Abstract
Aims:
Cysteine (Cys) is a major target for redox post-translational modifications (PTMs) that occur in response to changes in the cellular redox environment. We describe multiplexed, peptide-based enrichment and quantitative mass spectrometry (MS) applied to globally profile reversible redox Cys PTM in rat hearts during ischemia/reperfusion (I/R) in the presence or absence of an aminothiol antioxidant, N-2-mercaptopropionylglycine (MPG). Parallel fractionation also allowed identification of irreversibly oxidized Cys peptides (Cys-SO2H/SO3H).
Results:
We identified 4505 reversibly oxidized Cys peptides of which 1372 were significantly regulated by ischemia and/or I/R. An additional 219 peptides (247 sites) contained Cys-SO2H/Cys-SO3H modifications, and these were predominantly identified from hearts subjected to I/R (n = 168 peptides). Parallel reaction monitoring MS (PRM-MS) enabled relative quantitation of 34 irreversibly oxidized Cys peptides. MPG attenuated a large cluster of I/R-associated reversibly oxidized Cys peptides and irreversible Cys oxidation to less than nonischemic controls (n = 24 and 34 peptides, respectively). PRM-MS showed that Cys sites oxidized during ischemia and/or I/R and “protected” by MPG were largely mitochondrial, and were associated with antioxidant functions (peroxiredoxins 5 and 6) and metabolic processes, including glycolysis. Metabolomics revealed I/R induced changes in glycolytic intermediates that were reversed in the presence of MPG, which were consistent with irreversible PTM of triose phosphate isomerase and glyceraldehyde-3-phosphate dehydrogenase (GAPDH), altered GAPDH enzyme activity, and reduced I/R glycolytic payoff as evidenced by adenosine triphosphate and NADH levels.
Innovation:
Novel enrichment and PRM-MS approaches developed here enabled large-scale relative quantitation of Cys redox sites modified by reversible and irreversible PTM during I/R and antioxidant remediation.
Conclusions:
Cys sites identified here are targets of reactive oxygen species that can contribute to protein dysfunction and the pathogenesis of I/R.
Introduction
Post-
Oxidative Cys PTMs that can be reduced are considered “reversible” and include S-nitrosylation (SNO), S-glutathionylation (SSG), S-acylation, S-cysteinylation, S-sulfhydration, and S-sulfenylation (Cys-SOH), as well as intra- and intermolecular disulfides (8, 25, 26, 41, 45, 46, 62); while those with no known means of reduction (Cys-sulfinic and -sulfonic acid; Cys-SO2H/SO3H) are regarded as “irreversible,” are associated with oxidative “damage,” and are considered markers of protein dysfunction and degradation (41, 57). A single exception exists; enzymatic reduction of sulfinylated peroxiredoxins (Prx I–IV) to their active form by sulfiredoxin (4, 86).
Blockage of coronary flow leading to myocardial ischemia can result in temporary contractile dysfunction or irreparable cardiomyocyte necrosis depending on the location and severity of the occlusion and the duration of hypoxia. Ischemia is characterized by a dramatic reduction in high-energy phosphates, including creatine phosphate and adenosine triphosphate (ATP), and a switch from aerobic to anaerobic glycolysis resulting in increased intracellular lactic acid and reduced pH (33). While reperfusion is critical to survival posthypoxia, it comes with the burden of an influx of ROS generated by uncoupling of mitochondrial electron transport, elevated succinate, reduced antioxidant activity, and immune cell infiltration (2, 11, 14, 60). ROS are assumed to generate contractile and metabolic dysfunction via protein oxidation; however, the precise targets and extent of redox modification remain poorly understood.
Innovation
Redox-sensitive cysteines (Cys) are important in the response to reactive oxygen species during pathogenesis. A large-scale and multiplexed enrichment technique identified and quantified >4500 redox-modified Cys peptides in myocardial tissue during ischemia and ischemia/reperfusion (I/R). Cys oxidations were prevalent during ischemia even without subsequent reperfusion, and many functionally significant sites were “protected” from both reversible and irreversible oxidation by addition of an aminothiol antioxidant during reperfusion. Irreversible post-translational modifications (PTMs) were quantified using a novel parallel reaction monitoring mass spectrometry approach that showed I/R leads to large increases in overoxidation, which in most cases was attenuated by antioxidant intervention consistent with functional recovery. Our data show that Cys redox PTMs, particularly those involving loss of function, are a major contributor to the phenotypes of ischemia and I/R.
PTMs including phosphorylation (18, 38, 84), glycosylation (31, 42, 54, 64), acetylation (52, 55, 80), and protein cleavage (21, 44, 83) are associated with the pathophysiology of ischemia/reperfusion (I/R), and many such modifications are influenced by ROS; antioxidant interventions can reverse I/R-associated PTMs (85). Due to the crucial role of ROS in pathogenesis, substantial focus has been devoted to identifying redox PTMs, including SNO, SSG, and Cys-SOH, during I/R and cardioprotection (10, 19, 32, 34, 35, 48, 49). A number of methods exist for the delineation of Cys redox sites, but such approaches are often specific to only a single PTM [reviewed in Boronat et al. (6) and Yang et al. (87)].
Cys residues have low frequency (∼1.3% of all protein residues), and their PTMs likely constitute only a small fraction of site occupancy under physiological conditions. Enrichment is therefore required using various techniques, including (i) biotin switch, where specific PTM can be targeted by employing different reducing agents (e.g., ascorbate for SNO) (30); (ii) a variation of the biotin switch that employs chemical labeling, with the most frequently used labels containing a carbon electrophile (to alkylate Cys) and an enrichment handle {e.g., isotope-coded affinity tags [ICAT and oxICAT (22, 37, 40, 69, 78)] and Cys-reactive tandem mass tags [CysTMT and iodoTMT (50, 53, 56, 63)]}; (iii) resin-assisted capture by thiol-disulfide exchange (TDE) (20, 35, 58, 73); and (iv) incorporation of chemoselective tags (28, 88, 89).
We developed a peptide-based, quantitative, and multiplexed enrichment strategy incorporating TDE and isobaric labeling to monitor changes in reversible Cys redox PTM. Parallel analysis of irreversible Cys redox modifications [Cys-SO2H/SO3H (57)] enabled determination of site redox status in myocardial ischemia, I/R and I/R in the presence of a synthetic aminothiol-containing antioxidant (N-2-mercaptopropionylglycine or MPG), which attenuates contractile dysfunction after I/R in animal models (51, 85). Relative quantitation of irreversible Cys PTM was performed using parallel reaction monitoring mass spectrometry (PRM-MS) and changes associated with specific enzyme function. This study provides a global view of Cys redox PTMs that likely collectively contribute to the biochemical and pathophysiological phenotype of I/R.
Results
Contractile dysfunction and thiol status during I/R and antioxidant treatment
Recovery (I/R hearts) or maintenance of function (nonischemic time controls or NITCs) for Langendorff perfused rat hearts (n = 6, NITC; n = 9, all other groups) was measured as a percentage rate pressure product (RPP) at baseline (Fig. 1a). NITC hearts maintained function for the duration of the protocol (98.9 ± 3.6%), whereas hearts subjected to 15 min ischemia and 15 min reperfusion (15I/15R) recovered to only 70.1 ± 2.5% of baseline RPP. RPP improved significantly to 85.1 ± 3.4% when 1 mM MPG was included in reperfusion (15I/15R+MPG), which is consistent with previous studies (55, 85). Total free thiol assay (Fig. 1b) revealed that hearts subjected to 15 min ischemia without reperfusion (15I) showed no change in comparison with NITC; however, 15I/15R hearts showed a significant decrease in free thiols, which were recovered by addition of MPG. Similar data were obtained for the protein-free thiol assay (Fig. 1c).
