Identification of Nitroxyl-induced Modifications in Human Platelet Proteins Using a Novel Mass Spectrometric Detection Method*S

Nitroxyl (HNO) exhibits many important pharmacological effects, including inhibition of platelet aggregation, and the HNO donor Angeli's salt has been proposed as a potential therapeutic agent in the treatment of many diseases including heart failure and alcoholism. Despite this, little is known about the mechanism of action of HNO, and its effects are rarely linked to specific protein targets of HNO or to the actual chemical changes that proteins undergo when in contact with HNO. Here we study the presumed major molecular target of HNO within the body: protein thiols. Cysteine-containing tryptic peptides were reacted with HNO, generating the sulfinamide modification and, to a lesser extent, disulfide linkages with no other long lived intermediates or side products. The sulfinamide modification was subjected to a comprehensive tandem mass spectrometric analysis including MS/MS by CID and electron capture dissociation as well as an MS3 analysis. These studies revealed a characteristic neutral loss of HS(O)NH2 (65 Da) that is liberated from the modified cysteine upon CID and can be monitored by mass spectrometry. Upon storage, partial conversion of the sulfinamide to sulfinic acid was observed, leading to coinciding neutral losses of 65 and 66 Da (HS(O)OH). Validation of the method was conducted using a targeted study of nitroxylated glyceraldehyde-3-phosphate dehydrogenase extracted from Angeli's salt-treated human platelets. In these ex vivo experiments, the sample preparation process resulted in complete conversion of sulfinamide to sulfinic acid, making this the sole subject of further ex vivo studies. A global proteomics analysis to discover platelet proteins that carry nitroxyl-induced modifications and a mass spectrometric HNO dose-response analysis of the modified proteins were conducted to gain insight into the specificity and selectivity of this modification. These methods identified 10 proteins that are modified dose dependently in response to HNO, whose functions range from metabolism and cytoskeletal rearrangement to signal transduction, providing for the first time a possible mechanistic link between HNO-induced modification and the physiological effects of HNO donors in platelets.

which could be used as an antialcoholic treatment (14,18,19); and pretreatment of ischemic (oxygen-depleted) tissues with HNO has been shown to protect against ischemia-reperfusion toxicity (20). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH), a key enzyme in the carbohydrate metabolism pathway, has also been shown to be potently inhibited by HNO both in vitro (19) and in vivo (21), an effect thought to occur through the direct modification of its active site cysteine. At high concentrations (2-5 mM), HNO has been shown to be cytotoxic by eliciting DNA strand breaks and glutathione depletion, causing cellular toxicity due to oxidative protein damage (22). However, this toxicological effect is only relevant if physiological HNO levels are high, and it has thus far not been demonstrated to have any in vivo relevance (23).
These HNO-mediated pharmacological effects are dramatically different from those of NO (11) most likely because HNO tends to be much more thiophilic with cysteines being the major site of biochemical reactivity (24 -26). Therefore, it is no surprise that NO and HNO tend to have different targets. For example, in the vascular system, HNO can act through a cAMP signal transduction pathway, whereas the vascular activity of NO is primarily due to an elevation in cGMP (27).
Although a number of pharmacological and toxicological effects have been shown for HNO, the underlying mechanisms of action are largely unknown. HNO and cysteine are known to react to produce non-cross-linked sulfinamides and cause disulfide formation, and HNO can react with metals/ metalloproteins and oxygen and participate in reduction/oxidation reactions (23,25,26). However, the molecular targets of HNO have yet to be linked to its pharmacological and toxicological effects.
Here we describe a mass spectrometry-based method for the analysis of the major type of biologically relevant HNO reaction, the reaction with the thiol on cysteines to produce non-cross-linked sulfinamides as well as disulfide linkages (23,25,26). Although disulfides are produced through many different pathways, non-cross-linked sulfinamides are exclusively produced by HNO and can thus be used to analyze for the presence of HNO and its effects on cysteine-containing proteins. As well, the sulfinamide modification imparts a specific mass change to cysteines making sulfinamide analysis, and indirectly HNO analysis, very amenable to investigation by mass spectrometry. The sulfinamide modification has been observed in MS spectra (28), and in a recent mass spectrometric analysis, Shen and English (29) attributed a mass shift of 65 Da on prominent y-ions upon low energy CID to the elimination of the sulfinamide moiety from the molecule in their mass spectrometric comparison of nitroxyl products formed with free and protein-based cysteines. Here we investigate this mass shift and the formation of a previously unstudied neutral loss to determine an efficient method for the identification of the sulfinamide modification and demonstrate its utility on a sample generated by treatment of live platelets immediately post-isolation, that is ex vivo, with HNO.

EXPERIMENTAL PROCEDURES
Reagents-The peptide MHRQETVDCLK-NH 2 was provided as a gift from Phil Owen of the Biomedical Research Centre, and peptide EKPLQNFTLCFR-NH 2 was purchased from Bachem (Bubendorf, Switzerland). AS (Na 2 [ONNO 2 ]) was purchased from Cayman Chemicals (Ann Arbor, MI), and sequencing grade trypsin, formic acid, and acetonitrile were purchased from Fisher Scientific. All other reagents were purchased from Sigma-Aldrich.
Mass Spectrometry of Test Peptides-Analysis of the HNO-modified peptides was performed using the following instruments: a nano-ESI (nESI) source triple quadrupole instrument where the third quadrupole (Q3) has linear ion trapping capabilities (2000 Q TRAP), an nESI quadrupole time-of-flight (Q STAR XL) mass spectrometer, and a MALDI-TOF/TOF 4700 Proteomics Analyzer (all Applied Biosystems/MDS Sciex, Concord, Ontario, Canada). Alternatively an nESI linear ion trap coupled to a Fourier transform ion cyclotron resonance mass spectrometer (LTQ FT-ICR) with electron capture dissociation (ECD) capabilities and an nESI Orbitrap instrument were used (Thermo Electron Corp., Bremen, Germany). All analyses were performed with 1-5 M solutions of the modified peptides.
