Sulfenylation of Human Liver and Kidney Microsomal Cytochromes P450 and Other Drug Metabolizing Enzymes as a Response to Redox Alteration

in and showed consistent

reversing the modification and the roles that sulfenylated proteins play in biology (8). Current theories for the production of sulfenic acids include redox relays using peroxiredoxins, local sites of high oxidant production (e.g., NADPH oxidase, mitochondria), and direct diffusion across plasma membranes (9). Sulfenylation can lead to disulfide bond formation with glutathione or other free thiols or can be reversed by either glutaredoxin or protein disulfide isomerase (7).
Recently new evidence has emerged that ER-localized tyrosine-protein phosphatase non-receptor type 1 (PTP1B) sulfenyl groups can be reduced via thioredoxin reductase (10).
Here we describe the importance of cysteine oxidation in the context of other P450s and identify thiol sensitive and insensitive enzymes. Heme-thiol sensitivity is adopted as a term used here to describe enzymes that experience direct oxidation of the heme-thiolate center, thus altering heme-iron coordination and catalytic activity. Conversely, thiol insensitivity is used to describe enzymes that do not exhibit sensitivity to thiol oxidation up to 1 mM concentrations of H 2 O 2 .
Further findings from a proteomic analysis of sulfenylated proteins in both murine kidney and liver microsomes indicated that other P450 enzymes are also sulfenylated, including enzymes in the P450 2a, 2c, 2d, and 3a subfamilies, identified here. A subsequent screen in human liver and kidney microsomes yielded similar results. These analyses of human and murine microsomes from livers and kidneys revealed sulfenylation of many other important drug metabolizing enzymes including UDP-glucuronyltransferases (UGTs), epoxide hydrolase, flavin-containing monoxygenases (FMOs), monoamine oxidases (MAOs), and carboxylesterases. Using enzymatic activity assays, isotope-coded dimedone/iododimedone (ICDID) labeling (Fig. 1), and spectral studies, we further investigated the effect of H 2 O 2 on human recombinant P450s 1A2, 2C8, 2D6, and 3A4. We observed that P450s 2C8, 2D6, and 3A4 experienced heme-thiol sensitivity (with 6 the modification to P450s 2C8 and 2D6 being reversible) and that P450 1A2 is thiol-insensitive to H 2 O 2 . This redox phenomenon may represent a new regulatory mechanism for many P450s.
Murine Tissue Samples-All experiments using mice were conducted with approved protocols by the Institutional Animal Care and Use Committee of Vanderbilt University and in accordance with the NIH Guide for the Care and Use of Laboratory Animals. 129/Sv mice carrying one copy of the human cytochrome P450 4A11 gene (CYP4A11) (under control of its native promoter were generated as previously described (17)) were provided normal chow diet (Purina Laboratory Rodent 5001; Purina, St. Louis, MO) with free access to water and were housed in an Association for the Assessment and Accreditation of Laboratory Animal Care (AAALAC)accredited, temperature-controlled facility with a 12 h light-dark cycle. All studies were 7 conducted in mice aged 6-28 weeks of age. CYP4A11 transgenic mice were crossed with pure Sv129 wild type mice and offspring were genotyped for the presence of a single copy of the human CYP411 gene as previously described (17). Organs were collected from male transgenic mice immediately after sacrifice and used for microsomal preparations as described below.
Human Tissue Samples-Tissues were collected and stored by the Vanderbilt University Medical Center Tissue Repository using Cooperative Human Tissue Network (CHTN) approved standard operating procedures, under a waiver of consent and anonymized. After collection, areas of necrosis or cauterized areas were removed and then sectioned into normal and tumor tissue. A representative section of each tissue type and/or disease type was fixed in formalin and processed, and a hematoxylin and eosin stained section was obtained and reviewed by a board-certified pathologist to ensure sample integrity and usefulness in research. Kidney and liver samples (five each, collected within one year of analysis, decoded) used for this study were snap-frozen in liquid nitrogen and stored at -80 °C.