Western blotting using anti-SNO (anti-iodoTMT) antibodies confirmed increased SNO in 15I/15R (Fig. 1d; 3.6 ± 0.7-fold increase compared with NITC), with no change observed in 15I. SNO was attenuated after addition of MPG (2.2 ± 0.3-fold compared with NITC). Similar experiments probing proteins with anti-Cys-dimedone (anti-SOH) antibody revealed increased reactivity in both 15I and 15I/15R (both 2.2 ± 0.2-fold) compared with NITC (Fig. 1e). This overall densitometric quantitation was skewed somewhat by the appearance of a 15I-specific band at ∼35 kDa with shared bands showing largely greater intensity in 15I/15R. MPG diminished SOH reactivity to near NITC levels (1.4 ± 0.2-fold). These data show that while redox PTMs are predominantly associated with I/R, specific protein modifications also occur in ischemia alone.
We excised the sulfenylated 15I-specific 35 kDa band and performed a trypsin digest followed by liquid chromatography–tandem MS (LC-MS/MS), which identified three proteins: glyceraldehyde-3-phosphate dehydrogenase (GAPDH), isocitrate dehydrogenase, and NADH dehydrogenase [ubiquinone] 1 alpha subunit 10 (Ndufa10) (Supplementary Data S1), suggesting that the protein is one of these candidates. Since the thiol assays showed a loss of free thiols in 15I/15R, we next performed targeted metabolomics by LC-MS/MS to examine specific thiol-containing small molecules. We observed small, but not statistically significant, decreases in reduced glutathione (GSH) in 15I and 15I/15R that were reversed by MPG (Fig. 1f). We observed no significant change in free L-Cys (or the overoxidized free L-Cys-SO2H and L-Cys-SO3H forms) (Supplementary Fig. S1a–c) in 15I or 15I/15R. These results suggest that protein-associated thiol redox PTMs account for the majority of the change observed in total free thiols.
Multiplexed TDE method for enrichment and quantitation of Cys redox PTM
Since I/R results in a loss of protein thiols and globally increased reversible redox PTM, we next attempted to identify the Cys sites responsible for this effect. Proteins from the four tissue groups were trypsin digested, labeled with isobaric tags for relative and absolute quantitation (iTRAQ), and combined for TDE enrichment. Significant changes were made to our previous approach (58) to prevent thioacetylation by the iTRAQ tags. The complete series of derivatization steps included the following: (i) low pH alkylation of free thiols with N-ethylmaleimide (NEM); (ii) thiol-based reduction of all reversible Cys PTM with dithiothreitol (DTT); (iii) protection of newly free thiols as disulfides with methyl methanethiosulfonate (MMTS); and (iv) in-solution trypsin digestion (Fig. 2). Reduction of protecting disulfides with tributylphosphine (TBP) allowed for Cys peptide capture by TDE.
After enrichment, two fractions were obtained; peptides eluted from TDE (“bound”), which are those formerly modified by reversible Cys PTM, and those “unbound,” which include non-Cys peptides and NEM-alkylated Cys peptides that were used for normalizing Cys PTM quantitation and determining protein abundance.
Reversibly oxidized Cys peptides in ischemia, I/R, and antioxidant treatment
In the TDE bound fraction, we identified 4505 unique Cys peptides (Supplementary Data S2), of which 3147 were measured in ∼3/4 biological replicates. The interexperimental variance was compared with the variance observed across MS runs for all ratios between iTRAQ channels. Within MS replicates, the average coefficient of variance (CV) was 18.5%, while average CV across any two biological replicates was 21.8%. When all replicates were summed together, the interexperimental CV was 14.1%. Quantitative MS revealed 1372 regulated Cys peptides with a Loge(Prob) < −2 (i.e., <1% probability that the iTRAQ reporter distributions are identical) in comparison with NITC (Supplementary Data S3).
A total of 970 Cys peptides were altered in 15I compared with NITC, 737 in 15I/15R, and 455 in 15I/15R+MPG. Overall, 71% of regulated Cys peptides (n = 968) were increased in abundance (i.e., “oxidized”) and 21% (n = 284) were decreased in abundance (i.e., “reduced”) with respect to NITC (Supplementary Fig. S3), with the remainder varying across conditions. 15I and 15I/15R samples followed this trend with the majority of sites increased in abundance (76% and 78% oxidized, respectively). In 15I/15R+MPG however, 61% of altered sites were decreased in abundance (relative to NITC), indicating the “protective” nature of MPG against protein oxidation. The TDE unbound fraction contained no proteins that were considered statistically significantly changing using a ±2-fold change with p < 0.05 when comparing 15I, 15I/15R, and 15I/15R+MPG proteins with NITC (Supplementary Fig. S4).
We next subjected the data to cluster analysis with GProX (66). One thousand one hundred two of the 1372 (80.3%) significantly altered reversibly oxidized Cys peptides grouped into four major clusters (Fig. 3; Supplementary Data S4). All four clusters contained Cys peptides that demonstrated increased abundance in 15I and/or 15I/15R compared with NITC. Clusters 1 and 2 contained the largest membership (n = 422 and 395 Cys peptides, respectively). Cluster 1 contained peptides that showed elevated abundance in 15I and 15I/15R compared with NITC, and a reduction with the addition of MPG; while cluster 2 contained peptides with further increased abundance between 15I and 15I/15R (with both increases attenuated by MPG). Cluster 3 contained the smallest membership (n = 131 peptides) and consisted of Cys peptides increased in abundance in 15I alone, while cluster 4 contained reversibly oxidized Cys peptides that were elevated in abundance in both 15I and 15I/15R relative to NITC, but they were not influenced by the addition of MPG.
Each cluster was subjected to functional analysis using STRINGdb (Supplementary Fig. S5; Supplementary Data S5 and S6) (74). All clusters were significantly enriched for known interactions (STRING) and Kyoto Encyclopaedia of Genes and Genomes (KEGG) functional pathways, relative to the entire rat genome. Cluster 1 was enriched in mitochondrial proteins, as well as metabolic pathways, indicating that these are the early targets of oxidants during ischemia; cluster 2 was enriched in extracellular and membrane-bound proteins, and these were functionally associated with KEGG pathways containing contractile and extracellular matrix (ECM) proteins. Cluster 3 contained extracellular proteins mainly associated with blood plasma rather than ECM and finally, cluster 4 contained predominantly mitochondrial proteins.
Reversibly oxidized Cys peptides were next used to identify over-represented sequence motifs using MotifX (13). Motif analysis compared the “changed” sites versus those that were not significantly altered (“unchanged”) against the complete dataset; and second, within each specific heart group. Overall, 34 motifs were observed (Supplementary Fig. S6; Supplementary Table S1). Motifs containing vicinal thiols (CXC, CX2C, and CX5C) were enriched in the “changed,” but not “unchanged,” set indicating that such Cys residues are more likely to be oxidized during ischemia and I/R.
The “unchanged” Cys peptide set only contained over-represented motifs that included a single Cys in combination with Asp/Glu and/or Gly. We next examined motifs that were enriched in each of the heart groups. The most enriched motif, present almost exclusively as oxidized in 15I/15R, was CXDX2EC (>30-fold over-representation). Cys peptides reversibly oxidized in 15I were enriched in a single vicinal thiol motif (CX2CXC, ∼8.7-fold), while motifs associated with 15I and 15I/15R included GX2CX2C (7.8-fold and 10.3-fold over-represented in 15I and 15I/15R, respectively) and CXC (2.3- and 2.8-fold). Additional 15I/15R-specific motifs included KX5C (2.1-fold enriched) and CX2CX2G (7.6-fold). 15I/15R+MPG generated only 1 over-represented motif (CK, 3.4-fold enriched) that was significant only in the oxidized set, thus indicating that these sites were not protected from ROS by MPG.
Reversible redox PTMs alter TDE enrichment of ischemia and I/R proteins
While reversible redox PTMs at the site-specific level likely confer structural and/or functional alterations to their parent proteins, we also wished to determine whether these changes resulted in quantifiable differences at the protein level in hearts subjected to 15I and 15I/15R, and whether MPG could ameliorate such effects. Mitochondrial proteins were strongly linked with clusters 1, 2, and 4 that were associated with increased reversible Cys PTM in 15I and 15I/15R (Fig. 3; Supplementary Data S5). We therefore purified mitochondria and performed protein-based TDE to enrich reversibly redox-modified mitochondrial proteins from each of the 4 heart groups (TDE bound; Fig. 4a). No differentially abundant proteins were observed in the subset that were not reversibly redox modified (TDE unbound). In the TDE-bound set, we observed 11 protein bands that were reproducibly differentially abundant at mean > 1.5-fold levels as determined by densitometry in 15I and/or 15I/15R samples (compared with NITC) across n = 4 experiments (Fig. 4a). Many bands observed on the gels were also less intense in MPG-treated samples, despite no differences in the TDE-unbound or total mitochondrial lysates between heart groups (Supplementary Fig. S7), which was also consistent with our total protein quantitation (Supplementary Fig. S4).
