Chemoproteomics Analysis of S-Sulfenylated Cysteines in Arabidopsis Cells.

Based on the previously reported protocols (21⇓–23), an adapted chemoproteomics workflow was applied for in situ Cys-SOH detection and quantification in Arabidopsis cell cultures in the presence of H₂O₂ (Fig. 1A). First, we optimized the Cys-SOH labeling conditions with the BTD probe by varying the H₂O₂ concentrations. Whereas the protein S-sulfenylation levels, reflected by the biotinylation signals on Western blots, increased with the addition of 100 μM to 400 µM BTD, it decreased for H₂O₂ concentrations higher than 100 µM. This decrease in S-sulfenylation levels with increasing H₂O₂ concentrations suggests overoxidation of SOH to SO_2/3H (SI Appendix, Fig. S1). To capture and identify BTD-labeled peptides, samples were processed by a well-established chemoproteomics workflow (21) (Fig. 1A). Specifically, Arabidopsis cell cultures were incubated with 400 µM BTD and treated with or without (control) 100 µM H₂O₂ for 25 min. After protein extraction and tryptic digestion, BTD-modified peptides from H₂O₂-treated and control cells were conjugated with light and heavy azido-biotin reagents with an ultraviolet (UV)-cleavable linker, respectively, through a copper-catalyzed alkyne azide cycloaddition reaction (also known as click chemistry) and mixed together in a 1:1 ratio. After enrichment on streptavidin beads and photorelease at 365 nm, the resulting peptides were analyzed by liquid chromatography tandem mass spectrometry (LC-MS/MS) (Fig. 1A).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Quantitative Cys-SOH identification in Arabidopsis. (A) Workflow for in vivo S-sulfenylation labeling and analysis in Arabidopsis cell cultures. The BTD-based probe was sketched as an arrow (chemical warhead) followed by 3 lines indicating the alkyne handle. (B) Mapping of the 1,537 identified Cys-SOHs sites on 1,394 (Araport11) proteins, of which 200 proteins were identified to be S-sulfenylated previously (17⇓–19) (Right) and 1,194 were discovered as protein targets (Left) (Dataset S1). (C) Ratios (R_H2O2/Control, y axis) plotted for 463 Cys-SOH sites quantified in at least 2 out of 3 replicates (x axis). Ratios of at least 1.5-fold increases (red, 92 Cys-SOHs) or decreases (green, 3 Cys-SOHs) are shown. (Inset) For the catalytic Cys of MsrB3/7/8/9 (encircled in bold), the extracted-ion chromatogram is displayed (H₂O₂/light in red, and control/heavy in blue).

In total, we detected 6,399 peptide-spectrum matches, corresponding to probe-modified peptides in 3 biological replicates, resulting in the identification of 1,537 unique Cys-SOH sites on 1,394 proteins (Fig. 1B and Dataset S1). By comparison with the DYn-2 carbon nucleophile and the YAP1C genetic probes, in this study, we mapped 315 Cys-SOH sites of approximately half of the previously identified S-sulfenylation proteins (48.5%, 200 out of 412 proteins) (Fig. 1B and Dataset S1). More importantly, an additional 1,222 Cys-SOHs mapped to 1,194 proteins that had not been identified previously (17⇓–19). Taken together, this study represents a detailed compendium of protein S-sulfenylation sites in oxidation-challenged plant cells.

