A requirement for an active proton delivery network supports a compound I-mediated C – C bond cleavage in CYP51 catalysis

CYP51 enzymes (sterol 14 a -demethylases) are cytochromes P450 that catalyze multistep reactions. The CYP51 reaction occurs in all biological kingdoms and is essential in sterol biosynthesis. It removes the 14 a -methyl group from cyclized sterol precursors by first forming an alcohol, then an aldehyde, and finally eliminating formic acid with the introduction of a D 14 – 15 double bond in the sterol core. The first two steps are typical hydroxylations, mediated by an electrophilic compound I mechanism. The third step, C – C bond cleavage, has been proposed to involve either compound I ( i.e. FeO 3 1 ) or, alternatively, a proton transfer-independent nucleophilic ferric peroxo anion (com-pound 0, i.e. Fe 3 1 O 2 – ). Here, using comparative crystallographic and biochemical analyses of WT human CYP51 (CYP51A1) and its D231A/H314A mutant, whose proton delivery network is destroyed (as evidenced in a 1.98-Å X-ray structure in complex with lanosterol), we demonstrate that deformylation of the 14 a carboxaldehyde intermediate requires an active proton relay network to drive the catalysis. These results indicate a unified, compound I-based enzymes follow same compound I-mediated mechanism

; compound 0, step 4 in Fig. 2) (6-10), but the reliability of the 18 O isotope incorporation and MS data has been questioned because of technical concerns about ambient formic acid (and the difficulties in discerning 2 HCO 2 H from endogenous H 13 CO 2 H) (5,11). In more recent studies, a ferric peroxide mechanism has been supported by a kinetic solvent isotope effect (12) and resonance Raman spectroscopy (13,14) for the substrate-induced lyase reaction of CYP17A1, although evidence for a compound I pathway has also been presented (15,16). Compound I was identified as the active oxidant for the lyase step of CYP19A1 (steroid aromatase) using resonance Raman spectroscopy (17), kinetic solvent isotope effects (18), and 18 O isotope labeling (11) and for CYP11A1 (P450scc) using electron paramagnetic resonance/electron nuclear double resonance spectroscopy (19), 18 O isotope labeling (20), and resonance Raman spectroscopy (21). For the third step of the CYP51 reaction, theoretical studies have proposed the operation of a ferric peroxo anion (compound 0) (22), but definitive experimental evidence so far has been lacking.
A distinguishing difference between the two types of mechanisms is that the formation of compound I requires a specific proton relay network (to minimize the unwanted production of hydrogen peroxide via nonspecific proton donation to dioxygen in P450 enzymes) (23)(24)(25)(26), whereas the ferric peroxo-anion abstracts hydrogens from substrates and does not depend on proton delivery (13). It has been generally accepted (going back to the 1980s [27][28][29]) that the proton delivery network involves a conserved P450 threonine in the I helix, a preceding charged residue, and a solvent-accessible residue(s) of opposite charge in the middle of helix F (3,30,31). The conserved threonine is Thr-315 in human CYP51 (also called CYP51A1) and corresponds to Thr-252 in P450 cam (24). A preceding charged residue is always a His in CYP51 versus an acidic residue (Asp/Glu) in other P450s, and a solvent-accessible charged residue(s) in the helix F is always Asp/Glu in CYP51 versus Lys/Arg in other P450s.
In the ligand-free and inhibitor-bound CYP51 structures (which all display a very high overall similarity), this histidine and the carboxylate of an F helix acidic side chain are always salt bridged (e.g. His-294/Glu-205 in Trypanosoma cruzi and His-314/Asp-231 in human CYP51 orthologs). However, as we recently reported, binding of the substrate (obtusifoliol) to the T. cruzi CYP51 I105F mutant induces a large-scale conformational switch that causes the His-294/Glu-205 salt bridge opening, allowing the charged side chains to participate in proton delivery (32).