Nitroxylation of Test Peptides-A stock solution of 100 mM AS was prepared in 10 mM NaOH. Nitroxylation was performed on 10 nmol of either MHRQETVDCLK-NH 2 or EKPLQNFTLCFR-NH 2 with 1 l of AS stock solution and 99 l of 15 mM Tris-HCl (pH 7.4) buffer, producing a final AS concentration of 1 mM. Reactions were carried out for 25 min at room temperature. The peptides were then separated from the reactants, purified with C 18 ZipTips (Millipore, Billerica, MA), and reconstituted in 50% acetonitrile, 5% formic acid when analyzed by electrospray mass spectrometry or were spotted onto a MALDI target using the dried droplet method and ␣-cyano-4-hydroxycinnamic acid matrix.
Kinetics Analysis-The kinetics of the reaction of peptides MHRQETVDCLK-NH 2 and EKPLQNFTLCFR-NH 2 with HNO was investigated on a Q STAR XL using an ion accumulation time of 1 s, leading to the acquisition of one kinetics measurement per second. HNO is produced from the decomposition of AS upon its addition to the buffered peptide reaction solution with a half-life of 4.4 min at pH 7.4 and 22°C (30,31). For all kinetics experiments, a 40:1 molar ratio of AS to peptide was used with a final AS concentration of 2 mM. As there was a 40ϫ excess of AS to peptide in the reaction mixture, calculations using the first order rate of HNO decay while ignoring HNO consumption by reaction with thiols or by dehydrative dimerization of HNO showed that HNO may have been in excess compared with the peptide concentration as early as 10 s after AS addition.
Two different setups were required to monitor the kinetics of the reaction from 1.7 s to 15 min. For the 1-15-min time points, the reaction solution was prepared by adding AS to 2.5 nmol of either MHRQETVDCLK-NH 2 or EKPLQNFTLCFR-NH 2 in 12.5 mM Tris-HCl (pH 7.4), 20% methanol; inserted into a nanospray emitter; and then analyzed on the mass spectrometer. For such a process, it took between 1 and 1.08 min from AS mixing into the reaction solution to data collection. Three data sets were collected and averaged, and a five-point moving average was applied to the data. For the shorter time points, kinetics data in the range of 1.7-90 s were collected based on a continuous flow setup as described previously by Konermann et al. (32) with slight modifications. Briefly two syringes were operated simultaneously with syringe 1 (flow rate of 32 l/min) containing the buffered peptide mixture and an internal standard (osteocalcin fragment 7-19) and syringe 2 (flow rate of 1 l/min) containing the AS solution. Each syringe was connected to a fused silica capillary (100-m inner diameter, Polymicro Technologies, Phoenix, AZ) that was then connected together by a splitter on the electrospray source (MDS Sciex) to a third reaction capillary (also 100-m inner diameter fused silica) that led the sample to the tip of the emitter. The length of the third capillary was varied from 627 to 36 cm corresponding to reaction times from 89.5 to 5.1 s (32). To reach reaction times as low as 1.7 s, the flow rate of the two syringes was increased proportionately to obtain total flow rates up to 100 l/min. An accumulation time of 1 s per mass spectrum with a total acquisition time of 1 min was collected for one data set. The internal standard, osteocalcin fragment 7-19, was used to normalize the intensities of the precursor, intermediates, and final product as they may vary with the flow rate.
Treatment of Platelets with Angeli's Salt-Ethical approval for this study was granted by the University of British Columbia Research Ethics Board, and informed consent was granted by the donors. Whole blood was drawn from the antecubital vein of healthy human volunteers into 0.15% (v/v) acid-citrate-dextrose anticoagulant. Platelets were isolated by centrifugation, washed twice in physiological buffer (10 mM trisodium citrate, 30 mM dextrose, 1 IU/ml apyrase), and resuspended at physiological concentration (200 -350 ϫ 10 9 /liter) in Tris-buffered saline containing 5 mM EDTA. Platelets were treated with either AS (10 M, 100 M, 1 mM, and 10 mM) from a stock of 1 M AS in 10 mM NaOH or a vehicle control for 2 min at room temperature. Samples were spun at 1000 ϫ g for 3 min at room temperature to pellet the platelets. Platelet pellets were resuspended in 1ϫ lysis buffer (20 mM Tris, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM ␤-glycerol phosphate, 1 mM Na 3 VO 4 , 1ϫ protease inhibitor mixture), and lysates were immediately snap frozen in liquid N 2 .
Electrophoresis and Proteolytic Digestion-Samples were thawed slowly on ice, mixed with non-reducing sample buffer, and separated by one-dimensional (1D) SDS-PAGE. Gels were stained with Coomassie Brilliant Blue, and bands of interest were excised. Tryptic in-gel digestions were performed overnight at 37°C without reducing or alkylating the sample (33). Following digestion, peptides were extracted for MS analysis.
Liquid Chromatography-Tandem Mass Spectrometry-Identification of peptides with HNO-induced modifications from the in-gel digests of samples treated with 10 mM AS was performed by nano-HPLC MS/MS on an Agilent 1100 instrument (Agilent, Santa Clara, CA) coupled to an LTQ-Orbitrap using a 15-cm-long, 75-m-inner diameter fused silica column packed with 3-m-particle size reverse phase (C 18 ) beads (Dr. Maisch GmbH) using water:acetonitrile:formic acid as the mobile phase with gradient elution. Identification of modified peptides was performed by extracting the Mascot generic format files from the MS data using DTA Super Charge, part of the MSQuant open source project (SourceForge, Inc.) and then searching them against the human Swiss-Prot database (v.54.5; 589,473 sequences) using Mascot v.2.1 (34). The following search criteria were used: trypsin cleavage specificity with up to one missed cleavage site, no fixed modifications, variable modifications of oxidized methionine and dioxidized cysteine (sulfinic acid), Ϯ10-ppm peptide tolerance, Ϯ0.6-Da MS/MS tolerance, and the scoring scheme was ESI-TRAP. All MS/MS spectra of peptides that were identified as containing sulfinic acid modifications by Mascot were inspected manually to confirm the peptide assignment. These candidate modified peptides were then selected for multiple reaction monitoring (MRM) analysis for the dose-response experiment with the product ion spectra from the LTQ of the Orbitrap being used to determine the MRM transitions.