Isotope-Coded Dimedone/Iododimedone (ICDID) Labeling of Recombinant P450s-
Additional aliquots of oxidized protein (from above, 100 pmol) were incubated with 5 mM d 6dimedone (from a 50 mM stock suspended in 100 mM sodium 3-[4-(2-hydroxyethyl)-1piperazinyl]propanesulfonate (HEPPS) buffer (pH 8.0) containing 5% NaCl (w/v)) for 2 hours at 37 °C. Trichloroacetic acid was then added to a final concentration of 10% (w/v) and the samples were incubated on ice for 15 minutes. The enzymes in the samples were pelleted by centrifugation (12,000 ´ g, 15 min, 4 °C). The supernatant was removed from each sample, and the pellet was tolerance was set at 10 ppm, and fragmentation tolerance set at 0.5 m/z. The maximum Q values of peptide spectrum matches were adjusted to achieve a peptide false discovery rate ≤5%, using IDPicker software (Version 3.1.642.0) (25). A spectral library of peptides was then created with IDPicker and loaded into Skyline Software for confident identification and quantitation of precursors pertaining to cysteine-containing peptides. MS 1 precursor quantitation was performed as described previously (26,27). The retention time of the identified peptide was used to position a retention time window (± 2.0 min) across the run lacking the same peptide identification. Second, the resolution for extracting the MS1 filtering chromatogram of the target precursor ions with both light and heavy labeled peptides was set to 60,000 at 400 Th. Then extracted ion chromatograms for the top three isotopic peaks were manually inspected for proper peak picking of MS1 filtered peptides and those with isotopic dot product scores lower than 0.8 were rejected. Additional criteria were used to further ensure the high accuracy and precision of quantification: S/N> 3.0 and baseline separation was required between the isotopic peaks of a quantifiable peptide and unknown isobaric interference. The ratios of peptide areas of light peptides to their heavy isotopes (RL:H) were calculated automatically. Quantification results were obtained from five biological replicates for human microsomes, four biological replicates for murine microsomes, and two biological replicates for recombinant protein.
Spectroscopy-P450s were oxidized with H 2 O 2 as above but the procedure was adapted slightly for spectroscopic assays. The enzyme was diluted with oxidation buffer to 1. with NADPH (300 nmol, in aqueous solution) in a sidearm of the cuvette. Samples were degassed, 20 µM 3,4-dihydroxybenzoate (Sigma, substrate for protocatechuate dioxygenase) was added to remove oxygen (28), and samples were further degassed using a manifold attached to both vacuum and purified Ar (29,30), and placed under an anaerobic CO atmosphere. The valves of the cuvettes were sealed and multiple UV-visible absorbance spectra were recorded using an OLIS/Hewlett Packard 8452 diode array spectrophotometer (On-Line Instrument Systems, Bogart, GA). Spectra were collected from 380 to 600 nm before and after the addition of NADPH and then sodium dithionite (6).
Experimental Design and Statistical Rationale-All chemoproteomic studies using tissue were performed with at least four biological replicates to evaluate biological variability. Enzymatic activity and labeling assays using recombinant purified enzymes were performed in biological duplicate and showed consistent results. Values of means with standard deviation are presented.

RESULTS
Identification of Sulfenylated P450s in Murine Microsomes-Following our report that identified oxidation of P450 Subfamily 4 enzymes in murine kidney and liver microsomes (6), further analysis led to the identification of other sulfenylated cysteines from other P450s in Subfamilies 2, 3, and 26 (Table 1, Supplemental Table 1). Of these, the murine P450 3a, 2c, 2d, and 2e proteins are known drug metabolizing enzymes in mice (31). These results indicate that many P450s may be inhibited by oxidative posttranslational modification of cysteines.

