To obtain detailed chemical information regarding the potential reactivity of cellular -SOH with thiol-blocking electrophiles, we focused our studies on a recombinant bacterial peroxiredoxin, AhpC. The C165S mutant of this 2-cysteine enzyme from Salmonella typhimurium was chosen because it allows accumulation of -SOH at C46 and because of its documented stability across various pH and temperature conditions, its efficient MS-induced ionization, and the lack of other cysteine residues that may otherwise complicate the interpretation of results . These characteristics allow the direct observation of chemical reactivity that cannot be achieved in cells or with cell lysates due to sample complexity. Preliminary investigations with C165S AhpC revealed that treatment of C165S AhpC-SOH with the thiol-blocking reagent IAM led to an unexpected decrease in the effectiveness of subsequent -SOH labeling using cyclic or linear 1,3-dicarbonyl probes (e.g. dimedone). To examine suspected -SOH quenching by IAM, freshly prepared C165S AhpC-SOH (30 μm) was incubated with IAM (2 mm) in ammonium bicarbonate (50 mm, pH 7.5) at ambient temperature for 3 h (Fig. 2A, first step). The reaction was quenched by passing the mixture through a Bio-Gel P6 spin column pre-equilibrated with 0.1% formic acid in water. Analysis by ESI-TOF MS revealed formation of a protein adduct at 20 673 a.m.u. with a yield of 72% (Fig. 2B). Unreacted AhpC-SOH is observed in the gas phase as the presumed dehydrated sulfenamide (-SN) species at 20 598 a.m.u. . The deconvoluted adduct mass is +16 a.m.u. relative to the adduct formed via reaction of IAM with reduced cysteine (+57 a.m.u. relative to -SH), suggesting the presence of one oxygen atom and corresponding to addition of the carboamidomethyl moiety to AhpC-SOH.
To further characterize this reactivity with other electrophilic reagents, AhpC-SOH (30 μm) was similarly screened for reaction with NEM (2 mm) and MMTS (2 mm) under the conditions described above. A 3 h incubation period with each of these electrophiles resulted in formation of stable adducts at 20 742 and 20 662 a.m.u., respectively, with higher observed yields (89% for NEM and 100% for MMTS in terms of -SOH consumption) compared to IAM (Fig. 2B). Trypsin digestion of the three samples with subsequent nanoLC-MS/MS peptide analysis confirmed that the modification includes oxygen, presumably from the sulfenic acid, and that the modification occurs at the peroxidatic C46 (Fig. S1).
Rates for the reaction of AhpC-SOH with saturating amounts of IAM, NEM and MMTS were obtained by monitoring adduct formation as a function of time using ESI-TOF MS. Product yields at various time points were plotted against reaction time using an exponential increase equation to yield pseudo-first order rate constants (kobs) of 1.1 ± 0.3 min−1 (MMTS), 0.07 ± 0.02 min−1 (NEM) and 0.030 ± 0.003 min−1 (IAM) (all obtained at pH 7.5 and ambient temperature, Fig. 3). Accounting for electrophile concentration, the peroxidatic C46 of AhpC had similar reactivity towards IAM in both the thiol (0.15 m−1·s−1 at pH 7.0)  and sulfenic acid (0.10 m−1·s−1 at pH 7.5) states. Higher rate constants have been reported for other redox-sensitive proteins, and have typically shown the same order of reactivity (MMTS > NEM > IAM) as observed for AhpC-SOH [23-26]. More importantly for the objective of this study, the rate constants for electrophilic -SOH capture exceed those reported for labeling fully folded C165S AhpC-SOH with nucleophilic 1,3-dicarbonyl probes such as dimedone (e.g. 0.10 m−1·s−1 for the slowest electrophile IAM versus 0.05 m−1·s−1 for the dimedone-based DCP-Bio1 at pH 7, each calculated from pseudo-first order rate constants and reagent concentration) . Such comparable rate constants illustrate that blocking reagents and -SOH-targeted probes compete for the oxidized thiol site, emphasizing the risk of quenching cellular -SOH with blocking reagents when added prior to or alongside a 1,3-dicarbonyl probe.
Figure 3. Time course of adduct formation from the reactions of MMTS, NEM and IAM (each 5 mm) with C165S AhpC-SOH (30 μm). Concentrations of adducts were determined based on adduct abundance among the total ion abundances of the prominent species in the mass spectra as shown in Fig. 2. Reactions were performed at room temperature in ammonium bicarbonate buffer (50 mm, pH 7.5).