The dose-response analysis was completed with MRM experiments performed by nano-HPLC MS/MS on an Ultimate pump (LC Packings, Sunnyvale, CA) using a 15-cm-long, 75-m-inner diameter, 3-m-particle size reverse phase (C 18 ) column (LC Packings) coupled to an ESI 2000 Q TRAP. Each putative modified protein identified by the Orbitrap was then monitored by four MRM transitions (Table I): one to monitor the precursor ion with no fragmentation in the collision cell and three MRM transitions to monitor three different fragment ions produced from fragmentation of the precursor in the collision cell. Peak areas from these three fragment ions were collected from the extracted ion chromatograms and then averaged to increase selectivity as well as to allow for relative quantitation of the amount of sulfinic acid modification at different doses.
Two non-modified peptides from the same protein, each with two MRM transitions, were monitored to serve as internal standards: one MRM transition monitored the precursor ion (no fragmentation), and one MRM transition monitored the most intense fragment ion. Peak areas from the two precursor ions and two fragment ions were collected from the extracted ion chromatograms and normalized to the most intense peak area of the five dose-response samples for each transition. These four values were averaged to obtain a correction factor (CF) for the variation in absolute protein quantity from gel band to gel band caused by unequal protein loading or incomplete band excision between gel lanes. The average fragment peak area (A) of the modified precursor was divided by the CF to calculate the corrected average fragment peak area (A/CF). A noise correction was performed for each peak area used to calculate A and CF by drawing a base line from the midpoint of the noise on the left of the detected signal to that on the right. Due to integration of fluctuating noise levels, this process would result in non-zero values of A/CF even in the absence of a discernable peak.
Three dose-response data sets were collected by reacting three sets of platelets, collected from different donors, on different days using different batches of stock AS solution. To account for instrumental variation from data set to data set, the corrected average fragment peak areas (A/CF) for each data set were normalized to the most intense A/CF of the five dose-response samples of that data set with the maximal A/CF being set to 1. Three data sets were averaged and plotted along with the S.E.

Mass Spectrometric Method Development for HNO-induced Modification Analysis
Examination of the Products of the Reaction between HNO and Cysteine-In the initial experiments examining the reaction of HNO with cysteine-containing peptides, the major products, their relative abundance, and stability were investigated. To this end, the peptides MHRQETVDCLK-NH 2 and EKPLQNFTLCFR-NH 2 were reacted with AS and analyzed by mass spectrometry. As reported previously (25) and shown in Fig. 1, both the sulfinamide and disulfide modification were produced upon cysteine reaction with HNO (Scheme 1).
According to Scheme 1, reaction of the cysteine with HNO generates both the sulfinamide and disulfide modifications (products 4 and 5, respectively) through intermediates (25). As these products and intermediates have different masses, they can be distinguished by mass spectrometry (Fig. 1, a and b). If one or more of these intermediates are long lived, it could lead to a decrease in the final sulfinamide or disulfide yield, making the modification analysis more difficult. Thus, kinetics studies were conducted to determine whether the intermediates are long lived.
Nitroxylation of Cysteine-containing Peptides: Kinetics Analysis-The kinetics of the HNO/cysteine reaction was monitored mass spectrometrically on a Q STAR XL (Fig. 1) by examining the reaction of HNO with the cysteine-containing peptide MHRQETVDCLK-NH 2 . The kinetics of the cysteine/ HNO reaction was monitored from 1.7 s to 15 min using two different experimental setups. Fig. 1c displays kinetics data for the first 90 s of the reaction, whereas Fig. 1d tracks the reaction from 1 to 15 min. Within 1 min, the intensity of the unmodified peptide had dropped dramatically (Fig. 1c). From as early as 5 s, significantly detectable quantities of the ϩ14-Da intermediate (2; Scheme 1) and the final sulfinamide product (4) were produced. As the reaction proceeded, the intensities of both the ϩ14-Da intermediate (2), the final sulfinamide (4), and the disulfide (5) increased. At ϳ85 s, the sulfinamide (4) began to appear at a higher intensity than the ϩ14-Da intermediate (2)   The longer time points in the HNO reaction are shown in Fig. 1d. By 1.2 min into the reaction, intermediate 2 reached a maximum intensity and was thereafter converted to the sulfinamide final product (4) at a higher rate than it was produced. Within a time span of ϳ10 min, the intensities of the unmodified peptide (0) and intermediate 2 (ϩ14 Da) had dropped considerably, whereas the final disulfide (5) and sulfinamide (4) products had reached a plateau. As the species generating the ϩ14-Da peak was consumed in the reaction, it could not have been an intramolecular sulfinamide (29) but instead was an intermediate that was converted to the sulfinamide (4).
Intermediate 1 (Scheme 1; ϩ31 Da), the N-hydroxysulfenamide, could not be monitored as the mass shift is equal to the sulfinamide (4). Interestingly although this species has been implicated in both disulfide and sulfinamide formation, it has not been successfully isolated or characterized (25,35).
Because intermediate 3 (Scheme 1; ϩ32 Da) is 1 Da heavier than the sulfinamide (4; ϩ31 Da), it causes a nominal mass interference as the isotopes of the sulfinamide (4) modification ( Fig. 1b) interfere with the detection of intermediate 3, thus inhibiting its direct analysis. Therefore, the ratio of the ϩ31to ϩ32-Da peaks was monitored as a change in the ϩ31/ϩ32 ratio is considered indicative of the formation of intermediate 3. The ratio was plotted to determine the abundance and longevity of this intermediate (Fig. 1e). The ratio of the ϩ31/ ϩ32-Da sulfinamide isotopes stays constant throughout the experiment indicating that the peaks are the first and second isotopes of the sulfinamide (4) rather than the sulfinamide and intermediate 3. Thus, there was no significant buildup of intermediate 3, indicating that it is short lived.