Identification of Sulfenylated Drug Metabolizing Enzymes in Human Kidney and Liver
Microsomes-The murine data (Table 1, Supplemental Table 1) led to the expansion of the proteomic study to human microsomes. Frozen human kidney and liver tissues from five individuals were rapidly fractionated into microsomal fractions in deaerated buffer and labeled using an isotope-coded ICDID strategy for relative quantitation (Fig. 1) (32). After excising the 40-60 kDa P450 region and digesting the proteins with trypsin, we identified 347 modified proteins in the kidney microsomes at a 5% peptide false discovery rate (FDR) ( Table 2, Supplemental Table   2, Fig. S1). In addition, 380 modified proteins were identified in the liver microsomal fractions (5% peptide FDR, Table 2, Supplemental Table 3, Fig. S1). Of these proteins, 11 P450s were sulfenylated in the kidney microsomes and 24 P450s in liver microsomes. The identified P450s included enzymes that are important in the metabolism of both endogenous and xenobiotic substrates.
In addition to the P450 enzymes, we identified other modified proteins important in drug metabolism, including UGTs, epoxide hydrolase, FMOs, MAOs, and carboxylesterases. Other 14 previously known sulfenylated proteins were also identified including protein disulfide isomerases (33) and aldehyde dehydrogenases (34). MS1 precursor intensities for peptides were quantified using Skyline software (26).

Spectral Analysis of Recombinant P450s for Thiol Sensitivity to H
2D6, and 3A4 were selected for further analysis. To confirm that these P450 enzymes were affected by sulfenylation, we utilized their spectral properties to determine if heme coordination or interaction with NADPH-P450 reductase was disrupted. In the presence of CO, disruption of the heme thiol ligation either prevents reduction of the heme or yields the inactive form, cytochrome P420 (the five-coordinate ferrous-CO complex of which has a maximal absorbance at P420 nm) so that a characteristic 450 nm spectrum is not observed (35). tris-(Carboxyethyl)phosphine (TCEP)-pretreated P450s were treated with 500 µM H 2 O 2 and (after the removal of residual H 2 O 2 by catalase) reconstituted with NADPH-P450 reductase, deaerated, and placed under an anaerobic atmosphere of CO. Following the addition of NADPH, spectra were recorded for the TCEP-reduced (Fig. 2, left column) and H 2 O 2 -oxidized (Fig. 2, right column) samples. Sodium dithionite, which reduces both sulfenic acids and the heme iron, was subsequently added to both samples (Fig. 3). P450 1A2 was very insensitive to H 2 O 2 , with only a slight increase in 450 nm absorbance after dithionite addition ( Fig. 2A). P450s 2D6 (Fig. 2B) and 2C8 (Fig. 2C) both exhibited sensitivity to H 2 O 2 -dependent oxidation, with a noted decrease in 420 nm absorbance (indicative of the inactive form, cytochrome P420) and an increase in 450 nm absorbance indicative of reestablishment of the heme-thiol ligand (Fig. 3). P450 3A4 (Fig. 2D) exhibited a complete loss of 450 nm absorbance in the oxidized sample, which was not reversible upon addition of dithionite and increased absorbance at 420 nm.
Oxidative Inhibition of P450 1A2, 2C8, 2D6, and 3A4 Catalytic Activities-The enzymatic activity of P450 1A2 was largely uninhibited with preincubation of up to 1 mM H 2 O 2 , using phenacetin as a substrate, consistent with the spectral results (Fig. 4A). P450 2C8 showed sensitivity to H 2 O 2 , with an approximate IC 50 of 150 µM and a loss of 87% activity at 1 mM H 2 O 2 , using taxol as a substrate (Fig. 4B). P450 2D6 was also inhibited by H 2 O 2 , with an estimated IC 50 of ~300 µM (Fig. 4C) and loss of 70% activity at 1 mM H 2 O 2 , using dextromethorphan as a substrate. An IC 50 of ~300 µM H 2 O 2 was determined for P450 3A4, using testosterone as a substrate (Fig. 4D), with a loss of 95% of activity at 1 mM H 2 O 2 .

DISCUSSION
ICDID labeling of murine liver and kidney microsomes provided the interesting finding that P450s other than the Subfamily enzymes 4 previously described (6) also contained oxidatively modified cysteines. This result may seem inconsistent with previous activation studies involving other P450s and dithiothreitol, which found no significant differences in enzymatic rates (6).
However, this is probably because it is a regular practice to dialyze against dithiothreitol when removing imidazole after His 6 -nitrilotriacetic acid/nickel (NTA-Ni 2+ ) purification and storage.