Mutation of each of the corresponding salt bridge-forming residues in human CYP51 strongly affects catalysis of lanosterol, with the double mutant (D231A/H314A) displaying a decrease in k cat of more than two orders of magnitude. There were no changes in the stability, spectral properties (absolute absorbance and CO complex formation upon chemical reduction), or apparent substrate binding affinity of the mutants, but the amplitude of the type I spectral response to lanosterol (lowto-high spin state transition in the heme iron) increased in the order WT , D231A , H314A , D231A/H314A (32). The D231A/H314A mutant is the first example of a CYP51 that, upon the addition of substrate, is found completely in the highspin form (Fig. 3), although without structural evidence it has remained unclear whether it indeed forms a stoichiometric enzyme-substrate complex in solution or if the heme iron acquires the ability to exist in the high-spin (water-free pentacoordinated) form even without the substrate.
In this report, we determined the 1.98-Å X-ray structure of the human CYP51 D231A/H314A mutant in complex with lanosterol and applied comparative crystallographic and biochemical analyses of this mutant and WT human enzymes to probe the CYP51 mechanism. We found that the activity of the mutant toward the 14a-aldehyde substrate is affected as much as its activity with lanosterol, leading to the conclusion that proton transfer is critical for the third step of the CYP51 reaction and compound I is the active oxidant.

Crystallographic analysis
The enzyme-substrate complex was crystallized in the orthorhombic C2221 space group, and the final structure was refined to 1.98-Å resolution with an R free of 22% and R work of 19% (Table 1). The atomic coordinates and structure factors have been deposited in the Protein Data Bank under accession code 6UEZ. The asymmetric unit consisted of two P450 molecules, each of them revealing a clear electron density for one molecule of lanosterol bound within the CYP51 active site ( Fig. 4A and Fig. S2).
Substrate binding mode-The binding mode of lanosterol in the human structure was found to be essentially the same as the binding mode of obtusifoliol in the T. cruzi CYP51 I105F mutant (32) (Fig. 4, B and C). The C3-OH group of the sterol nucleus is directed toward the substrate access channel entrance approaching b-strand 1-4 and forming an H-bond with the main-chain oxygen of Ile-379 (corresponds to Met-358 in T. cruzi CYP51), the aliphatic arm occupies the deepest portion of the active site reaching helices C and I, and the C14a-methyl group is located 4 Å above the heme iron, indicative of a catalytically competent orientation.
Structural response to substrate binding-The active-site volumes of both protozoan and human substrate-bound CYP51 enzymes are very similar (1,200 and 1,300 Å 3 ), and so is the large-scale conformational switch seen in the superimposed ligand-free/obtusifoliol-bound T. cruzi CYP51 (of the four P450 molecules in the asymmetric unit in that structure, three are obtusifoliol-bound and one is ligand-free) and inhibitor-bound/ lanosterol-bound human CYP51 (Fig. 5). The rearrangements involve most of the structural segments of the CYP51 active site. The 5-6-Å inward movement of the HI arm and helix C changes the topology of the area to create a tighter interface with NADPH-CYP reductase (CPR), and the side-chain flip of  Binding of the substrate (1) displaces the water molecule from the sixth (distal) coordination site of the ferric heme iron, changing its spin state from hexa-coordinated low to penta-coordinated high, increasing its redox potential, and facilitating the first electron transfer from the protein redox partner (2). The ferrous iron complex binds molecular oxygen (3), which triggers acceptance of the second electron producing the nucleophilic ferric peroxo anion (4) (compound 0, or Fe 31 -O 2 2 ). The ferric peroxo complex is then protonated to form the ferric hydroperoxo state (5), the second protonation of the distal oxygen atom causing the O-O bond scission, release of a water molecule, and generation of a highly reactive electrophilic Fe 41 -oxo cation radical (FeO 31 , or compound I) (6). The compound I mechanism requires two protons delivered from the protein surface to the iron-bound dioxygen via the active proton delivery network. In the compound 0 mechanism, the dioxygen abstracts protons from the substrate, forming a peroxo-hemiacetal intermediate and, thus, being independent of proton delivery.