The 11 bands were excised, digested with trypsin, and subjected to LC-MS/MS for identification (Fig. 4b). Two bands (7 and 11) contained 2 highly scoring matches, such that 13 proteins in total were identified from the 11 gel bands. Twelve of the 13 altered TDE-bound proteins were represented by 30 reversible Cys redox-modified peptides (Supplementary Data S4), 24 of which (from 10 proteins) were altered in abundance and contained within clusters 1, 2, and 4, with another 5 found in the changing dataset but not clustered (cluster 0; all 5 elevated in abundance in 15I and/or 15I/15R). Consistent with our motif data, 10 of these 30 peptides were associated with vicinal Cys thiols. We also noted that band 4 (5.63-fold more abundant in 15I than NITC and not elevated in 15I/15R) shared similar mass and an identity (GAPDH) with one of the three proteins identified in the ∼35 kDa band that was heavily SOH modified in 15I alone (Fig. 1e; Supplementary Data S1). This strongly suggests that GAPDH is the ∼35 kDa 15I-specific S-sulfenylated protein. Collectively, these data confirm that reversible Cys redox PTM can occur at levels that influence the total pool of specific proteins, and that both ischemia and I/R can induce these changes.
Irreversibly oxidized Cys peptides in ischemia, I/R, and antioxidant intervention
A decrease in TDE capture of a Cys peptide may reflect (i) relative reduction of a Cys site, leading to increased free thiol or; (ii) overoxidation to Cys-SO2H/SO3H, which do not undergo TDE. While irreversible Cys PTMs exist at low physiological levels, they likely increase during oxidative stress in conditions such as I/R. Irreversible Cys redox PTM, however, cannot be enriched after isobaric labeling as the reagents change peptide charge distribution and alter fractionation by strong cation exchange (SCX) and hydrophilic interaction liquid chromatography (HILIC) (57). We therefore enriched for irreversible Cys PTM from each of NITC, 15I, 15I/15R, and 15I/15R+MPG samples individually, and in parallel with the reversible redox PTM experiments described above (Supplementary Fig. S8).
We identified 219 irreversibly oxidized Cys peptides (containing 247 Cys sites) from 150 proteins (Supplementary Table S2). These were comprised of 141 peptides identified in the SO3H form alone, 51 peptides identified in both SO2H- and SO3H-modified forms, and 27 identified with an SO2H modification alone. Across the heart groups, we observed 34 irreversibly modified Cys peptides in NITC, 128 after 15I and 168 in 15I/15R, providing further evidence that protein oxidation can occur in ischemia even before reperfusion. Addition of MPG to 15I/15R hearts lessened irreversible Cys PTM to just 24 peptides, highlighting MPG “protected” sites as candidates for explaining the functional deficiencies seen in I/R.
Analysis of the parent proteins from which irreversible Cys peptides were identified revealed that mitochondrial proteins (n = 70; 46.7%; Supplementary Table S2) were the most common targets of ROS during ischemia and I/R, followed by cytoplasmic proteins (20.7%), contractile apparatus (10.7%), membrane (8.0%), and extracellular and secreted proteins (6.7%). STRINGdb Gene Ontology (GO) further emphasized these results (Supplementary Data S7 and S8); analysis of the irreversibly oxidized Cys dataset from 15I and 15I/15R showed that 6 of the top 7 enriched GO subcellular locations were associated with mitochondria.
STRING analysis revealed that the complete irreversible dataset was predominantly defined by two clusters (Supplementary Fig. S9); one containing proteins associated with metabolic processes (as also highlighted by KEGG; Supplementary Data S8) and a second, smaller cluster associated with contractile proteins. Removal of proteins containing irreversibly modified Cys peptides remaining in the 15I/15R+MPG set (and those from NITC) resulted in a single cluster associated with mitochondria (Supplementary Fig. S9). Further confirmation resulted from analysis of the MPG set, which showed contractile proteins contained Cys sites that remained irreversibly oxidized even in the presence of MPG.
We next determined the influence of irreversible Cys redox PTM on their reversibly modified forms by aligning the data from each experimental set. Of the 219 irreversibly oxidized Cys peptides, 149 (68.0%) were also identified in the reversibly oxidized Cys peptide set (Supplementary Table S3); of these 115 (77.2%; 123 Cys sites) were significantly altered in abundance (and thus could be clustered (Fig. 3); these are here referred to as “group 1”), while 34 (22.8%; 38 Cys sites) were deemed not to be altered by ischemia, I/R, or I/R+MPG (“group 2”). Sixty-seven peptides (30.6% of the overall irreversible Cys peptide set; 86 sites [“group 3”]) were not identified in the reversible experiments, and 3 peptides corresponded to tryptic missed cleavage of sites identified in other peptides and were not considered further.
We undertook a site-specific analysis that consisted of database (UniProt) and literature searches to identify “catalytic” (those associated with protein function) or “structural” (those in known or predicted disulfides) sites; as well as those previously identified as modified by SNO (16, 19, 34, 35, 49, 50, 74) (Supplementary Table S4). Eighty-seven of 247 Cys sites have either experimentally proven (51 sites) or predicted (36 sites) roles in protein structure/function, with 68 sites associated with catalytic activity and 30 sites with disulfide bonds (11 sites have both potential roles; Supplementary Tables S3 and S4).
Groups 1 and 2 were enriched with Cys sites known to be targets of SNO (61.8% and 60.5% of sites, respectively), while group 3 contained significantly fewer SNO sites (14.0%; Supplementary Table S3). Sites associated with catalytic function across all groups were also enriched with known SNO (54 of 68 sites; 79.4%) particularly when compared with structural sites (11 of 30 sites SNO; 36.6%). Catalytic sites were not evenly distributed across groups; groups 1 and 2 contained 34.6% and 37.5% catalytic sites, respectively, in comparison with only 9.1% in group 3. Structural sites were less typically associated with groups 1 and 2 (9.2% and 12.5%, respectively) than with group 3 (14.8%), suggesting that disulfides protect against further redox PTM. The number of sites for which no role has been ascertained was high across all groups.
Relative quantitation of irreversibly modified Cys peptides by PRM-MS
We next attempted to provide relative quantitation for irreversibly modified Cys peptides by employing PRM-MS. Enrichment efficiency was normalized using dimethylated and “spiked in” Cys-SO3H peptides from bovine serum albumin (BSA) and total chromatographic peak areas summed from all irreversibly modified Cys peptides. We first employed a targeted strategy to examine peptides from peroxiredoxins 5 and 6 (Prx5 and Prx6) using a scheduled approach based on known LC retention times, precursor (peptide) mass and MS/MS fragment ions (transitions; Supplementary Table S5; Supplementary Fig. S10).
Prx5 is an atypical 2-Cys Prx with sites at Cys99 and Cys203. Cys203 provided unreliable data as it is contained within the C-terminal, nontryptic peptide (not shown). Cys99 was assayed by the peptide 85GVLFGVPGAFTPGC[SO3H]SK101, which showed a 1.8-fold increase in 15I/15R, with the addition of MPG lowering abundance to NITC levels (Fig. 5a). For Prx6, a 1-Cys Prx, we monitored the single redox site (Cys47; Fig. 5b) in the peptide 42DFTPVC[SO3H]TTELGR53, which showed a 1.4-fold increase in 15I/15R that was reduced to 1.2-fold with the addition of MPG.