Comparison of precursor intensities of light (H₂O₂-treated) and heavy (control) BTD-modified peptide pairs enabled the quantification of 72.4% (1,105) of S-sulfenylation events under H₂O₂ stress, among which 463 sites (∼42%) could be quantified in at least 2 out of 3 biological replicates, with median and average relative SD of 6.61% and 8.95%, respectively (SI Appendix, Fig. S2). Not unexpectedly, S-sulfenylation levels of ∼20% (92/463) of these cysteines increased in Arabidopsis cells upon H₂O₂ treatment [H₂O₂ to control ratios, R_H2O2/Ctrl ≥ 1.5, representing a threshold value commonly used in redox proteomics (22, 24, 25) as an indication of significant difference], whereas a majority (79.5%) of Cys-SOHs remained unchanged (Fig. 1C and Dataset S1). For instance, the BTD-modified peptide “HC_BTDVNSVSLK,” which entails the active sites (Fig. 1C) of the MsrB3, Msr7, Msr8, and Msr9, shows an elevated intensity upon H₂O₂ treatment (R_H2O2/Ctrl = 3.60; SI Appendix, Fig. S3). It is well known that the formation of a Cys-SOH in the MsrB-active site is an essential step in the catalytic cycle of MsrBs of several species (3, 11, 26). As such, this increased S-sulfenylation on the catalytic cysteine of MsrB results, probably, from an increased MsrB activity due to the applied H₂O₂ stress on the Arabidopsis cells. In addition, only 3 Cys-SOHs showed decreased S-sulfenylation levels under H₂O₂ treatment (R_H2O2/Ctrl < 0.67), as an indication of overoxidation to the sulfinic or sulfonic acids.

Notably, ∼58% (642/1105) of the S-sulfenylated sites could be quantified in only one replicate due to the intrinsic limitation of shotgun proteomics in a data-dependent acquisition mode; hence this result should be interpreted with caution. It is also important to note that our chemoproteomics approach is mainly designed for sulfenic acid identification and ratiometric quantification, rather than stoichiometric measurement. As such, a ratiometric increase for a specific site might only correspond to a minor stoichiometric difference. Without doubt, there is a continued need to develop methods that can globally measure site occupancy of this distinct type of cysteine oxidation, although the total cysteine oxidation content can be determined by a well-established chemoproteomics approach designated oxydative isotope-coded affinity tag (27). Furthermore, a more complete picture could be obtained by combining chemoselective probes for oxidized cysteines, as recently done for Cys-SOH (22).

Bioinformatics Analysis on the Target Specificity of S-Sulfenylation.

To find targeted processes or protein hotspots sensitive to S-sulfenylation, we first determined the subcellular distribution of the identified S-sulfenylated proteins (Fig. 2A and Dataset S1) (28, 29). The best-represented subcellular compartments are cytoplasm (48%), chloroplast (18%), mitochondrion (13%), and nucleus (12%). In comparison to the previous subcellularly oriented cytosolic and plastid YAP1C approaches (17, 18), and the DYn-2 approach (19), the current workflow provides a more diverse spectrum of proteins in terms of subcellular location (SI Appendix, Fig. S4).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Subcellular localization distribution and functional enrichment of S-sulfenylation events. (A) Subcellular distribution of 809 S-sulfenylated proteins defined as high-confidence markers (HCMs) by the Multiple Marker Abundance Profiling tool (28, 29). Organelles are colored proportionally to the S-sulfenylated HCMs. (B) Gene set enrichment of S-sulfenylated proteins for KEGG pathways (blue) and custom protein classes (SI Appendix, Materials and Methods) (Top). Fold enrichment (y axis) is plotted as a function of statistical significance (minus log Q value, x axis). Node size corresponds to the number of S-sulfenylated proteins. Bar chart outlining the number of Cys-SOHs for the enriched KEGG pathways and protein classes (Bottom). Induced Cys-SOHs (R_H2O2/Ctrl ≥ 1.5) are plotted, and their respective proportion is displayed. (C) Cys-SOH domain enrichment (SI Appendix, Materials and Methods). Fold enrichment (y axis) is plotted as a function of statistical significance (minus log Q value, x axis). Node size is scaled to the number of Cys-SOHs. ER, endoplasmic reticulum; PM, plasma membrane.