Lys-156 out of the active site is analogous to that of Arg-124 in T. cruzi CYP51, the residue we have shown to directly participate in the electron transfer (32). In both, the H-bond with the heme propionate is lost and a positively charged group is now exposed above the proximal P450 surface (

Biochemical analysis
Electron transfer-Both WT human CYP51 and the D231A/ H314A mutant were readily reduced chemically with sodium dithionite, regardless of the presence of the substrate, and the P450 concentrations determined from the difference spectra of the CO complexes corresponded to the concentrations calculated from the absolute absorbance spectra. The efficiency of enzymatic reduction with NADPH via CPR, however, was facilitated by the presence of the substrate, 100% in the mutant versus 65% in the WT sample, whereas without lanosterol the amount of detectable CO complexes did not exceed 15% in either case (Fig. 7A). These results support the conclusion that a crystallographically detectable substrate-induced large-scale conformational change does take place in solution, and the 35% lower enzymatic reduction of the WT sample is because of a lower lanosterol-bound fraction (33% versus 100% in the mutant).
Slightly faster rates of NADPH oxidation by the WT enzyme (24 versus 16 min 21 ) (Fig. 7B) were observed, because WT CYP51 performed lanosterol 14a-demethylation while the mutant was catalytically inactive. In the absence of the substrate, the rates of NADPH consumption were comparable for WT CYP51 and the mutant, both one order of magnitude slower. Thus, the structural and biochemical data indicate that the D231A/H314A mutation did not affect (but rather enhanced) the ability of human CYP51 to bind substrate, interact with the electron donor protein CPR, and accept electrons. The lack of charged residues that enable the proton flow is the only observed reason for the dramatic loss of enzymatic activity.
Deformylation of lanosterol 14a-carboxaldehyde-The mutant provided a unique opportunity to elucidate the mechanism of the C-C bond cleavage because, if this step in CYP51 catalysis is performed by a nucleophilic peroxo oxidant, it should not require protons from the solvent, and the ability of the D231A/ H314A mutant to catalyze deformylation of the 14a-carboxaldehyde intermediate of lanosterol (as opposed to its ability to hydroxylate lanosterol) should be comparable with or higher than that of the WT. If this reaction occurs via compound I, for which the formation transfer of protons from the solvent is critical (Fig. 2), the ability of the D231A/H314A mutant to catalyze deformylation of the 14a-carboxaldehyde intermediate should remain low.  We observed that catalytic turnover of the WT enzyme with the aldehyde intermediate as the substrate (k cat of 95 min 21 ) was twice as high as that with lanosterol (k cat of 46 min 21 ), which is expected in that the reaction occurs in one final step instead of three (K m of ;5 mM in each case). The activity of the mutant was low, as in the case of lanosterol (k cat of 0.6 min 21 ) (Fig. 8A), revealing the same decreased reaction velocity, with K m values being too low to measure accurately (32) and longer reaction time required for detectable product to be formed. The data strongly suggest that all three steps of CYP51 catalysis have the same compound Imediated mechanism.
It has been proposed that for a C-C bond cleavage reaction to proceed through the ferric peroxo oxidant instead of compound I, a P450 substrate must bear a hydroxyl group juxtaposed to the Fe-O-Ofragment, which was found to be the case for CYP17A1/17OH-pregnenolone but not for CYP19A1 (14,17). This is also not the case for CYP51, as in the CYP51 substrates the only OH group (at the C3 carbon) is located 8 Å from the C14a-aldehyde group and H-bonded with the main chain carbonyl oxygen in the b 1-4 strand (Fig. 4C). Thus, the current evidence supports compound I mechanisms in three of the four major steroid C-C bond cleavage reactions ( Fig.  1 and Fig. S1).