P450 4A11 is unusual in its ability to readily oxidize in the presence of air, at least under these conditions. This labeling strategy was then expanded to human liver and kidney microsomes.
Many other P450s were found to be sulfenylated in these samples ( Table 2, Supplemental Tables   S2, S3). While our knowledge of transcriptional regulation, sequence variation, and inhibition is extensive for many P450s (4), knowledge of post-translational regulation of P450s is limited, with research mostly focused on glycosylation, phosphorylation, ubiquitination, and nitration (24,36).
Cysteine oxidation may play a role in physiological post-translational regulation as well.
Cysteine sulfenylation was found in a total of 57 drug metabolizing enzymes. The group includes UGTs, FMOs, and carboxylesterases, suggesting that many microsomal enzymes are modified by oxidation. These results seem reasonable, in that many enzymes have been found to be regulated by oxidation to form cysteine sulfenic acids (34,37). Additionally, a comparison of the results presented here and a recent study on sulfenylated proteins (38) revealed that 21% (kidney microsomes) and 14% (liver microsomes) of proteins were identified in both datasets. This comparison may indicate that the RKO adenocarcinoma cell line studied by Gupta et al. (38) likely has similar basal proteins that are oxidized. More in-depth studies of the non-P450 enzymes will be required to further verify sensitivity of catalytic activity to oxidation and effects that oxidation may have on activity and regulation.
Relative quantitation of the modified cysteines ( Table 2, Supplemental Fig. S1) yielded similar levels of d 6 -dimedone labeling, which may be due in part to the one hour incubation time with microsomes. While dimedone has been shown to be selective, the low reactivity of dimedone with sulfenic acids requires extended incubation times to achieve sufficient labeling (rate of 0.8 min -1 under the reported conditions) (39,40). The modified cysteines identified are likely sensitive to oxidation, but the amount of labeling observed may be representative of the equilibrium of a saturated system. Overall, 35% and 33% of all cysteine-containing peptides identified were modified with d 6 -dimedone in kidney and liver microsomes, respectively. Since more than half of the peptides identified were not sulfenylated, the likelihood that this observation is an artifact would seem to be low. Furthermore, a control experiment was performed in which kidney and liver tissues were homogenized in 5 mM TCEP, fractionated into microsomes, labeled, and subjected to LC-MS/MS analysis as described in Experimental Procedures. Approximately 6% of all cysteine-containing peptides from both the kidney and liver samples were modified with d 6 -dimedone suggesting that some sulfenylated cysteines may have been inaccessible to the TCEP reducing agent (Supplemental Fig. S7). Also, the fractionation and labeling protocol may have introduced a slightly oxidative environment that modified highly susceptible cysteines.
Nevertheless, this experiment serves as a control for our positive results.
Four drug metabolizing P450s were chosen for more in-depth oxidative inhibition studies.
Interestingly, P450 1A2 was resistant to oxidation, as established by spectral, inhibition, and dimedone labeling studies (Figs. 2A, 4A, and 5A, respectively). Of note was the hyperoxidation of Cys-159, which did not affect the catalytic activity of P450 1A2 (Fig. 5A). This ancillary cysteine is positioned away from the active site, and its modification has a negligible effect on function (Fig. 6A). The resistance to oxidation is proposed to be related to access of oxidants or to stabilization of the heme-thiol system due to either the surrounding amino acid residues or the overall structure of the protein.
P450 2C8 showed an 87% loss of catalytic activity at 1 mM H 2 O 2, compared to the reduced control (Fig. 3B). This inhibition may seem surprising in light of the large number of cysteines contained in P450 2C8 (Fig. 7), but the heme thiol peptide was still modified. Despite the gradual loss of activity, the spectral and ICDID labeling studies (Figs. 2B and 5B) showed a high loss and recovery of the heme-thiol ligand spectrally and significant sulfenic acid labeling (Cys-435). Only four of the 16 thiols were quantified because of the clustering of thiols in the sequence; i.e., some tryptic peptides contained up to four cysteines and spanned >25 amino acids.