To dissect the biological processes targeted by S-sulfenylation, we performed gene set enrichment of the 1,394 identified S-sulfenylated proteins for the Kyoto Encyclopedia of Genes and Genomes (KEGG) metabolic pathways and custom protein classes (Dataset S2). It should be noted that, in this overrepresentation analysis, a certain bias toward abundant protein classes and pathways might arise due to the favorable detection of more-abundant peptide species by MS. Consistent with our previous observations (2, 18, 19), the S-sulfenylated proteins are overrepresented in the essential metabolic pathways (Q value ≤ 1E⁻⁶), such as “carbon metabolism,” “glycolysis/gluconeogenesis,” and “amino acid biosynthesis” (Fig. 2B and Dataset S2). RNA-binding proteins (RBPs) (247 out of 1,786 or 13.8%, Q value ≤ 1E⁻⁴⁶) were strongly overrepresented, directly corroborating the observed enrichment of KEGG pathways involved in translation, including “aminoacycl-tRNA (AA-tRNA) biosynthesis,” “RNA transport,” and “ribosome” (Fig. 2B and Dataset S2). Furthermore, several Cys-SOH sites of these RBPs showed an increased S-sulfenylation (R_H2O2/Ctrl ≥ 1.5), with, for instance, 7 out of 44 Cys-SOHs (15.9%) of AA-tRNA synthetases (SI Appendix, Table S1). Notably, similar RNA and metabolic processes had been specified as redox-sensitive in human colon cervical alveolar adenocarcinoma cells lines (22, 23), suggesting that these processes are a common S-sulfenylation target across species.

Using the precise knowledge of Cys-SOHs residues, we identified protein domain hotspots sensitive to Cys S-sulfenylation by performing a Cys-SOH protein domain enrichment (Fig. 2C) (SI Appendix, Materials and Methods and Dataset S2). The high prevalence of RBPs is emphasized further by the strong enrichment of protein domains involved in translation, such as “Translational GTPase,” “AA-tRNA ligase II,” and “RNA helicase adenosine 5′-triphosphate (ATP)-binding I” domains. In addition, 7 Cys-SOHs are present in the RNA recognition motif (RRM) domain (Dataset S2), including 5 Cys-SOHs that map to poly(A)-binding proteins (PABPs) and contain 4 highly conserved RRM domains in tandem (30). For PABP4, 3 identified Cys-SOHs were located within RRM domains, namely Cys163, Cys266, and Cys357. Given the increase in the Cys163 and Cys357 S-sulfenylation of PABP4 by H₂O₂ stress, Cys-SOH could affect the poly(A)-binding capacity and, hence, translation. Another enriched oxidation-sensitive domain is the Ras-related nuclear protein (Ran)-binding domain (Fig. 2C), which is essential for nuclear transport and facilitates nuclear assembly and cell cycle progression (31). Almost all Cys-SOHs in the Ran-binding domain show H₂O₂-increased S-sulfenylation levels (Dataset S1), suggesting a potential redox regulatory role of this Ran-binding domain in the Ran cycle. In addition, the WD40 and Ankyrin repeat domains are also highly enriched (Fig. 2C), both mediating protein−protein interactions in diverse signaling processes (32, 33).

S-Sulfenylation at Specific Cysteine Sites Affects Protein Functionality.

To estimate the consequences of the identified Cys-SOH sites, we cross-referenced them with known functional sites curated by the UniProt knowledgebase (https://www.uniprot.org/) (34), and obtained 28 functionally annotated Cys-SOHs (SI Appendix, Table S2), including 10 sites validated by mutagenesis studies, 3 sites recorded from protein structure analysis, and 15 sites that were inferred in silico based on homology.