We next applied PRM-MS to quantify 34 irreversibly modified Cys peptides, representing 20 proteins and corresponding to 27 sites of modification (some sites identified in both SO2H and SO3H forms; Supplementary Table S5; Supplementary Data S9). Only four peptides showed no significant change in abundance in any group compared with NITC, and just 6 of 34 peptides were not increased by >1.5-fold in 15I/15R (Fig. 6; Supplementary Fig. S11). Twenty-four peptides demonstrated attenuation of their 15I/15R levels by addition of MPG, with many being less abundant in the presence of antioxidant than in NITC, which is consistent with the lower numbers of irreversibly modified Cys peptides in the overall MPG dataset (Supplementary Table S2).
Ten peptides showed a >1.5-fold increase in abundance in 15I, as well as 15I/15R, again indicating that protein overoxidation can occur before the ROS burst typically associated with reperfusion. These PRM-MS data were also comparable with those achieved above for Prx5 Cys99-SO3H (2.0-fold increase in 15I/15R completely ameliorated by MPG; Fig. 5c) and Prx6 Cys47-SO3H (+1.8-fold in 15I/15R and +1.6-fold in 15I/15R+MPG, compared with NITC; Fig. 5d). We were also able to monitor the Cys-SO2H form of Prx6 Cys47 (Fig. 5e), which showed a large (3.0-fold) increase in 15I/15R but no change in 15I or 15I/15R+MPG, relative to NITC.
PRM-MS quantified a further eight peptides in their Cys-SO2H forms (Fig. 6; Supplementary Fig. S11). These were mainly present at reduced abundance in 15I relative to NITC (6 of 9 SO2H-modified peptides) and significantly higher abundance in 15I/15R (all SO2H-modified peptides). All Cys-SO2H abundance increases were ameliorated by MPG. The 15I-associated decrease in SO2H-modified peptides most likely reflected rapid conversion to SO3H; however of 6 peptides identified in both SO2H- and SO3H-modified forms, only two [containing electron transfer flavoprotein alpha (ETFA) Cys53 and fructose bisphosphate aldolase (FBA) Cys339] showed a rise in their Cys-SO3H forms at 15I consistent with this hypothesis (Fig. 6a; Supplementary Fig. S11a). For the remainder, we speculate that alternative PTM at another site may be responsible for the apparent change in redox PTM abundance.
Changes in irreversibly modified Cys peptides also appeared to be independent of the parent protein. We identified two or more peptides with different Cys sites from three proteins [triose phosphate isomerase (TPI; 4 peptides), ETFA, and long-chain acyl-CoA dehydrogenase (LCAD)] (Fig. 6a–c). While the SO2H-modified ETFA peptides containing Cys53 and Cys155 were consistent with each other, peptides from TPI containing Cys21,27, Cys42, Cys67, and Cys218 showed very different abundance changes across the heart groups (Fig. 6b), with the Cys67 peptide showing no change in any condition, while the doubly SO3H-modified Cys21,27 peptide was increased in 15I and 15I/15R, but not attenuated by MPG. Similarly, the LCAD Cys342 SO3H peptide, which was elevated in 15I/15R, did not show any effect of MPG addition (Fig. 6c), while the Cys351 SO3H peptide was attenuated by antioxidant.
Finally, we quantified three peptides that contain two modified Cys residues [TPI Cys21,27, dihydrolipoyl dehydrogenase (DLD) Cys80,85, and GAPDH Cys150,154]. For DLD, single SO3H modification of either site was elevated with respect to NITC across all conditions, irrespective of treatment (Fig. 6d), although the magnitude of change was less in 15I/15R, which suggests conversion to a doubly modified form. Although we did not observe this variant for DLD, we were able to monitor all three SO3H-modified forms of GAPDH Cys150,154, which responded similarly to 15I, 15I/15R, and were significantly decreased by MPG (Fig. 7a).
Association of irreversible Cys redox PTM and glycolysis metabolites
To determine whether irreversible Cys redox PTMs quantified on FBA, TPI, and GAPDH result in altered enzyme functions, we analyzed the relative abundance of glycolytic intermediates during I/R with or without MPG, using targeted metabolomics. First, we noted that both ATP and NADH levels were significantly compromised after 15I/15R (Fig. 7b), consistent with reduced glycolytic output and associated with pyruvate depletion and lactate accumulation (Supplementary Fig. S12). MPG restored NITC levels of these compounds, suggesting that oxidants play an important role in this effect.
Abundances of glycolytic intermediates for which targeted MS assays are available revealed significantly increased levels of glyceraldehyde-3-phosphate (G3P) in 15I/15R that were returned to NITC levels by MPG (Fig. 7b), consistent with reduced function of GAPDH that contains irreversible redox PTM displaying abundance changes consistent with this functional effect. Levels of fructose 1,6-bisphosphate were unchanged (Supplementary Fig. S12), suggesting that redox modifications of FBA Cys339 do not influence glycolytic flux at this step. MPG did not significantly alter 15I/15R levels of other glycolytic intermediates for which enzymes were not identified as redox modified (Supplementary Fig. S12), but did have a small, yet significant effect on 15I/15R levels of phosphoenolpyruvate, consistent with the observed but not quantified irreversible PTM of pyruvate kinase (PYK). Finally, we assayed GAPDH enzyme function in NITC, 15I/15R, and 15I/15R+MPG hearts, which showed that I/R resulted in a significant loss of GAPDH activity that was restored to NITC levels by addition of MPG, consistent with irreversible redox modifications resulting in loss of function (Fig. 7c).
Discussion
Large-scale quantitation of Cys redox PTM was performed during ischemia, I/R, and I/R in the presence of the antioxidant, MPG. Our method (57, 58) was optimized to include the following: (i) extraction conditions to “block” free thiols and limit artifactual oxidation, (ii) thiol-based reduction to ensure complete conversion of all reversible Cys PTMs to thiols, (iii) thiol protection by a methyl disulfide group, (iv) in-solution labeling, (v) TDE enrichment, and (vi) parallel identification of irreversible Cys PTM.
The method can occur in “one pot” as each reactant is terminated by an excess of the following reagent to limit sample loss. While NEM blocking in the first step is assumed to be 100% efficient (67), we cannot unequivocally rule out minor enrichment of nonredox-modified Cys sites if a very small proportion of that site pool is not blocked by NEM; however, since we observed no changes in total protein abundance, and assuming the contribution of any inefficiency would be equal across groups, any such sites will not generate quantitative differences in redox-modified peptides.
Irreversible Cys PTMs were quantified by a novel PRM-MS approach. Changes in irreversible PTM of proteins involved in the initial “payoff” phase of glycolysis (e.g., GAPDH) during I/R were ameliorated by MPG, consistent with metabolomics analysis of the corresponding glycolytic intermediates, as well as NADH and ATP, confirming that such PTMs play a pivotal role in injury-associated oxidative damage.
Reversible and irreversible Cys redox PTMs were observed in ischemia even before reperfusion. Western blotting for Cys-SOH also showed the presence of ischemia-specific targets. The production of ROS during ischemia has been shown (3, 79), and hypoxic stress is also activated by ROS (9). Our study shows the wide extent of oxidative Cys PTM in ischemia, and that modification of many sites appears specific to the ischemic period, with 15I resulting in both the largest number of profiled reversible changes and a significant number of irreversibly oxidized Cys sites, some of which were present at significantly higher abundance in that condition. I/R, however, did result in the largest total of irreversibly oxidized Cys sites, and ∼80% of quantified irreversible sites were increased >1.5-fold in I/R.
I/R also resulted in a loss of total and protein-free thiols, which were largely associated with protein rather than small molecules (e.g., GSH). Combined together, our data suggest a model in which ischemia results in elevated reversible PTM within the total Cys site “pool” of a given protein, with reperfusion then resulting in further oxidation and a greater proportion of the original site pool becoming reversibly oxidized. At the same time, the ischemia-associated reversibly oxidized pool (or indeed in models where modifications such as SNO are protective against irreversible modification, the nonmodified pool) is vulnerable to reperfusion-associated “over” oxidation (to Cys-SO2H/SO3H).
The time point profiled here (15R post-15I) is later than the ROS peak, which is thought to occur <10 min after onset of reflow (5). Therefore, the redox state at 15R could reflect some recovery, particularly for small molecule thiols and reversible Cys PTM. Relative depletion of protein Cys has been observed earlier (e.g., 5 min) in reperfusion compared with later (30 min) (37). As 15R was the time point showing a statistical difference in functional recovery between I/R and I/R+MPG it was chosen for this study, although earlier time points may also be relevant for the elucidation of redox signaling upon reperfusion.