It should be noted that in CYP17A1, some of the evidence suggests the interaction of the substrate hydroxyl group with compound I (5,16). However, in the case of CYP17A1, the 18 O labeling work is not unambiguous in discerning between the two pathways (5,15), and the reported spectral intermediates have not been shown to be catalytically competent (5,13,14,17), so more experimental work on this enzyme is in order. Overall, the number of P450 reactions that can be attributed to compound 0 ferric peroxo anion C-C bond cleavage and other types of transformation has decreased  5 s) for the active-site area of the D321A/H314A mutant of human CYP51. Lanosterol, heme, and some of the protein residues are marked. The snapshot was taken in Coot. B, superimposed structures of obtusifoliol-bound I105F T. cruzi CYP51 (pink, PDB code 6FMO) and lanosterol-bound D231A/H314A human CYP51 (yellow). Shown is the distal P450 face. The RMSD of Ca is 1.2 Å. C, substrate-contacting residues in human (left) and T. cruzi CYP51 (right). The active-site volumes are 1,200 Å 3 and 1,300 Å 3 , respectively. The complete list of the residues located within 4.5 Å of the sterol molecules can be found in Table S1. (5,(33)(34)(35), although the possibility exists that some facile reactions do involve this species.

Conclusions
To clarify whether the third step of CYP51 catalysis can proceed through a proton-independent ferric peroxo anion-mediated mechanism, we applied a combined crystallographic and biochemical approach. The work was conceived based on our previous findings that substrate binding induces a large-scale conformational switch in the structure of CYP51 from the protozoan pathogen T. cruzi (32). Among the changes was a 4-6-Å movement of the FG arm that led to the opening of the His-294/Glu-205 salt bridge, a process expected for the activation of the CYP51 proton delivery network.
Mutation of the corresponding residues in human CYP51 (His-314 and Asp-231) to alanine, without changing the apparent lanosterol binding affinity, caused a decrease of two orders of magnitude in the rate of lanosterol 14a-demethylation yet increased the amplitude of the mutant spectral response to la-nosterol binding. Cocrystallization of the mutant with lanosterol produced a 1.98-Å structure of the enzyme-substrate complex, providing convincing evidence that the large conformational switch accompanying substrate binding must be general for CYP51 across phylogeny, physiologically relevant, and required for catalysis.
The D231A/H314A mutation did not cause any apparent disturbances in substrate binding (including the spectral K d values), protein folding, stability, conformational dynamics, or the ability to accept electrons from NADPH via its electron donor partner CPR, and the disruption of the proton delivery network (Fig. 8B) is judged to be the only observable reason for the loss of catalytic activity. Because proton delivery is a critical step in the activation of iron-linked molecular oxygen, required for splitting of the O-O bond and formation of compound I, the finding that the mutant catalytic rates are dramatically affected both for lanosterol (hydroxylation, the first step of the three step reaction) and for the 14a-carboxaldehyde intermediate (C-C bond cleavage, the third step [a proposed reaction scheme is shown in Fig. 9]) strongly supports the view that CYP51 . Middle, the overall distal P450 view. The arrows mark the areas of the molecules enlarged above (rearrangements for electron delivery) and below (rearrangements for proton delivery). The conserved salt bridges are shown as red dashes. enzymes follow the same compound I-mediated mechanism during all three steps of the reaction.
On the basis of our work here and previous studies with CYP11A1 and CYP19A1, we conclude that the C-C bond cleavage mechanisms involve compound I in at least three of the four major steroid P450s that perform multistep reactions The approach we used should be applicable in distinguishing between Fe 31 -O 2 2 and FeO 31 as catalytically active species for other P450 cytochromes that catalyze multistep reactions.

Experimental procedures Protein expression and purification
All proteins were expressed in Escherichia coli (HMS-174, Novagen) in a pCW expression vector. The expression and purification of human CYP51 (full-length WT and the D231A/ H314A mutant) was performed as described previously (32). For crystallization purposes, the 50-amino-acid membrane anchor sequence at the N terminus of D231A/H314A was replaced with an MAKKTSSKGKL fragment (36). The expression and purification of rat NADPH-cytochrome P450 reduc-tase (CPR) and T. brucei CYP51 and CPR were performed as described previously (37,38).