Taken together, these results indicated that the sulfinamide (4) is the final product in the reaction as none of the intermediates 1-3 are seen past 12-14 min of the reaction, which is advantageous for monitoring the sulfinamide modification. Similar kinetic behavior was observed for the longer time points of the second peptide, EKPLQNFTLCFR-NH 2 , suggesting that the reaction on peptides is largely governed by the availability of HNO rather than the peptide sequence.

Identification of Characteristic Fragmentation Patterns of Nitroxylated Peptides by Tandem Mass
Spectrometry-A recent study attributed a mass shift of 65 Da on prominent y-ions upon low energy CID of a multiply charged peptide to the elimination of the sulfinamide moiety from the molecule (29). To determine whether, along with this mass shift, the sulfinamide modification fragmented to produce characteristic marker ions, product ion analyses of peptides EKPLQNFTLC*FR-NH 2 and MHRQETVDC*LK-NH 2 (C* denotes the presumed cysteine modification) were performed. Product ion spectra of the singly, doubly, and triply charged precursor ions were collected and analyzed (Fig. 2).
Fragmentation of singly charged EKPLQNFTLC*FR-NH 2 ( Fig. 2a) and MHRQETVDC*LK-NH 2 (Fig. 2b), produced from MALDI-TOF/TOF PSD, led to a low yield of y-and b-sequence ions. For both peptides, a 65-Da neutral loss of HS(O)NH 2 from the sulfinamide on the precursor ion, believed to be produced from the sulfur-mediated abstraction of the ␣-hydrogen, was observed. However, there was also the appearance of a 66-Da neutral loss in these spectra.
On closer inspection, an unusually large second isotope peak, 1 thomson heavier than the monoisotopic peak of the sulfinamide-modified precursor, suggested that an additional species may be present. Indeed any fragment ion containing a modified cysteine also showed this pattern, whereas fragment ions that either did not contain the cysteine or had undergone a neutral loss showed isotope ratios that corresponded to a regular amino acid sequence. The cause of this isotope ratio skewing on ions containing the modified cysteine is hypothesized to be due to the additional presence of a sulfinic acid modification generated upon conversion of the sulfinamide via a deamidation reaction as illustrated in Scheme 2. Upon collision-induced activation, this species would then produce a neutral loss of HS(O)OH (66 Da). This appears to be the case, as the complex isotope pattern observed for the precursor ions of peptides MHRQETVDC*LK-NH 2 and EKPLQNFTLC*FR-NH 2 was not apparent after the neutral loss and was replaced by a peak with expected isotope ratios 65 Da lower than the SCHEME 1 monoisotopic peak and 66 Da lower than the large second isotope peak.
Fragmentation of the doubly charged (Fig. 2, c and d) and triply charged (Fig. 2, e and f) precursors, both obtained from a Q STAR XL, led to neutral losses from the precursors as well as a greater production of y-and b-sequence ions. Similar to what was observed in the MALDI-TOF/TOF PSD spectra, neutral losses of both 65 Da from the sulfinamide-modified precursor and 66 Da from the sulfinic acid-modified precursor were apparent. Furthermore when the singly, doubly, and triply charged forms of the unmodified precursor were analyzed, no 65-or 66-Da neutral losses or mass shifts were observed (data not shown). The neutral losses from the precursor ions, the mass shifts on the fragment ions, and y-and b-fragment ions from the doubly and triply charged precursor ions were all reproduced when analyzed It is worth noting that the difference between sulfinamide and sulfinic acid and the corresponding change in proton affinity of the precursor ion seemed to influence the overall charge state of the peptides. The ratio of the first and second isotope was markedly different between the ϩ2 and ϩ3 charge states of a given modified peptide with the second isotope (sulfinic acid modification) being considerably more prevalent in the lower charge state. This effect complicates modification quantification as the ratio of sulfinamide/sulfinic acid modification is not accurately displayed in any one charge state but requires complete analysis of all charge states.
Overall these findings suggest that the neutral loss of HS(O)NH 2 or HS(O)OH is readily apparent and noticeable for low charge state precursors, indicating that neutral loss monitoring may be a preferred method of detection for HNOinduced modification but would be hampered by the difficulty of quantifying the extent of modification when analyzing only one charge state. To determine the fragmentation characteristics of the modification, collision energy profiling of the neutral loss fragment ion [M ϩ nH Ϫ NL (neutral loss)] nϩ and precursor ion was performed on a Q STAR XL. The neutral loss began to be produced at relatively low collision energies (ϳ30 eV) with maximum production occurring between 60 and 70 eV (supplemental Fig. 2). Initial neutral loss production at this low collision energy indicates that the characteristic fragmentation is amenable to any type of MS instrument as it is available by low and medium energy conditions.
Determination of the Site of Sulfinamide Modification on Nitroxylated Peptides Using Tandem Mass Spectrometry-As the CID product ion spectra (Fig. 2) lacked a complete b-/yion series, ECD was used on a FT-ICR mass spectrometer to verify that the cysteine is in fact the modified residue. Because post-translational modifications often remain intact after peptide backbone fragmentation by ECD, it is a very useful fragmentation technique for the localization of modification sites (36). ECD product ion spectra were collected for triply charged MHRQETVDC*LK-NH 2 (Fig. 3a), doubly charged EKPLQNFTLC*FR-NH 2 (Fig. 3b), and doubly charged MHRQETVDC*LK-NH 2 (Fig. 3c). A virtually complete c-ion series was observed for doubly and triply charged MHRQETVDC*LK-NH 2 (Fig. 3, a and c). All three precursor ions fragmented to produce c-ions on both sides of the cysteines (c 8 /c 9 in Fig. 3, a and c, and c 9 /c 10 in Fig. 3b) with a mass difference of 135.01 Da between c-ions for MHRQETVDC*LK-NH 2 and 134.01 Da between c-ions for EKPLQNFTLC*FR-NH 2 . The 134.01 Da corresponds to the mass of cysteine plus HNO or alternatively to the mass of sulfinamide-modified cysteine, whereas the 135.01 Da corresponds to the mass of a sulfinic acid-modified cysteine.