Motif analysis of modified Cys sites during ischemia and I/R revealed a bias toward vicinal thiol motifs, including the highly reactive CX2C motif present in many TDE-active proteins [e.g., thioredoxin (12)]. This motif is not enriched in the unchanged or I/R+MPG sets, which indicates that it is specifically targeted by oxidants during 15I and I/R. The enrichment of motifs in the “changed” peptide set alone (e.g., GX2CX2CX2G) likely indicates that they are common to both 15I and I/R, but not frequent enough to be revealed in either set alone. Motifs not containing vicinal thiols may indicate sites of specific PTM such as SNO. Indeed, one motif, CX5K, has been suggested as a SNO motif (35). Overall, the motifs reflect that Cys reactivity is tuned by electrostatic factors, structural constraints, and the presence of proximal thiols.
Cys-SO2H/SO3H are extremely low abundance PTM under physiological conditions and represent a major analytical challenge for enrichment (57). Thus, the number of sites identified here is few when compared with large-scale analyses of more abundant PTM (e.g., phosphorylation). The major drawback of Cys-SO2H/SO3H enrichment is an incompatibility with labeling techniques (e.g., iTRAQ), which alter charge distribution of modified peptides and reduce enrichment efficiency [to ∼2% from 10% to 20% (57)]. This can be overcome by stable isotope incorporation; however, such approaches remain largely incompatible for most tissue-based studies.
We developed a PRM-MS strategy for quantitation of irreversible Cys redox PTM. Prx5 and Prx6 were used to establish this approach as they are highly redox active and their response to oxidative stress is crucial in I/R injury. PRM-MS showed that Cys sites in both proteins were overoxidized in I/R, and these were reversed to varying degrees by MPG. Further validation of the Prx data was not possible; while several antibodies are available for SO2H/SO3H modification of Prxs I–IV, such reagents are rare for Prx5/6, and a single commercially available Prx6-SO3H polyclonal antibody demonstrated significant nonspecific binding (data not shown).
Constraints that are known in targeted proteomics (e.g., presence of methionine, missed tryptic cleavage) reduced the list of peptides considered reliable for PRM-MS, although 34 quantitative assays were ultimately generated, and additional replicates for Prx5 and Prx6 generated data near identical to that achieved during method establishment. The “holy grail” remains to determine occupancy of free thiol and each redox PTM at Cys sites; however, absolute quantification using peptide standards for most such PTMs is not currently possible (e.g., due to rapid reactivity of SOH to SO2H and SO3H), and thus full redox site stoichiometry remains elusive.
Functional and subcellular location analyses of proteins containing Cys sites most affected by redox PTM during 15I and 15I/15R were reflective of processes known to show dysfunction after I/R, indicating that such deficiencies may be related to Cys PTM. Mitochondrial proteins were those most significantly targeted by ROS. Antioxidants limit the burden of I/R in animal models and although they have shown limited clinical efficacy, this may be due to the lack of suitable delivery to mitochondria (36). Energy generation is affected by ischemia and I/R and Cys PTM were identified in enzymes involved in high-energy phosphate production {e.g., glycolysis, fatty acid β-oxidation (FAβox), tricarboxylic acid (TCA) cycle, and creatine kinase [oxidation of Cys317 inactivates the enzyme (65, 81)]}. Several such proteins have been observed as redox “sensitive” in previous studies (7, 17, 32, 35, 37, 49).
Site-specific analysis showed that >50 experimentally verified Cys sites involved in catalysis were a target of irreversible PTM in ischemia and/or I/R. We identified 17 sites (14 peptides) of irreversible Cys PTMs, of which 13 (11 peptides) were also identified as reversibly oxidized, representing all 7 proteins from the FAβox pathway (Supplementary Fig. S13). PRM-MS showed that SO3H modification of the active site (Cys92) of 3-ketoacyl-CoA thiolase was induced in 15I and 15I/15R (>2-fold in both), and this increase was not observed in MPG hearts. SNO of very long-chain acyl-CoA dehydrogenase (VLCAD) Cys237 increases activity (16), and our data show an ischemia- and I/R-related rise in reversible Cys PTM at this site, as well as SO2H/SO3H PTM that is reversed by MPG. Protein-level TDE showed that VLCAD was significantly more enriched in 15I, suggesting “protective” SNO at this site to maintain function and to limit overoxidation, as observed elsewhere (35). Irreversible PTMs were also quantified by PRM-MS for other FAβox proteins, including hydroxyacyl-CoA dehydrogenase Cys201 and LCAD Cys342 and Cys351, while no significant changes could be detected for Cys111 from enoyl-CoA hydratase.
Six of 8 enzymes from the TCA cycle contained 13 Cys sites (13 peptides) that were irreversibly oxidized, as well as being contained within the significantly altered reversibly oxidized Cys peptide set. These sites included aconitase Cys385, which is part of the Fe-S cluster and required for function (71, 77). Both aconitase and citrate synthase (Cys211 identified here) show greatly reduced activity in I/R, with antioxidant intervention restoring function (71). Our data align with these studies, as both sites belong to clusters showing reduction with MPG and neither could be identified as Cys-SO2H/SO3H after antioxidant intervention. Three Cys sites were identified in isocitrate dehydrogenase (Cys113,402,418). Cys113 and Cys418 are associated with activity; alkylation of Cys113 inactivates the enzyme, presumably by blocking NADP binding, while peroxynitrite induces inactivity via Cys418 SNO (39, 72). Mitochondrial malate dehydrogenase (MDH) is also targeted by oxidants; we identified five irreversibly modified sites (Cys89,93,212,275,285), all of which were contained in peptides found in cluster 4 of the reversible Cys peptide set, and MDH was significantly associated with both 15I and 15I/R in the protein TDE analysis. PRM-MS of MDH Cys275-SO3H confirmed that even in the presence of MPG this site remained irreversibly oxidized at >1.8-fold the levels seen in NITC hearts.
Redox-modified Cys were identified in 5 of 9 glycolytic enzymes (14 sites). Irreversibly oxidized sites including Cys119 (enolase), Cys358 (PYK), Cys127 and Cys218 (TPI), Cys150 (GAPDH) and Cys339 (FBA) are all considered important for activity (1, 24, 29, 47, 70, 76), and several are related to oxidant-generated functional changes (1, 47, 70). We show here that ischemia specifically generates significant SOH modification of full-length GAPDH that is not observed in other conditions. Loss of this reversibly modified GAPDH in I/R strongly suggests overoxidation to SO2H/SO3H at one or more sites, and we observed and quantified 3 irreversibly modified peptides containing Cys150,154 in our PRM-MS data.
Our data also show that antioxidant intervention reverses both I/R-associated changes to irreversible PTM on specific sites within glycolytic enzymes, and allows maintenance of nonischemic levels of corresponding glycolytic intermediates. Addition of MPG during I/R resulted in the preservation of ATP, NADH, pyruvate, and lactate levels consistent with NITC hearts and functional (hemodynamic) recovery, which also suggests restoration of the glycolytic payoff phase through G3P catalyzed by GAPDH. The return of NITC G3P levels in hearts subjected to I/R in the presence of MPG was confirmed by GAPDH enzyme assay that confirmed I/R resulted in the loss of ∼33% of GAPDH activity, which could be restored by MPG. Consistent with this, we observed no effect of MPG treatment on glycolytic intermediates for which the corresponding enzymes were not identified as redox modified (e.g., G6P/F6P and 2-PG/3-PG).
These data indicate that ischemia and I/R result in oxidative inhibition of several enzymes that result in perturbation of essential functional pathways that contribute to the molecular pathogenesis of injury. Since the time frame of our experiments precludes protein synthesis allowing recovery, it is likely that the functional outcome is dependent on the total pool of each protein and the proportion of that pool that is oxidized (particularly irreversibly). Post-I/R recovery is thus likely to be dependent on the rate of (i) reduction of reversibly oxidized sites to their functional form; (ii) proteolytic degradation of potentially toxic overoxidized, nonfunctional proteins; and (iii) synthesis of new protein.
