in the Soret band was accompanied by increased absorbance at 370 and 695 nm (Fig. 2a). These changes suggested compound I formation, and, indeed, through the use of target testing (TT, vida infra), a compound I-like spectrum, which we assign to SeCP119-I, could be extracted from the stopped-flow data (Fig. 2b). Maximum formation of SeCYP119-I was observed approximately 12.5 ms after mixing with one equivalent of m-CPBA, after which time the intermediate rapidly decayed to a ferric species with a Soret maximum at 406 nm (Supplementary Fig. 3). The 406 nm-absorbing SeCYP119 species was previously observed by Ortiz de Montellano and co-workers, who attributed the 406 nm absorbance to selenenic acid ligation (i.e. oxidation of the selenolate ligand)30. The selenolate-ligated compound I species was also examined by Mössbauer and EPR spectroscopies. The SeCYP119-I spectra obtained from these techniques are shown in Fig. 3. Experiments have shown that the electronic structure of compound I is best described as an S=1 ferryl unit exchange coupled to S=1/2 ligand-based radical1. The exchange coupling, J, between the spin systems generates doublet and quartet states that are mixed by the zero-field splitting, D, of the ferryl moiety to yield three Kramers doublets38. Only the lowest of these is populated at the cryogenic temperatures required for EPR and Mössbauer measurements. As a result, the spectra can be fit in either an effective S=1/2 or a spin-coupled representation (Supplementary Figs. 4-7)1. The parameters obtained from these fits are listed in Table 1. Although the Mössbauer spectrum of SeCYP119-I shows similarities to the CYP119-I spectrum1, it is clear that selenocysteine ligation has altered the heme environment. The quadrupole splitting of the selenolate-ligated system is significantly larger than the value obtained for thiolate ligated intermediates (Table 1). Additionally, variations in the isomer shift and 57Fe hyperfine couplings suggest greater electron donation from the selenolate ligand. The isomer shift decreases across the series CPO-I > CYP119-I > SeCYP119-I consistent with increasing sigma donation, while the increasingly rhombic 57Fe hyperfine couplings across this series are consistent with stronger π donation39. The effective g-values of the compound I spectrum are determined by the ratio of |J/D|38. The g-values values indicate that |J/D| is larger for SeCYP119-I than CYP119-I. Fits in the spin-coupled representation yield |J/D| = 1.4, as compared to |J/D| = 1.3 for CYP119-I. A recent study linked an increase in |J/D| to an increase in electron donation from the axial ligand, suggesting that this quantity could be a marker for compound I reactivity11. To investigate the reactivity of SeCYP119-I towards substrates, we could not use the stopped-flow techniques previously employed for the wild type intermediate: SeCYP119-I is too unstable. The intermediate cannot be prepared and subsequently mixed with substrate to obtain reaction rates, as the decay to the 406 nm species is too rapid. To obtain insight into the reactivity of SeCYP119-I, we found it necessary to generate the intermediate in the presence of substrate. This increases the rate of SeCYP119-I decay, but it also provides a window into the reactivity of SeCYP119-I that cannot be accessed with sequential mixing studies. Although quantitative reaction rates could not be determined from data sets generated