From the 10 Cys-SOH sites studied by mutagenesis (SI Appendix, Table S2), 6 are located within the protein catalytic site. Redox regulation by S-sulfenylation has already been shown for the catalytic sites of GAPDH C SUBUNIT 1 (GAPC1) (8) and STARCH-EXCESS 4 (SEX4) (4), but potential redox regulation of METACASPASE4 (MC4), NITRILASE 4 (NIT4), UBIQUITIN-PROTEIN LIGASE 1 (UPL1), and LIKE-SEX FOUR 2 (LSF2) remains to be investigated. Besides active sites, Cys-SOHs in Mg-CHELATASE SUBUNIT 1 (CHI1) (35), GLUTAREDOXIN16 (GRXS16) (36), and the peptidyl-prolyl cis-trans isomerase CYP20-3 (37) have been implicated in intradisulfide or interdisulfide bonds crucial for protein conformation and enzymatic activity (SI Appendix, Table S2). Similarly, S-sulfenylated Cys165 forms a disulfide bridge with Cys150 of TenA_E, based on protein structure information (38) (SI Appendix, Fig. S5A). In both the ADH1 and ADH2 structures, Cys47 is involved in the coordination of Zn²⁺ in the catalytic site (SI Appendix, Fig. S5 B and C) and was found here as S-sulfenylated. Recently, oxidation of yeast ADH1 Cys47 has been suggested to cause enzymatic inactivation (39), and the redox sensitivity of Arabidopsis ADH1 Cys47 has been shown (40). Whereas supply of NAD⁺ or NADH cofactors prevented enzymatic inactivation by H₂O₂, Cys47 mutation to Ser abolished the enzymatic activity (40). Hence, the catalytic Zn²⁺ coordination by Cys47 probably acts as a H₂O₂-sensitive switch that can fine-tune the enzymatic activity. Taken together, the identified Cys-SOH in several proteins implies either the formation of a disulfide bond with a neighboring cysteine, loss of metal coordination, or inactivation of a specific activity. However, the detailed functional impact on protein and cellular levels remains largely unspecified and awaits further analysis.

S-Sulfenylation in Arabidopsis and Human Protein Orthologs.

To compare the Arabidopsis study with the reported BTD-labeling profile in human (20), we examined the linear sequence motifs in the environment of the BTD-identified Cys-SOHs by constructing an iceLogo (41), visualizing overrepresented and underrepresented amino acids (Fig. 3A). Cysteine residues were clearly underrepresented and mostly predominant at position −3 and +3 relative to the BTD-labeled position. The absence of a cysteine is a stabilizing factor for a SOH, because its close presence mediates disulfide formation (42) (Fig. 3A). Remarkably, several basic amino acids surround the BTD cysteine sites. Whereas lysine is prevalent in human, histidine is primarily present in Arabidopsis (Fig. 3A). Positively charged histidine and lysine residues are the preferred hydrogen-binding partners for cysteines, as they reduce their pK_a and increase their sensitivity to oxidation (43). For instance, the His preceding the active site of SEX4 (SI Appendix, Table S2) has been suggested to decrease the pK_a of the Cys198 and to facilitate a thiolate nucleophilic attack (44).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Evolutionary comparison of Arabidopsis and human Cys-SOH sites. (A) Overrepresentation motifs of Cys-SOH sequence windows (−6 to +6 residues) of human (20) and Arabidopsis BTD-labeled sites constructed by iceLogo (39). Basic residues Lys, Arg, and His are in blue, and Cys is in red. (B) Overview of Arabidopsis Cys-SOH alignment scenarios and their occurrence. (C) Examples of Arabidopsis (green)–human (blue) protein sequence alignments with exact alignments of S-sulfenylation events (red, with protein position indicated). Ratios (R_H2O2/Ctrl) were derived from this study (Dataset S1), with the number of biological replicates indicated between parentheses, and as reported for human RKO cells treated with H₂O₂ (23).

Further, we compared 1,537 Cys-SOH sites from this BTD dataset and those identified by the DYn-2 and BTD probes from human studies (20, 22, 23). After protein sequence alignment of human−Arabidopsis orthologs, the relative positions of Cys-SOHs were evaluated, and 4 Cys-SOH alignment categories were distinguished (Fig. 3B). In addition to the Arabidopsis Cys-SOH sites that lack human protein orthologs (313 Cys-SOHs, 20%) or without S-sulfenylated human orthologs (350 Cys-SOHs, 23%), the majority of the Arabidopsis Cys-SOHs does not align exactly with the human Cys-SOH positions (719 Cys-SOHs, 47%). However, 155 Arabidopsis Cys-SOH (10%) correspond to those in one or multiple human protein orthologs (Fig. 3B and Dataset S1).