To unambiguously determine the source of the characteristic neutral loss, a comparative analysis of the fragmentation pattern of the intact precursor ion ([M ϩ 2H] 2ϩ ) and the product ion of the neutral loss ([M ϩ 2H Ϫ HS(O)NH 2 ] 2ϩ ) of peptide EKPLQNFTLC*FR-NH 2 was performed. This was sufficient to localize the source of the neutral loss to the cysteine amino acid as the y-, b-, and other fragment ions demonstrated a mass shift to occur on this residue upon neutral loss (Fig. 3d). Taken together, this verified cysteine thiols as the sole functional group that is modified by HNO while excluding other functional groups such as amines as possible alternative reaction sites and demonstrated that sulfinamide and sulfinic acid modifications are the source of the neutral losses of 65 and 66 Da, respectively, identifying these as suitable means of specifically detecting these modifications.

Targeted Proteomics Study for Method Validation: Analysis of Nitroxylated GAPDH from Platelets
Optimization of Experimental Conditions-The treatment of platelets with AS and subsequent mass spectrometric analysis of platelet proteins were performed to examine HNOinduced modification in an ex vivo setting. To obtain modified peptides in a peptide mixture of limited complexity, a standard proteomics work flow was used to separate and digest proteins from platelet lysates. Both protein separation and digestion typically involve reduction of the sample to generate free thiols. To determine whether the modification could withstand the reducing conditions normally encountered in 1D SDS-PAGE and protein digestion, the standard sulfinamide-modified peptides were treated with DTT, which was shown to drastically reduce the abundance of the sulfinamide modification while fully reducing the disulfide bonds (supplemental Fig. 3). It is for this reason that all preparatory steps were performed under non-reducing conditions, and the gel bands were digested without reduction or alkylation.
Analysis of GAPDH-The MS-based detection method was validated by analyzing the protein GAPDH, which is known to be nitroxylated in vitro upon treatment with HNO, generating a sulfinamide-modified cysteine (29). Platelets were treated with 10 mM AS for 2 min, lysed, and separated by 1D SDS-PAGE, and the GAPDH region (ϳ37 kDa) was excised and trypsin-digested. Analysis of the GAPDH peptides was carried out by an information-dependent acquisition whereby an MS scan was followed by a product ion analysis on a Q STAR XL with the most abundant doubly to quadruply charged ions present in the MS scan selected for fragmentation in the product ion scan.
Two of the three cysteines in human GAPDH (Cys-152 and Cys-156) were identified as being disulfide-linked in the tryptic fragment 146 -162 (Fig. 4a). The Cys-247-containing tryptic peptide (235-248) could not be found either as a disulfide or in sulfinamide-modified or unmodified form. However, a peptide 1 Da heavier than the sulfinamide-containing form was consistently observed. Fragmentation of this peptide showed a neutral loss of 66 Da and a fragment ion series matching tryptic peptide 235-248 (Fig. 4b), suggesting the presence of a sulfinic acid modification. Indeed subjecting the sulfinamide-containing peptide MHRQETVDC*LK-NH 2 to the entire sample preparation procedure resulted in a near quantitative mass increase of 1 Da (supplemental Fig. 4) and subsequent neutral loss of 66 Da upon fragmentation, indicating that sample preparation results in deamidation of the modification as outlined in Scheme 2. Although the observed near quantitative conversion of the peptide makes this a plausible pathway, another possible mechanism of generating sulfinic acid modifications is the direct conversion of sulfhydryl groups by oxidants.

Global Proteome Analysis for Discovery of HNO-induced Modifications on Platelet Proteins
In continuation of the GAPDH studies, a complete, global proteome analysis was used to elucidate the role of HNOinduced modification of human platelet proteins. Initially platelet samples were treated with the highest dose of AS (10 mM) for 2 min, then lysed, separated by 1D SDS-PAGE, and in-gel tryptically digested (Fig. 4). The resulting peptides were analyzed by an Orbitrap mass spectrometer and searched using Mascot to determine sulfinic acid-modified peptides and identify the corresponding platelet proteins based on the 32-Da mass shift on cysteine residues (supplemental Table 1). Modified peptides determined from the database searches were verified by manual inspection of the MS/MS spectra to confirm cysteine modification and sequence matching with 22 modified peptides identified in 21 proteins (Table I). Fourteen of the 22 modified peptides also showed the characteristic neutral loss of 66 Da.
Based on these findings, the 21 proteins were further examined to determine the dose dependence of the sulfinic acid modification. The modified peptides that had been identified in the initial high dose study were further analyzed by MRM transitions (Table I) set up to monitor the modified peptides at each dose of AS. Doses included 10 M, 100 M, 1 mM, and 10 mM AS with a buffer-only vehicle control also performed and graphed as 0 mM dose. This approach led to the identification of 10 peptides from 10 proteins that were found to be sulfinic acid-modified by AS in a dose-dependent manner (Table II and Fig. 5). None of the other candidate peptides found by the high dose study could be identified in the MRM transitions as the fragment ions of the modified peptides were not observed possibly because of the differences in sensitivity and fragmentation mechanisms between the mass spectrometers used for the discovery and identification (LTQ-Orbitrap) and the MRM analyses (2000 Q TRAP). Interestingly of the 10 modified proteins identified in the MRM transitions, nine had previously been found to produce the characteristic sulfinic acid neutral loss of 66 Da (Fig. 4).  The 10 proteins identified as modified upon HNO treatment of platelets all showed a dose-dependent increase in the level of sulfinic acid modification with increasing AS concentration as measured by the area of the corresponding fragment ion peaks from the MRM transitions (Fig. 5). The dose-response curves of proteins showing similar response trends have been grouped together in Fig. 5. The most dramatic responses were seen with the platelet surface receptors, glycoprotein 1b␣ and integrin ␤3 (Fig. 5, a and b), which showed a steady increase in modification with increasing doses of AS. Adenylyl cyclase-associated protein 1 and the enzyme pyruvate kinase isozyme M1/M2 were found to have similar concentrationresponse curves upon treatment with AS (Fig. 5c). Both showed slight responses to lower doses of AS and a dramatic effect at the highest AS concentration (10 mM). GAPDH and vinculin were also found to have similar dose-response curves to AS, showing little response at doses of 1 mM and below before climbing steeply at the 10 mM AS dose (Fig. 5d). Finally actin, gelsolin, and ␣-enolase are shown together in Fig. 5e. All showed a trend of increasing modification with increasing Vinculin Involved in cell adhesion; may be involved in the attachment of the actin-based microfilaments to the plasma membrane dose of AS. However, there was greater variation in the data for these three proteins possibly due to a lower intensity of the signals from the modified peptides seen from these proteins.