Quantitation of multiple sites within a single protein confirmed that structural constraints and site exposure dictate oxidative PTM. The best example of this could be seen in TPI, for which we quantified 4 peptides containing 5 Cys redox sites (Cys21,27 and Cys42, Cys67 and Cys218). Peptides containing Cys21,27 and Cys42 appeared susceptible to oxidation with both SO3H forms elevated in 15I and 15I/15R, while the Cys67 peptide was not altered under any condition. In addition, the Cys21,27 peptide, unlike Cys42- and Cys218-containing peptides, remained highly oxidized relative to NITC even in the presence of antioxidant.
We mapped all five sites onto the three-dimensional structure of human TPI (hTPI). This showed that Cys21,27 (which are not present in hTPI and were thus treated as mutations) are on the exposed surface of the protein and hence likely to be readily oxidized in the presence of ROS (Supplementary Fig. S14). Similarly, surface exposure of Cys42 and Cys218 could be seen, while Cys67 appears buried at the interface of the hTPI homodimer. A previous report suggested that Cys67 was more solvent accessible than Cys42 or Cys218; however, the same study identified Cys218 as the only site of SNO modification (68). Quantitation of redox-modified sites in other proteins with known structure will further enable confirmation of the structural constraints for reversible and irreversible Cys redox PTMs.
This study represents the largest myocardial Cys redox dataset to date, and is the first to identify and quantify sites of irreversible Cys PTM during ischemia and I/R. Antioxidant intervention ameliorated many reversible and irreversible Cys redox events, which correlated with improved functional recovery and for glycolysis enzymes, the metabolic activities of proteins modified by redox PTM. Cys redox PTMs, particularly those involving loss of protein function, are likely to be a major cause of the pathophysiology of ischemia and I/R.
Materials and Methods
Langendorff perfusion of rat hearts
Animal studies were approved by the Animal Care and Ethics Committee, The University of Sydney (approval K20/4-2001/3/5517). Lewis rats were euthanized with pentobarbital (200 mg/kg), hearts rapidly excised and subjected to Langendorff perfusion with nonrecirculating Krebs buffer as described (54). Contractile function was assessed by insertion of a latex balloon attached to a pressure transducer into the left ventricle. Heart rate (in bpm) and left ventricle developed pressure (LVDP) were recorded (LabChart; ADI Instruments, Australia) as a RPP (bpm × LVDP in mmHg). After a 20 min stabilization period (baseline), hearts were randomized into 1 of 4 protocols: (i) 30 min of perfusion (NITC; n = 6); (ii) 15 min of global no-flow ischemia (15I; n = 9); (iii) 15I/15R (n = 9); or (iv) 15I/15R with 1 mM MPG added to the reperfusion buffer (15I/15R+MPG; n = 9). Hearts were snap frozen in liquid nitrogen and stored at −80°C.
Free thiol assays
Frozen tissue (30 mg per heart) was homogenized (Omni International, Kennesaw, GA) in 0.1 M NaH2PO4, pH 6, containing 20 mM 4,4′-dithiodipyridine, 1% (w/v) sodium dodecylsulfate (SDS), and 2 mM diethylenetriaminepentaacetic acid (DTPA). The homogenate was centrifuged at 13,000 g for 15 min at 4°C, and the supernatant removed before heating at 80°C for 5 min. Aliquots were diluted in assay buffer [0.1 M NaH2PO4, pH 6, 0.2 mM ethylenediaminetetraacetic acid (EDTA)], and the absorbance read at 324 nm (dilution such that absorbance <1 U) to assay total free thiols. The remaining protein extract was subjected to chloroform/methanol precipitation, and the pellets resuspended in 0.1 M NaH2PO4, pH 7.5, 1% SDS, 0.2 mM EDTA, followed by reduction with 20 mM DTT for 30 min at 57°C. Absorbance was taken at 324 nm to assay protein-associated free thiols.
Targeted metabolomics by LC-MS/MS
Frozen tissue (30 mg) was homogenized by 2 × 1 min (50 cycles/s) bead beating with 5 mm stainless steel beads (TissueLyser LT; Qiagen, Hilden, Germany) in 2:1 methanol:chloroform. Samples were centrifuged at 14,000 g for 20 min at 4°C, and the aqueous phase collected and dried by centrifugal evaporation. Metabolites were resuspended in 2:1 methanol:water. LC-MS/MS was performed on a 5500 Q-TRAP (SCIEX, Framingham, MA) coupled to an Agilent 1260 HPLC (Agilent Technologies, Santa Clara, CA).
The MS was operated in scheduled multiple reaction monitoring mode (Supplementary Table S6) with a target scan time of 1 s. Positive ion mode was used in conjunction with a HILIC column (Atlantis HILIC, 100Å, 3 μm, 2.1 × 150 mm; Waters Corp., Milford, MA) and a 24.5 min gradient from 0.5 mM ammonium formate, 95% acetonitrile (MeCN), 0.1% formic acid (FA) to 6 mM ammonium formate, 40% MeCN, 0.1% FA. Negative ion mode was used in conjunction with an amide column (XBridge Amide, 130Å, 3.5 μm, 4.6 × 100 mm; Waters Corp.) and a 16 min gradient from 3 mM ammonium acetate, 3 mM ammonium hydroxide, 85% MeCN to 19.5 mM ammonium acetate, 19.5 mM ammonium hydroxide, 38% MeCN. Metabolite extracted ion chromatograms were processed in Skyline (see the Parallel reaction monitoring MS section). A minimum of n = 9 samples were processed per metabolite per heart group.
Myocardial protein extraction, trypsin digestion, and iTRAQ labeling
One hundred milligrams tissue (pooled from n = 6 hearts per group for 4 biological extractions) was homogenized in 0.1 M NaH2PO4, pH 6, containing 10 mM NEM, 1% SDS, 1 mM phenylmethylsulfonylfluoride, 1 μM aprotinin, 20 μM leupeptin, 1.5 μM pepstatin A, and 2 mM DTPA. The homogenate was centrifuged at 13,000 g for 15 min at 4°C. Additional homogenization buffer was added, and the pellet underwent 2 × 20 s cycles of tip-probe sonication (Branson, Darbury, CT), followed by centrifugation at 13,000 g for 15 min at 4°C, whereupon the supernatants were pooled and an additional 20 mM NEM was added and allowed to react for 10 min before addition of 25 mM DTT. The pH was adjusted to >7, and proteins were reduced for 1 h at 25°C. Excess DTT was quenched, and newly formed free thiols were reversibly modified with MMTS (60 mM in MeCN) for 1.5 h at 30°C.
Proteins then underwent chloroform/methanol precipitation before resuspension in 50 mM NaH2PO4, pH 7.5 and trypsin digestion for 16 h at 37°C with a 1:25 ratio trypsin: protein. The digest was acidified to pH <3 (with FA), and a final concentration of 0.1% trifluoroacetic acid (TFA) added before concentration and desalting of peptides by solid phase extraction (SPE) on hydrophilic-lipophilic balance cartridges (Waters Corp.) and the eluent evaporated to dryness. Peptides were resuspended in 0.5 M triethylammonium bicarbonate (TEAB), pH 8.5, with 0.1% SDS; and quantified by Qubit fluorometry (Thermo Scientific). For each sample, 180 μg peptide was labeled with 4-plex iTRAQ (SCIEX) per the manufacturer's instructions, for 2 h at 25°C, ensuring the final organic percentage was >70 and the reaction pH was >7.6. The reaction was quenched, and any O-acetylation reversed with 0.5 M hydroxylamine for 1.5 h at 25°C. The 4 labeled samples were pooled before drying the combined solution volume by 3/4 to reduce the organic concentration.