Download figure to PowerPoint
In addition to sulfenic acid, thiol-blocking electrophiles were screened for reactivity with protein sulfinic acid (AhpC-SO2H) and S-nitrosothiol (AhpC-SNO). AhpC-SO2H was incubated in the presence of each electrophile for 2 h at pH 7.5, then analyzed by ESI-TOF MS (Fig. S2). A small amount of adduct at the expected mass was observed only in the case of NEM, although subsequent digestion and nanoLC-MS analysis showed that the succinimide modification occurred at histidine sites and not at C46. Indeed, slow alkylation of non-cysteine amino acid residues by NEM has been reported at longer reaction times [28-31]. AhpC-SNO was generated via trans-nitrosation by S-nitrosocysteine (CySNO), then incubated at pH 7.5 in the presence of each electrophile after CySNO removal. Product formation corresponding to electrophile addition to AhpC-SNO was not observed by ESI-TOF MS for the three electrophiles surveyed (Fig. S3).
Reduction of oxidized species and tagging of the nascent thiols is performed subsequent to thiol blocking in the biotin-switch assay and related methods for detecting oxidized cysteine (Fig. 1). This approach is often tailored to detection of disulfides (including sites of glutathionylation), sulfenic acids and S-nitrosothiols through use of chemoselective reducing agents . Additionally, reduction chemistry is an integral component of protein sample processing for in-gel and in-solution digestion followed by mass spectrometric analysis. As a result, we investigated the stability of the -SOH/electrophile (-SOR) adducts toward reducing agents at concentrations relevant to the biotin-switch assay and other biochemical techniques in an effort to understand the implications of inadvertent -SOH reactivity with thiol-blocking electrophiles on overall outcome of these assays (Fig. 2A, second step).
Freshly generated AhpC-SOH was first incubated with IAM, NEM or MMTS (5–10 mm) at pH 7.5 for 2 h to maximize the abundance of -SOR. Unreacted electrophiles were removed by gel filtration, and the resulting protein species were treated with biochemical reductants (10 mm) at room temperature. Product ratios were measured at time points ranging from 10 min to 3 h by ESI-TOF MS (Fig. 4, Fig. S4 and Fig. S5). Cleavage of IAM -SOR adducts by equal concentrations of dithiothreitol (DTT), tris(2-carboxyethyl)phosphine (TCEP) and sodium ascorbate proceeded similarly, leading to a 40–50% reduction in adduct concentration (Tables 1 and 2). Succinimide adducts, generated with NEM, were more susceptible to reduction compared with the IAM -SOR adducts. With DTT, there was an ~ 87% loss of adduct upon a 1 h incubation, while with TCEP, the loss of -SOR adduct was ~ 60% (Fig. S4). Least stable were the MMTS adducts, with full reduction by DTT and TCEP, and 85% loss by ascorbate (Fig. S5). DTT and TCEP reductions of all -SOR adducts proceeded with concomitant generation of C165S AhpC-SH. In contrast, all ascorbate-mediated reductions were instead accompanied by an increase in the amount of AhpC-SO2H. Sulfinic acid formation was also observed upon direct treatment of AhpC-SOH with sodium ascorbate (10 mm) under the same reaction conditions, suggesting an ascorbate-mediated oxidation of thiol and sulfenic acid to AhpC-SO2H catalyzed by trace metals in the reaction medium as previously described .
Table 1. Percentage of AhpC-SOR adducts remaining after room temperature incubation (1 h) with reducing agents DTT, TCEP and sodium ascorbate
|Entry||-SOR adduct||+10 mm DTT||+10 mm TCEP||+10 mm ascorbate|
Table 2. Susceptibility of AhpC-SOR (R = succinimide) to reduction by sodium ascorbate (1 and 10 mm). Each reaction contained 1 μm CuSO4 and was incubated for 1 h at room temperature
| ||NH4HCO3 pH 7.5||HEPES pH 7.7|
|AhpC-SOR (% remaining)||38||37||35||39||77||89||77|
Figure 4. ESI-TOF MS demonstrating a, upper panels the formation of sulfenic acid–electrophile (-SOR) adducts, and b, lower panels subsequent reduction by DTT (10 mm, pH 7.5). Adducts in the upper panels were generated via incubation of C165S AhpC-SOH (30 μm) with electrophile (2 mm): IAM (1a), NEM (2a) or MMTS (3a) for 3 h at room temperature. Excess electrophile was removed prior to DTT treatment. The signal at 20 598 amu corresponds to sulfenamide (-SN), and the signals at 20 566 and 20 646 amu (3a,b) correspond to cysteine conversion to dehydroalanine (20 566) and AhpC-SH addition to MMTS (20 646). Signals labeled with single and double asterisks represent 20 632 (-SO2H) and 20 648 (-SO3H) a.m.u., respectively. Evidence for thiol regeneration is indicated by the appearance of a mass peak at 20 600 a.m.u. (-SH), which occurred with concomitant loss of signal corresponding to the -SOR adducts.