Method Development and Validation Using a Targeted Proteomics Study to Analyze HNO-induced Modification of GAPDH from Platelets
This study describes the development and application of a mass spectrometric approach for the identification and analysis of HNO-induced protein modification. Initially developed using HNO-modified test peptides, a 65-Da (HS(O)NH 2 ) neutral loss was originally identified. Subsequently a 66-Da (HS(O)OH) neutral loss associated with modified cysteines due to slow conversion of sulfinamide to sulfinic acid was also identified. The approach was then applied to GAPDH, a protein known to be nitroxylated by HNO, extracted from AStreated human platelets. The sample preparation associated with the separation, isolation, and digestion of proteins from a physiologically relevant sample was found to produce a peptide that fragmented to only produce a neutral loss of 66 Da (HS(O)OH) rather than both the 65-Da (HS(O)NH 2 ) and 66-Da (HS(O)OH) neutral losses observed previously in the MS/MS analysis of the modified test peptides. This neutral loss is thought to be the result of quantitative deamidation of the modified residue, a reaction previously proposed for sulfinamide (25) and similar to a process previously observed on other amines such as asparagine and glutamine (37,38).
The deamidation reaction (Scheme 2) is a plausible mechanism as carbon and sulfur have very similar electronegativities, the sulfinic acid produced is a thermodynamically stable product, and an intermediate forming a low energy intramolecular five-member ring transition state, similar to the succinimide intermediate seen with asparagine and glutamine deamidation, can be obtained. This low energy transition state lowers the activation energy of the deamidation reaction and increases the rate of reaction. Furthermore combined with the high temperature from boiling the sample at 99°C prior to SDS-PAGE separation, which has been shown to increase the rate of deamidation (38), and the vast excess of water in the reaction solution, these conditions yield a high rate constant that would explain the quantitative conversion of the sulfinamide to sulfinic acid (supplemental Fig. 4). Although this conversion was not observed in the kinetics experiments ( Fig.  1) in which the reaction products were analyzed in real time, it was already evident with the test peptides, which were often analyzed after a short storage period. This indicates that the conversion of the sulfinamide to sulfinic acid may occur under less extreme conditions than those experienced by the sample prior to SDS-PAGE. It further complicates quantitative studies of the HNO reactions as it generates additional products that need to be analyzed and also requires taking into account multiple charge states. To identify molecular targets of HNO-induced modification, simplifying the product distribution by minimizing the contribution of one of these species, e.g. by quantitative conversion of sulfinamide into sulfinic acid, therefore seems beneficial. In the current study, quantitative conversion of sulfinamide to sulfinic acid was observed, and this process appears to readily occur using standard sample preparation procedures.
MS analysis of modified human GAPDH, which contains three cysteines, revealed disulfide bond formation between Cys-152 (the active site cysteine) and nearby Cys-156 (Fig.  4a) and a 66-Da neutral loss associated with Cys-247 (Fig.  4b). These results parallel those of a previous in vitro study examining rabbit GAPDH (which contains four cysteines) in which mass spectrometric analysis revealed the formation of an intrasubunit disulfide between the active site cysteine (Cys-149) and the nearby Cys-153 upon treatment with HNO. The remaining two cysteines, Cys-244 and Cys-281, were converted to sulfinamides (29). Our analysis revealed the same modification pattern as the previous study, thereby validating this MS approach as an effective means to identify proteins subject to HNO-induced modification in a physiologically relevant setting. It also lends further credibility to the presumed sulfinamide to sulfinic acid conversion in our studies as the sites of sulfinamide modification in vitro match those of the sulfinic acid modification ex vivo.

Global Proteome Analysis for Discovery of HNO-induced Modifications on Platelet Proteins
Utility of the Neutral Loss for Monitoring the Modification-Applying this approach to a more global analysis of sulfinic acid-modified proteins in AS-treated platelets led to the identification of 21 modified proteins based on 22 modified peptides with 14 of the 22 modified peptides showing the characteristic neutral loss of 66 Da. Proteins identified covered a wide range of functions and abundances, including cytoskeletal and cytoskeleton-associated proteins (e.g. vinculin, gelsolin, and actin), metabolic proteins (e.g. pyruvate kinase isozyme, malate dehydrogenase, and lactate dehydrogenase), signaling proteins (e.g. 14-3-3 /␦, Rab 1a, and Rab 3), and platelet surface receptors including glycoprotein 1b␣ and the ␤3 subunit of integrin ␣IIb␤3. As the detection of 32-Da mass shifts and 66-Da neutral losses were linked to sulfinic acid, other oxidative pathways leading to the same product could not be ruled out.
For more detailed analysis, platelets were treated with either doses of AS ranging from 10 M to 10 mM or a vehicle control. The concentrations of AS were chosen to encompass those shown previously to cause inhibition of platelet function (between 1 and 40 M (12)) as well as those used in other biological systems, such as rat cardiac muscle (50 -1000 M (39) and 100 -1000 M (40)). Treatment with AS was for 2 min, corresponding to the time at which inhibition of platelet aggregation by AS was observed previously (12). It is important to note that these are AS concentrations, whereas the actual concentration of HNO will be considerably lower than that of AS after a 2-min incubation period at physiological pH. These parameters were used to identify those proteins whose modification may be involved in the inhibition of platelet aggregation by HNO.