Labeled peptides were reduced with 10 mM TBP at 25°C for 30 min, before acidification and dilution to >1 mL with 0.1% TFA for SPE as described above. Protein digestion was performed before TDE to limit missed tryptic cleavage, which were only ∼6.5% for the reversible redox-modified peptide set and 1.4% of the irreversible Cys peptide set. In comparison, on-resin digestion of TDE-captured proteins (27, 35) can result in missed cleavage in excess of 10%. We also noticed that iTRAQ appeared to make samples more hydrophobic (TDE enrichment ∼70%) in comparison with unlabeled peptides [Cys >90%; data not shown and (58)]. Thus, higher concentrations of SDS were utilized in binding (0.75% vs. 0.5%) and wash steps (0.5% vs. 0.1%) to obtain comparative selectivity.
Western blotting
Proteins were prepared as described above. Samples to be used with anti-Cys-dimedone antibody (anti-SOH; Millipore, Billerica, MA) had 1 mM dimedone added to chemoselectively label the SOH group. Proteins were collected at 14,000 g for 15 min at 4°C. Pellets underwent 2 × 1 min bead-beating cycles with glass beads (≤106 μm, acid washed; MP Biomedical, Santa Ana, CA), and chilling on ice between cycles. Samples were centrifuged at 14,000 g for 15 min at 4°C, and the supernatants collected and pooled. For SNO, proteins (100 μg) were precipitated with 6 volumes of chilled acetone and left at −30°C overnight. Samples were centrifuged at 10,000 g for 10 min at 4°C. Pellets were resuspended in HENS Buffer (100 mM [4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid HEPES], 1 mM EDTA, 0.1 mM neocuprione, 1% SDS, pH 7.8) to 1 mg/mL. IodoTMTzero™ labeling reagent (0.4 mM; Pierce Biotechnology, Rockford, IL) and 38 mM sodium ascorbate were added to each sample and incubated at room temperature for 2 h. For negative controls, water was added instead of sodium ascorbate.
Twenty micrograms of protein from each heart group was treated with either nonreducing sample buffer [anti-iodoTMT (anti-SNO), Pierce Biotechnology, 62.5 mM Tris, 2% SDS, 10% glycerol, 0.02% bromophenol blue], or reducing sample buffer (anti-SOH, nonreducing sample buffer with 5% β-mercaptoethanol). Samples that were prepared with nonreducing sample buffer were heated at 95°C for 5 min. Proteins were separated on 12% SDS-polyacrylamide gel electrophoresis (PAGE) gels, and transferred onto polyvinylidene difluoride (PVDF) membranes (Bio-Rad, Hercules, CA) and blotted against a 1:1000 dilution of primary antibody as described (55).
Membranes were washed and incubated with 1:10,000 secondary antibody [goat antirabbit (Enzo Life Sciences, Farmingdale, NY) for anti-SOH, and goat antimouse (Abcam, Cambridge, MA) for anti-SNO] for 1 h. Membranes were incubated in SuperSignal West Pico Chemiluminescent Substrate (Thermo Scientific) for 5 min and imaged on a Chemidoc XRS+ (Bio-Rad). Densitometric analysis was performed in ImageLab (Bio-Rad). Membranes incubated with anti-SNO were stripped, washed, blocked, and then probed with anti-GAPDH primary antibody (Enzo Life Sciences) for 2 h at room temperature.
Multiplexed batch-based peptide TDE chromatography
Thiopropyl-sepharose 6B beads (100 mg) were rinsed in Milli-Q water (2 mL per 10 mg beads) and then with 0.1 M NaH2PO4, pH 7.5 (3 mL). Samples were resuspended in loading buffer (0.1 M NaH2PO4, pH 7.5, 0.75% (w/v) SDS and 1 mM DTPA) and allowed to bind to the resin at room temperature for 2 h with gentle tumbling. The resin and captured peptides were washed once with loading buffer (1 mL), twice with 1 mL of 0.1 M NaH2PO4, pH 7.5, 0.5% SDS for 10 min, and then 6 times with 2 mL 0.1 M NaH2PO4, pH 7.5 for 5 min each. Captured peptides were eluted with 5 × 200 μL of 10 mM DTT in 10 mM NaH2PO4, pH 7.5 each for 15 min. The unbound fraction and first 5 washes were combined, desalted by SPE, and fractionated by SCX to remove any remaining detergent. Elution fractions were pooled, desalted by SPE, and subjected to HILIC as described previously (58).
Offline HILIC peptide separation
Fractionation of peptides was performed on an Agilent 1200 series capillary LC system. Samples were resuspended in HILIC Buffer B (90% MeCN, 0.1% TFA) and separated on a home-packed column (3 μm, 20 cm × 320 μm, 100Å pore size) of TSKgel Amide-80 beads (Tosoh, Tokyo, Japan) using a gradient of 0%–100% HILIC Buffer A (0.1% TFA) for 29 min at a flow rate of 6 μL min−1.
SCX for separation of irreversibly oxidized Cys peptides
Enrichment of irreversibly oxidized Cys peptides was performed as provided in a previous study (57). In brief, peptide samples (70 μg taken from digest before labeling) were diluted in 2 mL SCX loading buffer (5 mM KH2PO4, pH 2.5) and loaded onto an SCX cartridge (SCIEX) equilibrated with the same buffer (3 mL). For PRM-MS, before loading onto the SCX cartridge, a 0.5% “spike-in” of dimethyl labeled oxidized BSA peptide standard (below) occurred. The flow through (FT) was collected, and the cartridge washed with an additional 1 mL SCX loading buffer and pooled with the FT. The cartridge was then washed with organic wash buffer (5 mM KH2PO4, pH 2.5, 25% MeCN), and the wash collected separately (fraction “MeCN”). Retained peptides were eluted (each 1 mL) with varying concentrations of 100% KCl buffer (5 mM KH2PO4, pH 2.5, 25% MeCN, 0.5 M KCl) diluted in organic wash buffer to 10, 25, 40% KCl. All fractions containing MeCN were dried by ∼1/2 to 3/4 to reduce the MeCN concentration, and then diluted with 0.1% TFA to >1 mL before concentration and desalting by SPE as described above. Peptides from the two flow-through fractions (FT, MeCN) were subject to HILIC fractionation (as described in the Offline HILIC peptide separation section).
Liquid chromatography–MS/MS
LC-MS/MS was performed using a 6600 Triple-TOF mass spectrometer (SCIEX) coupled online to an Eksigent nanoLC ultra (Eksigent, Dublin, CA). Samples were resuspended in reversed phase buffer A (0.1% FA) and separated on a home-packed column (3 μm, 15 cm × 75 μm, 120Å pore size) of ReproSil Pur C18 AQ beads (Dr. Maisch GmbH, Ammerbuch-Entringen, Germany) at 2%–35% reversed phase buffer B for 45 or 75 min (TDE unbound) at 300 nL/min−1. The instrument was operated in data-dependent positive ion mode over a m/z range of 300–2000 with the 12 most intense precursor ions (top 15 for unlabeled samples; charge states 2+–4+) selected for collision-induced dissociation fragmentation, and MS/MS transient summing for 250 or 150 msec for iTRAQ-labeled and unlabeled peptides, respectively. Dynamic exclusion was enabled with a 6 s window.
Raw data were scripted to .mgf in Mascot distiller (Matrix Science, London, UK) without MS/MS/MS time domain summing. Data from unbound (total proteome) and TDE bound were searched separately against the UniProt Rattus norvegicus database (taxonomy number 10116; release 2014_07) using an in-house Mascot server. Parameters were one missed cleavage, and variable modifications: methionine oxidation, NEM (Cys), and carbamidomethyl (Cys), and with quantitation by 4-plex iTRAQ selected as a nonvariable modification, where relevant. Another search was performed with the iTRAQ tags searched as variable modifications to ensure labeling efficiency was >90% (data not shown). Peptide/fragment mass tolerances were 30 ppm and 0.1 Da, respectively. All searches were performed against a decoy database, which was used to filter identifications to a <1% false discovery rate (FDR; total proteome) and <2% FDR for modified peptides. Quantified peptides were saved in tab-delimited (.txt) format and processed utilizing an in-house Java script (see the Ratio estimation and Bayesian probability analysis section).