These 155 Cys-SOH sites match exactly the human−Arabidopsis orthologs, possibly indicating that these sites are conserved oxidation-sensitive sites. Moreover, for some of these proteins, Cys redox regulation has been documented in different species, such as for the catalytic Cys site of GAPDH (8, 10) and MsrB (3, 11) (Fig. 3C). Remarkably, in the case of GAPDH, oxidation-sensitive sites were assigned to 2 adjacent cysteines (Cys156 and Cys160) (SI Appendix, Fig. S6). Furthermore, the NKCD motif of Ras GTPases has been reported to be redox-sensitive (45, 46), and S-sulfenylation of Cys126 of recombinant human Rab-1A was confirmed with the carbon nucleophilic probe DAz-2 (47), but no information is available yet for the plant orthologs. The conserved Cys-SOHs of the human Ran-BINDING PROTEIN 1 (RANBP1) and its 3 Arabidopsis orthologs are located within the Ran-binding domains (Fig. 3C). In human, cellular stress and elevated H₂O₂ levels disrupt the Ran system, and this was partly attributed to cysteine oxidation of the nucleotide exchange factor RCC1, although independent mechanisms were suggested (48). Cys-SOH in the active sites of AtLSF2 and AtSEX4 (SI Appendix, Table S2) align with the nucleophilic Cys of 2 human Protein Tyr Phosphatases (PTPs), VACCINIA H1-RELATED PHOSPHATASE (HsVHR) and SLINGSHOT HOMOLOG3 (HsSSH3) (Fig. 3C). Low H₂O₂ concentrations were reported to oxidize and inactivate HsVHR reversibly (49). Another human PTP, HsSSH1, is activated under oxidative stress due to Cys oxidation of its negative interactor Hs14-3-3ζ (50). The Cys94 of Hs14-3-3ζ was identified as oxidation-sensitive, and its mutation to Ala or Ser causes significant protein destabilization (51). Notably, 3 S-sulfenylated cysteines from the Arabidopsis orthologous proteins, general regulatory factors, align with Cys94 of Hs14-3-3ζ (Fig. 3C). To explain potential functional roles of these cysteines, comparison of Cys-SOH sites between protein orthologs can help prioritizing cysteine functional sites for mutagenesis studies.

Functional Importance of Cys181 in AtMAPK4.

Arabidopsis mitogen-activated protein kinases (AtMAPKs) play important roles in plant signaling (52) and have been shown to be activated by H₂O₂ (53⇓⇓–56), although the precise mechanism remains elusive. Among the 155 Cys-SOH sites that exactly match the human−Arabidopsis orthologs (Fig. 3B and Dataset S1), a Cys-SOH site occurred in a unique peptide “DLKPSNLLLNANC_BTDDLK” (SI Appendix, Fig. S7) that can be mapped to several AtMAPKs, including AtMAPK3, AtMAPK4, AtMAPK6, and AtMAPK11, of which AtMAPK4 had previously been trapped by the YAP1C probe (18). The identified S-sulfenylated Cys from AtMAPK4, Cys181, resides in a conserved motif (MAP Kinase signature) that distinguishes MAPKs from other kinases (Fig. 4A) (57). Our comparative analysis with the human Cys-SOHs indicated an exact alignment with Cys161 of HsMAPK1 (Fig. 4A), also known as EXTRACELLULAR SIGNAL-RELATED KINASE2 (HsERK2). Besides sequence alignment, superimposing a structural model of AtMAPK4, based on the structure of AtMAPK6 (58), with HsMAPK1 (59) revealed an almost identical structural environment of Cys181 in AtMAPK4 and Cys161 in HsMAPK1 (SI Appendix, Fig. S8). For both structures, a conserved proximal His (AtMAPK4 His145, HsMAPK1 His125) acts as a possible hydrogen bond donor and could thereby decrease the pK_a of the thiol by stabilizing it as a thiolate (43), which increases the oxidation sensitivity of the cysteines (SI Appendix, Fig. S8). Notably, Cys161 resides in a conserved docking site that governs the regulatory protein interactions of HsMAPK1 (60⇓–62).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Sulfenylation of Cys181 of AtMAPK4 that is crucial for the kinase activity. (A) S-sulfenylation of the H₂O₂-sensitive Cys in the MAP kinase signature (PS01351 PROSITE pattern) in Arabidopsis (Cys176 in AtMAPK3; Cys181 in AtMAPK4; Cys201 in AtMAPK6; and Cys178 in AtMAPK11) and human (Cys161 in HsMAPK1). (B) Dimedone blot showing AtMAPK4 S-sulfenylation under 1 mM H₂O₂ for 1 h. (C). Annotated spectrum match of ‘DLKPSNLLLNANC₁₈₁DLK’ with Cys181 modified by dimedone (+138.0 Da). (D) Kinase activity of AtMAPK4 and its C181S variant as measured on phosphorimage displaying myelic basic protein (MBP) phosphorylation with [³²P]ATP. (E) Kinase activity of autophosphorylated (P_i) and nonautophosphorylated AtMAPK4 and its C181S variant as measured by quantitative Cerenkov counting after 45 min of incubation with MBP.