Download figure to PowerPoint
The structural identity of the -SOR adducts, whether a sulfoxide or a thioperoxide (also called a sulfenate ester, or a thiosulfinate when derived from MMTS), is not well-defined. Structures of both types have been proposed for nucleophilic addition of sulfenic acids, and formation of the thermodynamically stable sulfoxide product is the favored proposed mechanism . However, reduction of -SOR adducts to thiols by these relatively mild reductants suggest a less stable bond connectivity and/or the possibility of β-elimination. Sulfoxide reduction to thioethers is well-defined enzymatically and has been documented for chemical reductants such as TCEP; reduction of these species to thiols typically requires boron or other Lewis acids [33, 34]. With the exception of MMTS adduct reduction, the products observed subsequent to -SOR reduction do not suggest that the β-elimination mechanism dominates, as expected products include NEM-blocked cysteine and the conversion of cysteine to dehydroalanine . The molecular structure and stability of protein -SOR adducts are certainly site-dependent and warrant further investigation.
The accuracy of biochemical methods for cysteine oxoform detection relies profoundly on the chemoselectivity of both the thiol-blocking and chemical reduction steps prior to downstream analysis (spectroscopy, western blotting, mass spectrometry, etc.). The rates of -SOH reaction with electrophiles and subsequent adduct reduction are undoubtedly protein-dependent, and are expected to vary according to the microenvironment of each -SOH moiety, complicating the view of how this chemistry ultimately affects the interpretation of results obtained using thiol-blocking and reduction techniques. Clearly, the effects of the sulfenic acid reactivity described here on the outcomes of oxoform detection will vary according to specific experimental details, most critically sample complexity and cysteine accessibility. A few general scenarios are examined below based on the chemical data obtained with AhpC.
Implications for the global detection of reversibly oxidized cysteine species
In the case of biotin-switch techniques, thiol blocking is immediately followed by a presumably chemoselective reduction of a given oxidation product to generate nascent thiols (Fig. 1) . We have demonstrated that the thiol-blocking step also consumes the sulfenic acid of AhpC, and anticipate that similar cross-reactivity may exist with other protein sulfenic acids. Reductants such as DTT and TCEP, which are often utilized to quantify the content of reversibly oxidized cysteines, cleave the majority of -SOR species to an extent determined by the blocking reagent and the reduction conditions to generate a pool of cysteine thiols with varying proportions originating from disulfide (-SS-), -SOH and -SNO species (Fig. 1, path 2a) . In this experimental scenario, and given the low preponderance of the -SOH species relative to disulfides, the -SOR chemistry is not expected to significantly affect the global qualitative or quantitative analysis of reversibly oxidized cysteines in the cellular proteome. (Note: This is aside from caveats related to indirect methods of detection: e.g. incomplete blocking of free -SH during the first workflow step.)
Implications for detection of -SOH species in complex samples using the biotin-switch assay
The biotin-switch assay has been tailored for -SOH detection via use of sodium arsenite as reductant, and, although unexplored here, this reagent is expected to similarly reduce -SOR species based on its demonstrated selectivity for S-O bonds [38-40]. In this case, and given the caveats described above, the -SOR chemistry will not affect the qualitative identification of -SOH sites, but may affect the quantitative analysis by underestimating the -SOH content. However, it is possible that -SOR formation in this assay could enable the identification of -SOH sites otherwise lost to adventitious hyperoxidation during sample workup had they not been alkylated in the initial step.
Implications for the direct detection of -SOH species in complex samples using chemical probes
Uncovering protein sulfenic acids by mass spectrometry or western blotting involves use of nucleophilic 1,3-dicarbonyl probes, often in conjunction with a thiol-blocking reagent [2, 41-44]. As noted above, the cross-reactivity of the -SOH species and nucleophilic -SOH probes with thiol-capturing electrophiles was suspected of decreasing the overall yield of probe incorporation in our cellular oxidation studies. Indeed, incubation of equimolar amounts of dimedone with IAM, NEM and MMTS resulted in formation of covalent adducts as observed by LC-MS (Fig. S6). With the relative abundance corresponding to a 1 : 1 stoichiometric addition, proposed structures involve the new bond connecting the nucleophilic methylene C2 of dimedone to the blocking group.