P450 2D6 showed less loss of activity (70%) than P450 2C8 but still exhibited inhibition after treatment with H 2 O 2 (Fig. 4C) and the ability of the heme-thiol to re-ligand the heme iron in the spectral analysis (Fig 2C). This diminished response was reflected in ICDID analysis, which showed lower amounts of sulfenylation of the heme-thiol Cys-443 and also Cys-191 (Fig. 5C).
Oxidized P450 3A4 showed an inability of the heme-thiol to re-ligand to the heme-iron (after dithionite reduction), as judged by the spectra (Fig. 2D). This irreversible inactivation is proposed to be related to stabilization of the oxidized thiol by surrounding residues and is a unique feature among P450s tested. ICDID labeling showed high amounts of sulfenylation on several cysteine thiols, including the heme-thiol Cys-442, indicating a high susceptibility to oxidation.
This sensitivity of the non-heme thiols can be considered in light of the report that a cysteinedepleted variant of P450 3A4 has increased activity compared to the wild-type enzyme (41). We concur that the cysteine residues are not essential but that the presence of the (non-heme peptide) sulfenic acids appears to be inhibitory to catalytic activity, through an unknown mechanism.
There has been difficulty designing probes that can directly measure local bursts of H 2 O 2 and not just a cumulative or average amount over time, especially in prominent production areas such as the plasma membrane, endoplasmic reticulum, and peroxisomes (42). When looking at redox issues, one mainly focuses on a concentration of H 2 O 2 for which the effect can be reversed.
This is the case for three of the four P450s tested spectrally, suggesting that this treatment does not irreversibly destroy the protein.
These spectral studies have been challenging to interpret because the proper conditions have not yet been identified to produce a homogenous sample of stable sulfenylated heme-thiol cysteine P450 protein in large quantities. The spectral studies in conjunction, with the observation that sulfenylated cysteines cause enzymatic inhibition, lead us to believe this phenomenon is blocking steps in the P450 catalytic cycle. This blocked step may be reduction of the heme iron.
This could be occurring directly (most likely) or through a peripheral oxidation that limits reductase binding. It may also reduce the ability for the heme iron to bind oxygen (or carbon monoxide in the case of the inhibitory or spectral studies, respectively). The spectral studies also 20 require physical changes (i.e., vacuum and slight bubbling) to remove the existing oxygen in the sample. These manipulation conditions likely affect some protein irreversibly, accounting at least in part for the cytochrome P420 seen in the pre-reduced samples. What is most relevant is the change in the amount of P450 seen between the reductase-and dithionite -reduced spectra (Fig.   2).
It has been known that some P450s can catalyze reactions with the use of H 2 O 2 alone for quite some time (43)(44)(45)(46). Also, mutating residues around the proximal heme ligand allows for alterations in both the heme redox potential and reactivity (47). P450s are also known to produce H 2 O 2 as a byproduct of catalysis (48). CYP4A11 H 2 O 2 production was measured to be 5 µM min -1 , which would be 25 nmol H 2 O 2 min -1 under our experimental conditions (6). Typical incubation times would likely not produce enough H 2 O 2 to produce an inhibitory effect. These data point to the significance of H 2 O 2 in the catalytic cycle of P450s and of the residues surrounding the hemethiol ligand. The presence of a biological mechanism that limits potentially harmful H 2 O 2 shunting in certain mammalian P450s, but is not present in others (e.g. P450 1A2), seems reasonable but more studies are needed to evaluate the significance of this phenomenon.
Our finding of both thiol-sensitive and insensitive P450s expands the knowledge of potential post-translational modifications found in drug metabolizing enzymes. In general, cysteines are reactive, underrepresented in the proteome, and conserved among proteomes (49). In an accounting of all human P450s, the number of cysteines varies from two to 15 (Fig. 7). Using the average length of all human P450s (501 amino acids) and the overall percent of cysteines found in the human proteome (2.26% (49)), the expected number of cysteines in human P450s would be 11. Due to the deviation from this number, these cysteines may play important roles other than heme coordination and further investigation in cellular systems may be of interest.
* Single amino acid variation (leucine/isoleucine) in peptide sequences that cannot be determined from MS/MS fragmentation.