In recombinant HsMAPK1, S-sulfenylation was reported to decrease kinase activity toward a substrate protein, Elk1, with Cys161 and/or Cys166 proposed as potential oxidation-sensitive sites that could influence kinase activity given their location within the docking site (63). Noteworthy, the HsMAPK1 Cys166 corresponds to Gly in Arabidopsis MAPKs; hence, we decided to investigate the functional effect of AtMAPK4 Cys181 that corresponds to the HsMAPK1 Cys161. First, we confirmed the S-sulfenylation of H₂O₂-oxidized recombinant AtMAPK4 with dimedone blots (18) (Fig. 4B) and the maintenance of its overall structure upon oxidation with far UV circular dichroism (CD) (SI Appendix, Fig. S9). MS/MS analysis showed that Cys181 of the recombinant AtMAPK4 was consistently detected as oxidation-sensitive (Fig. 4C) as other cysteines (Cys6, Cys181, Cys232, and Cys341) (SI Appendix, Table S3 and Dataset S3). To study the functional effect of only Cys181, which was identified as redox-sensitive in vivo, we mutated Cys181 into serine (AtMAPK4_C181S). Similarly to the oxidized AtMAPK4, CD spectra indicated a conserved secondary structure composition without drastic structural alterations, due to the Cys181 mutagenesis of AtMAPK4 (SI Appendix, Fig. S10). Next, kinase activities were evaluated by the incorporation of [γ-³²P] ATP in the myelin basic protein (Fig. 4 D and E). In the AtMAPK4_C181S variant, the kinase activity was significantly lower than that of the wild-type AtMAPK4 (Fig. 4 D and E), implying that Cys181 affects the AtMAPK4 kinase activity. As MAPK family proteins can also be phosphorylated on specific Tyr/Thr/Ser residues (64), we identified Tyr203 of recombinant AtMAPK4 to be prone to autophosphorylation (SI Appendix, Table S3 and Dataset S3). Tyr203 belongs to a conserved TEY amino acid activation segment of the TEY subgroup of MAP kinases. Therefore, we made sure that both enzymes were compared in exactly the same experimental setup to allow the control of autophosphorylation (Fig. 4E). Control of the autophosphorylation confirmed the decrease of the kinase activity for the AtMAPK4_C181S variant. Taken together, the oxidation-sensitive Cys181 of AtMAPK4 seems to play an important role in the kinase activity, but how Cys181 regulates the AtMAPK4 activity is still unclear, especially because Cys181 is located remotely from the active site. To investigate the precise role of Cys181 in the kinase mechanism, experiments with AtMAPK4 native substrates and its MAP kinase kinase within its in vivo redox environment will be required.