Ratio estimation and Bayesian probability analysis
For normalization, the median was calculated across all nonmodified peptide reporter ion intensities and brought to a ratio of 1:1:1:1 across all four samples using NITC as the control. We determined the relative abundances of each peptide via iTRAQ reporter tag intensities using every discrete MS/MS spectral measurement as a replicate. To do so, we assumed that the measured intensity (abundance) for a given reporter tag in each MS/MS measurement was the product of the expected intensity for that measurement, and the relative abundance [ratio relative to m/z = 114 (NITC)] of that tag. The parameter values (ratios) reported are those that achieved the nonlinear sum of least squares, between the expected and measured intensities, summed across all measurements and reporters for that peptide (Supplementary Data S1).
The relative abundance for the analyzed peptide was taken to be the summed intensity of all measurements for that peptide. Bayesian analysis was then used to determine the uncertainties of these parameters. Furthermore, it was applied to determine whether two tags were significantly different in abundance. This was achieved by considering two hypotheses: (i) that their corresponding ratios are the same within the same error (H1); and (ii) that both have unique ratios and separate errors (H2). If we found that one ratio was r1 = μ1 ± σ1 and the other ratio was r2 = μ2 ± σ2, then, after marginalizing all other parameters, it was found that
A value of zero corresponds to when H1 and H2 are equally plausible, whereas negative values indicate a greater likelihood that the ratios are significantly different. A Loge(Prob) < −2 is equivalent to a <1% probability of H1, and was the significance threshold applied to all ratios.
Bioinformatics
Cluster analysis of regulated reversibly redox-modified Cys peptides was undertaken using GProX (66). Functional analysis was performed using STRINGdb
Protein TDE of heart mitochondrial proteins
Heart tissue was homogenized in 8 volumes of sucrose-mannitol buffer (220 mM mannitol, 70 mM sucrose, 2.0 mM HEPES, 10 mM NEM, 2 mM DTPA pH 7.4). Mitochondria were isolated by differential centrifugation as previously described, with the addition of 10 mM NEM and 2 mM DTPA (82). Isolated mitochondria were reconstituted in 0.1 M NaH2PO4 pH 6.0, 1% SDS, 10 mM NEM, 2 mM DTPA, and additional alkylation performed for 10 min before excess NEM was quenched with addition of 25 mM DTT. The pH was adjusted to >7.0, and proteins reduced for 1 h at 25°C. Proteins then underwent chloroform:methanol precipitation. TDE was performed as previously described (58). TDE-bound proteins were concentrated by molecular weight cutoff filters (Vivaspin 500, 3000 MWCO) and separated by SDS-PAGE.
Gels were stained in Sypro Ruby (Thermo Scientific) and imaged on a Chemidoc XRS+ imager (Bio-Rad). Image densitometry was performed using ImageLab (Bio-Rad, v.6.01, build 34). Bands were excised and destained (25 mM NH4HCO3, 50% MeCN) before drying by vacuum centrifugation. Gel pieces were rehydrated for 15 min with 100 ng trypsin in 25 mM NH4HCO3. An additional 2.5 volumes of 25 mM NH4HCO3 were added, and gel pieces were incubated at 37°C for 16 h. Digests were acidified with 0.1% TFA, and peptides concentrated and desalted using home-packed C18 columns (POROS 20 R2 reversed phase resin; Thermo Scientific). Bound peptides were washed twice with 0.1% TFA and eluted with 0.1% TFA, 50% MeCN. Eluents were dried by vacuum centrifugation, reconstituted in 0.1% FA, and analyzed by LC-MS/MS as described above.
Parallel reaction monitoring MS
BSA was resuspended in FA to 1 mg/mL and 25 μg treated with 3 volumes of perfomic acid. This was left to react on ice for 2 h, after which 5 volumes of water were added and the solvent removed by vacuum centrifugation. Oxidized BSA was resuspended to 1 mg/mL in 100 mM TEAB (pH 7.4) and digested with trypsin (1:35 trypsin to protein) overnight at 37°C. BSA digest was acidified with 0.1% TFA and desalted on C18 Zip-Tip cartridges (Millipore). Before elution, 5 volumes of dimethyl labeling solution [0.2% (v/v) formaldehyde, 30 mM sodium cyanoborohydride, 50 mM NaH2PO4 pH 7.5] were passed through the column, with each volume allowed to react for 10 min with bound peptides. Columns were washed with a further 5 volumes of 0.1% TFA before elution with 1.5 volumes of elution buffer (70% MeCN, 0.1% TFA). LC-MS/MS was performed on a Q-Exactive HF (QE-HF) coupled online to an Easy-LC UHPLC (Thermo Scientific).
Peptides were separated on a home-packed column (1.9 μm, 40 cm × 75 μm 120Å pore size) of ReproSil Pur C18 AQ beads using a 140 min gradient stepping from 5% to 100% Buffer B (90% MeCN, 0.1% FA). The MS was operated in data-dependent acquisition (DDA) positive ion mode, taking the top 20 most intense precursor ions with charges 2+–4+ for higher energy collisional dissociation MS/MS. This occurred with a dynamic exclusion list of 20 s. The Orbitrap was operated with a range of 300–1650 m/z for a resolution of 60,000 (QE-HF) at m/z 200 for full MS scans, with a range of 100–2000 m/z and a resolution of 15,000 at m/z 200 for MS/MS. PRM occurred with the instrument operating in data-independent acquisition mode with a 1.4 m/z isolation window. This consisted of a full MS scan followed by a loop of the scheduled inclusion list with 7.5 min windows; MS parameters were as stated above (n = 3 injection replicates per n = 3 biological replicates).
Raw data were processed using Proteome Discoverer 2.2 with Cys-NEM (+125.048 Da), Met oxidation (+16.00 Da), Cys-SO2H (+32.00 Da), and Cys-SO3H (+47.98 Da) as variable modifications. Data were searched against the UniProt R. norvegicus proteome (release 2016_07) using an in-house Mascot server with a 2% FDR. To identify the internal standards, each dataset was also searched against the UniProt Bos taurus proteome (taxonomy number 9913; release 2016_07) with the addition of N-terminal (+28.03 Da) and Lys dimethylation (+28.03 Da). Tryptic peptides without Met and with Cys-SO2H or Cys-SO3H PTM were analyzed.
The DDA results from each biological condition were used to generate a combined spectral library in Skyline. Raw data from scheduled PRM were imported into Skyline, which matched peptides against the spectral library. The window of integration for each peptide was set to the base of the chromatographic peak, and the integral of each transition was summed across each peptide. The sum of the peak areas for each peptide was then normalized to the sum of all/BSA standard peptide peak areas per injection replicate. These were then expressed per group as a ratio to NITC. Peak areas were then expressed per heart treatment group as a ratio to NITC for each of n = 3 independent experiments and the mean ratio ± standard deviation used to calculate significance (Supplementary Data S9).
GAPDH assay
Tissues were homogenized by 1 × 1 min (50 cycles/s) bead beating with 5 mm stainless steel beads (TissueLyser LT; Qiagen) in ice-cold GAPDH Assay Buffer (10 mg tissue/120 μL buffer), and assays conducted using a GAPDH Activity Assay Kit (Sigma-Aldrich) according to the manufacturer's instructions.
Footnotes
Author Disclosure Statement
No competing financial interests exist.
Funding Information
This work was supported in part by a grant from the National Health and Medical Research Council of Australia (571002 to S.J.C.). M.Y.W. was supported by an Australian Research Council Discovery Early Career Researcher Award (DECRA). A.W.R., J.P., K.A.L., M.P., and D.K.L. were supported by Australian Postgraduate Awards.
Supplementary Material
Supplementary Data S1
Supplementary Data S2
Supplementary Data S3
Supplementary Data S4
Supplementary Data S5
Supplementary Data S6
Supplementary Data S7
Supplementary Data S8
Supplementary Data S9
Supplementary Figure S1
Supplementary Figure S2
Supplementary Figure S3
Supplementary Figure S4
Supplementary Figure S5
Supplementary Figure S6
Supplementary Figure S7
Supplementary Figure S8
Supplementary Figure S9
Supplementary Figure S10
Supplementary Figure S11
Supplementary Figure S12
Supplementary Figure S13
Supplementary Figure S14
Supplementary Table S1
Supplementary Table S2
Supplementary Table S3
Supplementary Table S4
Supplementary Table S5
Supplementary Table S6
Abbreviations Used
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