Further examination of the 21 proteins for dose dependence of the sulfinic acid modification was performed using MRM transitions that led to the identification of 10 proteins that were modified dose-dependently, nine of which had been found previously to produce the characteristic neutral loss of 66 Da (Fig. 4c). Indeed six of the 10 proteins found to contain an HNO-induced modification in this study have previously been reported with redox-based modifications on their cysteine residues often in pathological or stressed situations leading to alterations in protein function (actin (41)(42)(43), ␣-enolase (44), GAPDH (21,45,46), integrin ␤3 (47-49), pyruvate kinase isozyme (50), and vinculin (51)). Redox regulation of cysteine residues has emerged over recent years as a key modulator of protein function, and recently evidence has begun to come to light regarding the processes involved in regulating these redox-based post-translational modifications, showing that many of these are specific and reversible modifications that play a key role in protein function.
This observation that the majority of the modified proteins identified in the MRM transitions were previously found to produce the neutral loss indicates its utility as a characteristic marker of this modification. Furthermore the one remaining protein (integrin ␤3) that did not produce the neutral loss had its modified precursor ion in a highly charged state (5ϩ), which is not conducive to neutral loss formation. This is because there is an inverse correlation between the prevalence of the neutral loss and the charge state of the precursor ion that was observed in the product ion analysis of the sulfinamide modified-test peptides (Fig. 2) as well as in a previous study (52). If the modification has a low proton affinity and the molecular ion is in a low charge state, the probability that the modification fragments as a neutral fragment increases. The generation of the neutral fragment will also depend on the kinetics of the fragmentation mechanism and the thermodynamic stability of the neutral fragment. The finding that the majority of the modified proteins identified by the MRM transitions fragment to produce the neutral loss indicates that monitoring this neutral loss is a more accurate and specific method for discovering HNOinduced modification in proteins than solely monitoring for the modified peptide by mass shift of unmodified to modified peptide, which may generate false positives through incorrect peptide assignments.
Relative Quantitation of the Sulfinic Acid Modification in Platelet Proteins-Absolute quantitation of the degree of sulfinic acid modification was not possible as the compulsory omission of reducing agents will inadvertently result in disulfide bond formation and a decrease of the free cysteinecontaining form of the peptide to an extent that depends on experimental variables such as HNO dose. This results in the unmodified form of the peptides not always being identifiable and excludes the use of its concentration as an internal reference. Relative quantitation was thus used to analyze the dose responses; however, plotting the average fragment peak areas of the sulfinic acid-modified precursor from the MRM transitions does not account for experimental variability from differing amounts of protein in the excised gel bands and differing amounts of peptide produced from the digestion. To correct for the variability from the quantity of protein from gel band to gel band, two non-modified peptides from the same protein were monitored and used to determine a CF. The average fragment peak area (A) was then divided by this correction factor (A/CF). In the case of the variability in the production of the modified and two non-modified peptides from the same protein by tryptic digestion, the assumption that using similar digestion conditions from gel band to gel band would not produce statistically significant differences in quantities of the peptides from experiment to experiment was made. This is a common assumption that underlies all quantitative proteomics experiments, certainly those that rely on label-free quantitation, e.g. spectral counting (53). Lastly to account for instrumental variation from data set to data set, the corrected peak areas (A/CF) of each set were normalized (Fig. 5). Taken together, these normalization steps allowed for the comparison of the dose responses of different proteins. It should be noted, however, that due to this normalization the absolute extent of modification of proteins even within the same group may differ significantly as we could not use the unmodified peptide as an internal reference for quantitation.
Based on the above work flow, 10 proteins were found to be sulfinic acid-modified dose-dependently. Although the 10 modified proteins have all been previously identified in platelet proteome studies (54 -57), it is noteworthy that they span a wide range of abundances in human platelets as determined by their rankings in our global proteome analysis where they appear interspersed among the 400 highest ranking proteins (data not shown). These proteins are involved in a variety of platelet processes, including cytoskeletal and cytoskeletonassociated proteins (actin, adenylyl cyclase-associated protein 1, gelsolin, and vinculin), metabolic enzymes (␣-enolase, fructose-bisphosphate aldolase A, GAPDH, and pyruvate kinase isozyme M1/M2), and platelet surface receptors (integrin ␤3 and platelet glycoprotein 1b␣). Many of the proteins identified play key roles in the regulation of platelet activation, and although the responses of each of the proteins to AS varied, the modification was clearly seen in all proteins at higher doses of AS (1 and 10 mM).
The dose-response curves of the various proteins were grouped together based on similar response trends to gain some insight into the specificity and selectivity of the HNOinduced modification about which little is currently known. Interestingly the most striking dose-response curves were seen in the two platelet surface receptors, integrin ␤3 and glycoprotein 1b␣ (Fig. 5, a and b), which showed increasing sulfinic acid modification with increasing AS dose even at concentrations as low as 0.1 mM in the case of ␤3 and 1 mM in the case of glycoprotein 1b␣. As the HNO donor was added to intact platelets, the extracellular receptors would have encountered HNO relatively early in its concentration gradient and would have been more likely to be subject to modification at low AS/HNO concentrations thus explaining, in part, the differences in response to AS seen from protein to protein.
As both of the platelet surface receptors mentioned above play key roles in platelet activation, their sulfinic acid modification induced by HNO may be an important element of the inhibition of platelet aggregation by AS. Platelet glycoprotein 1b␣ and integrin ␤3 are involved in platelet adhesion and aggregation, respectively. Glycoprotein 1b␣, as part of the glycoprotein 1b-IX-V complex, binds von Willebrand factor as its primary ligand and plays a key role in platelet adhesion and subsequent platelet activation (58). Integrin subunit ␤3 exists in platelets as a heterodimer with subunit ␣IIb, together comprising the platelet-specific integrin ␣IIb␤3. Its primary ligand is fibrinogen, which acts as a bridge between adjacent platelets, supporting aggregation (59). Of particular interest, integrin ␤3 is a highly cysteine-rich protein, which was recently shown to be S-nitrosylated (49). The cysteines found to be modified in both glycoprotein 1b␣ and ␤3 were extracellular, suggesting that the modification may interfere with ligand binding and/or subsequent outside-in signaling functions of these receptors. Indeed treatment of platelets with AS has been shown previously to inhibit PAC-1 binding, a marker of integrin ␣IIb␤3 activation (12), and the results of the present study suggest that this inhibition may in fact result from a direct interaction between HNO and the receptor.