These observations led to an experiment designed to investigate the cross-reactivity among several probes and electrophiles when one reactive partner is protein-bound. Samples of AhpC-SOH (30 μm) were incubated separately with dimedone and the 1,3-cyclopentanedione probe BP1 (2 mm each) for 1 h to generate protein thioethers . Excess probe was removed, and the proteins were treated separately with either IAM or NEM (10 mm each) for 45 min, then digested using trypsin followed by nanoLC-MS analysis. The MS2 spectra acquired for the cysteine-containing peptide in each sample provided evidence for cysteine modified by dimedone or BP1 only; no evidence for subsequent addition of IAM or NEM was observed with either of the probes once bound to AhpC. Such results corroborate the dimedone reactivity described above, suggesting that the 1,3-dicarbonyl enol attacks both the -SOH and the electrophiles at C2. Once the C2 site had formed a thioether at a protein cysteine residue, the dicarbonyl adduct was no longer reactive with electrophiles despite opportunities to form other less reactive enols.
As a result of the reactivity of thiol-blocking electrophiles with both -SOH itself and the 1,3-dicarbonyl probes used for its labeling, adjustments to the workflows for -SOH labeling and detection are required. There is a clear need for more chemoselective thiol-blocking reagents. A recent report describes the use of a new thiol-quenching reagent that has been shown to circumvent the cross-reactivity of IAM and NEM with non-cysteine amino acids, but this reagent has not yet been tested with oxidized cysteine species . While new approaches are in development, the addition of thiol-blocking reagents after -SOH tagging with the 1,3-dicarbonyl probes is recommended. More stringent efforts to prevent lysis-associated oxidation (e.g. lowering the lysis buffer pH, using degassed lysis buffers supplemented with catalase and superoxide dismutase, etc.) are also critical.
Implications for the detection of -SNO species in complex samples using the biotin-switch assay
The replacement of DTT/TCEP with ascorbate as the reductant has been widely utilized as a method for indirect detection of protein S-nitrosothiols (Fig. 1) . The limitations of this technique have been critically investigated in recent years, with particular attention given to the chemoselectivity of ascorbate and its Cu(I) dependency in the reduction of S-nitroso moieties [47-49]. Cumulatively, these studies question the capacity of ascorbate to reduce -SNO species without affecting disulfide content and generating false-positive identifications of -SNO species.
Here, ascorbate is shown to be a promiscuous reductant with respect to -SOH and the -SOR adduct (Fig. S5 and Table 1). These results suggest that the use of ascorbate in the presence of protein -SOH may well lead to false-positive identifications of -SNO content by generating cysteine thiol sites derived from -SNO, -SOH and -SOR functional groups with an inability to distinguish their precursor states. In fact, the potential for ascorbate to reduce protein sulfenic acids has been long established, although evidence to the contrary has also been reported [50-52]. Nevertheless, the current results clearly indicate conversion of -SOH species to -SOR during the thiol-blocking step of the biotin-switch assay. We next investigated the chemical response of -SOR to ascorbate reduction under conditions described in the biotin-switch workflow.
In contrast to earlier biotin-switch assays requiring 10–20 mm ascorbate, more recent investigations have uncovered a role for Cu(I) ions in promoting nitrosothiol reduction, allowing a decrease to 1 mm ascorbate with catalytic amounts of Cu(I) or Cu(II) . The C165S AhpC-SOR, where R is N-ethylsuccinimide, was treated with 1 mm ascorbate in the presence of 1 μm CuSO4 at pH 7.5 for 1 h in the absence of light. In comparison to the 73% decrease of this -SOR species when 10 mm ascorbate is used (Table 1), ESI-TOF MS analysis of the 1 mm ascorbate reduction showed less -SOR reduction at the lower ascorbate levels (38% remaining Table 2). Again, AhpC-SO2H was the primary product of -SOR reduction (and oxidation) by ascorbate, and neither the degree of -SOR cleavage nor the amount of -SO2H formed was altered by performing the reduction in the presence of catalase and superoxide dismutase.