Spectroscopic characterization-UV-visible spectra were recorded at room temperature (23°C) using a dual-beam Shimadzu UV-240IPC spectrophotometer. P450 concentrations were determined from the Soret band absorbance in the absolute spectrum, using an absolute molar extinction coefficient (e 417 ) of 117 mM 21 cm 21 for the low-spin oxidized form of the protein or a difference molar extinction coefficient (De 446-490 ) of 91 mM 21 cm 21 for the reduced carbon monoxide complex in the difference spectra (Fe 21 -CO versus Fe 21 ). The spin states of P450 samples were estimated from the absolute spectra as the DA 393-470 /DA 418-470 ratio, with values of 0.4 and 2.1 corresponding to 100% low-and 100% high-spin iron, respectively (36,39). NADPH oxidation rates were monitored at 340 nm (De 340 , 6.22 mM 21 cm 21 ).
Crystallization, structure determination, and analysis-The human CYP51 D231A/H314A mutant [5 mM in 20 mM potassium phosphate buffer (pH 7.4) containing 500 mM NaCl, 5.6 mM tris(2-carboxyethyl)phosphine, and 10% (v/v) glycerol] was gradually saturated with lanosterol (using a 1 mM stock solution in 45% [w/v] 2-hydroxypropyl-b-cyclodextrin [HPCD]), incubated for 20 min at room temperature, centrifuged to remove the precipitate, concentrated using an Amicon Ultra 50K (Millipore) to 500 mM, diluted 2-fold with 5 mM phosphate buffer (pH 7.4), and mixed with 23 mM cyclohexylpentanoyl-N-hydroxyethylglucamide (Ana-trace). Crystals were obtained at 26°C by vapor diffusion in hanging drops using 2 ml protein mixture and 2 ml of mother liquid consisting of 100 mM calcium acetate (pH 7.3), 18% (w/v) PEG 3350, and 0.1 mM EDTA. The crystals were transferred in three steps of increasing glycerol concentration to a cryoprotectant solution containing 25% (v/v) glycerol in mother liquor and flashcooled in liquid nitrogen. X-ray diffraction data were collected from a single crystal at the 21-ID-D beamline at the Advanced Photon Source, Argonne National Laboratory, at a wavelength of 1.12713 Å and using a Dectris Eiger 9M detector. The diffraction images were processed with HKL-2000, and the initial electron density map was obtained by molecular replacement using the coordinates of the VFV-bound human CYP51 (PDB entry 4UHI) as a search model in Phaser MR (CCP4 Program Suite [40]). The structure was built with Coot (41) and refined with Refmac5 (CCP4 Suite). Details of the data collection and refinement statistics are listed in Table 1. Structure superimposition and root mean square deviation (RMSD) calculation were performed in lsqcab (CCP4 Suite). Active-site volumes were calculated in BioVia Discovery Studio Visualizer 2019. Molecular graphics were rendered using Chimera.
Preparation of lanosterol 14a-carboxaldehyde-The 14acarboxaldehyde intermediate of lanosterol (i.e. the substrate for the third step of the CYP51 reaction, the C-C bond cleavage) was prepared using a feature of T. brucei CYP51, i.e. that of producing a large amount of this intermediate if lanosterol is used as the substrate (38). The reaction mixture contained 10 mM T. brucei CYP51, 5 mM T. brucei CPR, and 50 mM lanosterol (for this purpose, 0.5 mM solutions of unlabeled and [3-3 H]-labeled lanosterol [;4000 dpm/nmol] were mixed in 45% HPCD at a 10:1 molar ratio). The radiolabeled lanosterol was added to be able to quantify the amount of aldehyde by its peak area. The reaction was started by the addition of 1 mM NADPH, conducted at 37°C for 2 h, and the 14a-carboxaldehyde intermediate was separated by reverse-phase HPLC and collected as described previously (38), except that the time of the linear gradient (vide infra) was extended to 40 min (Fig. S5). The aldehyde fraction was dried under a nitrogen stream, dissolved in 45% (w/v) HPCD, and used as a substrate for comparative analysis of WT and D231A/H314A human CYP51 (Fig. S6).

Data availability
The atomic coordinates and structure factors (code 6UEZ) have been deposited in the Protein Data Bank (http://wwpdb. org/), and all other data are contained within the manuscript and the supporting information.