HNO is a small, easily diffusible molecule that can cross biological membranes (60), and thus it is not surprising that several intercellular proteins were also found to be modified in response to AS treatment. These could primarily be divided into platelet cytoskeletal and cytoskeleton-associated proteins and metabolic proteins. Both categories are instrumental in platelet activation; activation of platelet surface receptors such as integrin ␣IIb␤3 leads to transduction of signals to the platelet cytoskeleton, its subsequent polymerization, and reorganization, leading to shape change (61), whereas carbohydrate metabolism and glycolysis are thought to be important energy sources for platelet activation (62). Because of their various roles in platelet activation, interference with the function of these proteins through ASinduced modification could have ramifications for the ability of platelets to aggregate. It is also interesting to note that several glycolytic enzymes were found to be modified as inhibition of glycolysis by HNO has been identified previously as a key mechanism of action of HNO in yeast (21).
The monomeric actin-binding protein adenylyl cyclase-associated protein 1 (63) and the enzyme pyruvate kinase isozyme M1/M2 (and fructose-bisphosphate aldolase A (data not shown)) were found to have similar concentration-response curves upon treatment with AS (Fig. 5c). Both showed slight responses to doses of AS of 1 mM and below (Fig. 5c,  inset) within the range in which platelet inhibition has been shown to occur. However, there was a dramatic effect at an AS concentration of 10 mM. This dramatic increase in sulfinic acid modification may be related to the cytotoxic effects of AS, known to occur with high mM concentrations, and therefore may not be physiologically relevant. In a similar manner, the enzyme GAPDH and focal adhesion protein vinculin (64) (Fig. 5d) were grouped as they were found to have similar dose responses to AS. Both GAPDH and vinculin showed little response to AS at doses of 1 mM and below before climbing steeply at the 10 mM AS dose, possibly also relating to a cytotoxic dose of AS.
The final three proteins found to give similar concentrationresponse curves to AS were actin, gelsolin, and ␣-enolase (Fig. 5e). Gelsolin is an actin-binding protein involved in actin severing and capping (65), whereas ␣-enolase is an enzymatic protein. Sulfinic acid modifications of these proteins were identified; however, although there was a trend of increasing modification with increasing dose of AS, there was no steep increase in the slope of the curve. There was greater variation in the data for these three proteins largely due to low intensity peak areas of the fragment ions. The 10 sulfinic acid-modified proteins showed five different general trends in response to the reaction with AS (Fig. 5, a-e), possibly indicating that HNO targets different proteins with different selectivities and specificities.
In addition to the proposed pathway via HNO-induced sulfinamide modification and subsequent deamidation, nonspecific oxidation of free thiol groups in cysteines may also lead to the formation of sulfinic acid modifications. Indeed it has been shown that the hydrolysis of Angeli's salt results in the generation of singlet HNO, which in addition to free thiols can react with a variety of other chemicals, including HNO and NH 2 OH (66). Under aerobic conditions, HNO has also been shown to react with oxygen to form an as of yet unknown intermediate of 1:1 stoichiometry that is distinct from peroxynitrite (67). Not only has this species been implicated with the oxygen-dependent cytotoxic effects of AS, it has also been suggested to be sufficiently long lived in cellular membranes to initiate oxidation reactions. As a result, HNO may activate two separate chemical reaction pathways: an oxygen-independent mechanism resulting in oxidation of thiols to sulfinamide and disulfides and an oxygen-dependent mechanism that presumably leads to oxidation, hydroxylation, nitrosation, and nitration. The first reaction will rapidly quench HNO in the presence of thiols, however. Depletion of intracel-lular glutathione upon HNO release has indeed been reported, which has led to postulating a membrane-residing HNO-oxygen intermediate as the culprit of the second mechanism as it is presumably protected from this quenching reagent in this location (60).
In the current study, we identified membrane and cytosolic proteins that are subject to sulfinic acid modification of cysteines upon HNO exposure and gel electrophoresis. Furthermore we demonstrated that the initial sulfinamide product is subject to slow conversion into sulfinic acid, which is accelerated under gel electrophoretic sample preparation conditions. This provides a plausible pathway of the formation of sulfinic acid modifications observed in the ex vivo experiments but does not rule out the involvement of the HNOoxygen intermediate. As the reaction was halted only 5 min after addition of AS by platelet lysis, the subsequent release of intracellular thiols would rapidly quench the soluble portion of the HNO-oxygen intermediate but not that residing in the membrane. This implies that direct oxidation of cysteine residues of membrane proteins, but not cytosolic proteins, may also occur in these experiments. The possibility of two pathways being active in parallel with the contribution of each being dependent upon the location of the protein as well as the local concentration of the protein, HNO, oxygen, and quenching reagents such as free thiols, HNO, and NH 2 OH is intriguing. Not only would this explain the complex modification patterns and dose-response curves observed in this study, it would mandate more detailed mechanistic studies to shed more light on this reaction. Our identification of cellular targets of HNO-induced modification will, for the first time, allow such targeted studies to be performed. Conversely a pathway independent of HNO leading to direct sulfinic acid formation can be ruled out. The vehicle control experiment, performed at a zero dose of AS, showed relative intensities that correspond to base-line noise, i.e. the absence of sulfinic acid modification in the absence of HNO.
Given the current excitement regarding the potential of HNO as a therapeutic agent, for example for the treatment of heart failure (10,11,16), and its recent success in inhibiting angiogenesis and breast cancer growth in vitro (15), the approach outlined in this study should be of use to researchers in the biomedical community interested in further elucidating the targets and mechanism of action of nitroxylation. This mass spectrometry-based approach, applicable to a single protein or at a proteome-wide level, has for the first time allowed the identification of HNO-induced modifications on proteins in a biologically relevant setting.