A notable difference between reported tag-switch procedures and the -SOR investigations described here is the composition of reaction buffer for ascorbate reduction. The reduction step in the tag-switch workflows is often performed in HEPES in the presence of 1% SDS, whereas -SOR chemistry has been largely investigated in MS-compatible NH4HCO3. To assess buffer effects on ascorbate reduction, the succinimide-based AhpC-SOR was incubated with 10 mm ascorbate in HEPES (25 mm, pH 7.7) or NH4HCO3 (50 mm, pH 7.5). Interestingly, the reductive cleavage of -SOR by ascorbate is weakened in HEPES buffer compared to NH4HCO3: 33% cleavage in HEPES compared to 73% cleavage in NH4HCO3. However, the AhpC-SOR formation was unchanged when comparing reactions performed in HEPES versus NH4HCO3 buffers. The use of denaturing conditions (1% SDS) did not affect -SOR stability in HEPES as measured by ESI-TOF MS. Furthermore, the principal product of AhpC-SOR reduction by ascorbate in the presence or absence of Cu(I) was sulfinic acid, which is unreactive toward N-[6-(biotinamido)hexyl]-3′- (2′-pyridyldithio)propionamide (biotin-HPDP) and other biotinylated electrophiles (Fig. S5). Together, these results imply that -SOR reduction by ascorbate may affect the analysis of S-nitrosation in tag-switch assays. A full biotin-switch assay with western blot detection was performed using MS-identified oxoforms of AhpC to further test these conclusions.
Cysteine oxoforms of interest, -SOH, -SOR (R = N-ethyl succinimide) and -SNO, were generated using mutant AhpC, and the mass of each modified protein species was confirmed by ESI-TOF MS (Fig. S7). AhpC-SH (40 μm) was also treated with biotin-HPDP or NEM (5 mm) as positive (lane 1) and negative controls (lane 3), respectively. Analysis by ESI-TOF MS confirmed that the C46 thiol was fully blocked in both control samples. Biotin-switch blocking was performed with 20 mm NEM and 2.5% SDS as described previously [53, 54]. After precipitation, resuspended samples were treated in the dark with ascorbate (1 mm) in the presence of CuSO4 (1 μm) and 1% SDS. Non-reducing SDS/PAGE followed by western blottting shows that, together with -SNO (lane 6), both -SOH and pre-generated -SOR (lanes 4 and 5, respectively) are reduced by ascorbate under biotin-switch conditions, confirming the extensive cross-reactivity of these species with ascorbate and illustrating the potential for misinterpretation using this assay (Fig. 5). These results also reveal a relatively low overall yield of the biotin-switch assay, perhaps as a result of nascent AhpC-SH oxidation to AhpC-SO2H out-competing thiol capture by biotin-HPDP.
Figure 5. Biotin-switch assay for C165A AhpC reduced and oxidized cysteine species provides evidence of biotinylation at levels similar to those observed for AhpC-SNO. Where indicated, the initial blocking step was performed using NEM (20 mm); thiols generated by sodium ascorbate (1 mm) reduction were tagged with biotin-HPDP (5 mm). Steps 1 and 2 reflect the order of treatment. ESI-TOF MS spectra confirming the identities of the initial AhpC species are shown in Fig. S7.
Download figure to PowerPoint
Despite the wide use of biotin and other tag-switch techniques, the fate of -SOH in such approaches has been largely under-investigated other than in the study by Forrester et al.  in which the biotin-switch specificity was assessed using cysteine oxoforms of human protein tyrosine phosphatase 1B (PTP1B). Oxidatively modified PTP1B, presumably a mixture of -SOH/-SN, was confirmed by a sharply decreased phosphatase activity, of which < 40% was restored upon DTT treatment, suggesting the presence of hyperoxidized enzyme as well as the reversibly oxidized species . The lack of evidence for biotinylated PTP1B in the described biotin-switch assay with up to 100 mm ascorbate is perhaps a result of the mildly oxidized enzyme's propensity to reside in the sulfenamide (-SN) state, as has been described in great detail [21, 55]. Our results together with these previous observations illustrate the critical importance of careful control experiments and the challenge of assessing specificity in such indirect detection assays.
In summary, we have demonstrated the promiscuity of several thiol-blocking reagents by detailing their chemical reactivity with the sulfenic acid of peroxiredoxin AhpC. Treatment of AhpC-SOH with IAM, NEM and MMTS at concentrations relevant to the biotin-switch assay and other biochemical methods produces covalent adducts that are partly or fully susceptible to reduction and thiol generation. This reactivity has profound effects on approaches for detecting cellular cysteine oxidation, in particular methods for selective detection of labile signaling species such as sulfenic acid and S-nitrosothiol.