Thiol redox proteomics: Characterization of thiol‐based post‐translational modifications

Redox post‐translational modifications on cysteine thiols (redox PTMs) have profound effects on protein structure and function, thus enabling regulation of various biological processes. Redox proteomics approaches aim to characterize the landscape of redox PTMs at the systems level. These approaches facilitate studies of condition‐specific, dynamic processes implicating redox PTMs and have furthered our understanding of redox signaling and regulation. Mass spectrometry (MS) is a powerful tool for such analyses which has been demonstrated by significant advances in redox proteomics during the last decade. A group of well‐established approaches involves the initial blocking of free thiols followed by selective reduction of oxidized PTMs and subsequent enrichment for downstream detection. Alternatively, novel chemoselective probe‐based approaches have been developed for various redox PTMs. Direct detection of redox PTMs without any enrichment has also been demonstrated given the sensitivity of contemporary MS instruments. This review discusses the general principles behind different analytical strategies and covers recent advances in redox proteomics. Several applications of redox proteomics are also highlighted to illustrate how large‐scale redox proteomics data can lead to novel biological insights.


Reversible Irreversible
F I G U R E 1 Thiol-based redox PTMs. Note that SO 2 H can be reversed by sulfiredoxin [8]. ROS, Reactive oxygen species.

INTRODUCTION
Reactive oxygen species (ROS) are critical secondary messengers for cellular signaling and regulation [1][2][3].  [7] and more broadly for over 50 other proteins [8]. Many of these redox PTMs play an important role in the regulation of protein function, localization, and/or activities (i.e., redox switches) [9,10]. Dysregulation of redox signaling, which is caused by an imbalance between ROS and antioxidant defense systems, can lead to various pathological conditions including diabetes, cancer, and neurodegenerative diseases [11][12][13].
With an increasing appreciation for redox signaling, there has been a growing interest in characterizing redox PTMs at the proteome scale (i.e., redox proteomics). Mass spectrometry (MS) has become the leading platform for large-scale PTM profiling, including redox PTMs [14][15][16]. The emerging field of redox proteomics is focused on advancing analytical approaches for enabling quantitative, site-specific measurements of various types of redox PTMs to decipher redox-dependent mechanisms in various biological systems [17][18][19]. Redox proteomics approaches can be categorized into several general strategies according to their chemical principles. The first strategy is "selective reduction and tagging," which was first demonstrated by the biotin-switch technique (BST) for detecting SNO [20,21]. With this strategy, all free thiols are initially blocked by an alkylation reagent (e.g., N-ethylmaleimide, NEM). Redox PTMs of interest are then selectively reduced to free thiols, which are then tagged with thiol reactive reagents that also bear an affinity tag (e.g., biotin) for enrichment. Given the many versions of this strategy developed to date, it is the most widely applied method due to its simplicity in chemistry, adaptability to different PTM types, and seamless compatibility with MS-based proteomics [17,[22][23][24]. The second major strategy employs chemoselective probes that directly label specific types of redox PTMs to enable further tagging and enrichment for LC-MS measurements [15,25,26]. Due to the distinct chemical nature of various redox PTMs, various chemoselective probes have been developed, with many adapted for proteomic profiling [4]. In addition to these two general strategies, recent studies have demonstrated direct MS detection of redox PTMs without chemical derivatization, which had been a long-standing interest in the field [27].
Recent advances have made redox proteomics an indispensable tool in fundamental biological studies and applied biomedical research. The purpose of this review is to summarize recent research endeavors in advancing analytical approaches including chemistries that permit one to label, quantify, and enrich certain types of redox PTMs, chemical tools for labeling specific PTM types, and approaches for quantifying site-specific PTM occupancies. We also highlight several recent examples of redox proteomics applications that have generated novel biological insights.

GENERAL CONSIDERATIONS FOR THIOL REDOX PROTEOMICS
Most protein cysteine thiols in living cells exist in a reduced state (80%-90%) [28] making many cysteine thiols prone to artificial oxidation during sampling, storage, and processing. Thus, preserving the endogenous redox state of the proteome samples necessitates efforts to minimize artificial oxidation with several general considerations.
Firstly, one needs to carefully consider the sampling process to avoid introducing unnecessary redox activity. For instance, components of cell culture media such as phenol red [29], certain nutrients (or lack thereof) as well as cultivation conditions (e.g., light exposure) can influence the redox proteome and need to be well controlled for comparative studies. Secondly, free thiols can be easily oxidized during sample processing steps such as lysis and tissue homogenization. Several methods have been developed to preserve (at least in part) redox states; these include trichloroacetic acid (TCA) quenching of thiolate reactivity during cell harvesting and free thiol blocking with highly reactive alkylating reagents such as NEM or iodoacetamide (IAM) [17,[30][31][32]. The choice of alkylating reagents and buffer pH are critical for trapping free thiols and minimizing artificial oxidation during the initial blocking steps. For example, the kinetics of NEM blocking are faster than IAM and can effectively alkylate protein thiols at lower pH (4.3 vs. 8.0) in comparison with IAM [33]. Thirdly, other factors can potentially cause artifacts during sample processing. For instance, metal ions, such as Cu 2+ and Fe 3+ (even in trace quantities) can catalyze thiol oxidation [34,35] or decomposition, as is the case with SNO [36]. The inclusion of metal chelating reagents (e.g., EDTA or neocuproine) in cell extracts or tissue homogenates is important for mitigating these undesirable side-reactions. Exposure to indirect sunlight (from an adjacent window) resulted in artifactual ascorbate-dependent signals for SNO detection using the BST [37]. Due to these factors, it is important to include appropriate negative and/or positive controls when designing experiments involving analysis of redox proteomes [31]. Finally, it is important to consider how the redox states of samples can be preserved if they are not processed immediately. In general, cell pellets or tissue samples can be preserved through flash freezing and −80 • C storage [38], but it is still unclear how long samples can be stored without significant impact on redox states.

SELECTIVE REDUCTION AND TAGGING-BASED APPROACHES
In principle, MS is an ideal tool to directly detect and identify PTMs.
However, direct detection of redox PTMs has proven to be challenging due to the transient and labile nature of certain PTMs. For example, SOH, an important modification in redox signaling, is very labile and has the potential for further oxidation to SO 2 H, conversion to SSG (and protein disulfides), or reduction back to a free thiol [39]. Another example: SNO is UV-sensitive and unstable at high temperatures [40]. To address this, a major redox proteomics strategy is to quantify "thiol oxidation" (i.e., all reversible redox PTMs) instead of specific types of redox PTMs. This provides an overall measure of the changes to the redox proteome and is achieved by using a strong reducing agent such as DTT or TCEP (i.e., using the selective reduction and tagging strategy).
One caveat of note is that measurements of thiol oxidation will also include S-acylation (the reversible attachment of fatty acids through a thiol-ester group, mainly known as palmitoylation [41]), which can also be removed by DTT [42]. In this section, we review recent advances in redox proteome sample processing and applications of this strategy for broad identification and quantification of thiol oxidation or specific types of redox PTMs.

General principles and methods for thiol oxidation
The selective reduction and tagging strategy for profiling redox PTMs is built upon the reversible nature of many redox PTMs. Figure 2A illustrates the general workflow for this strategy. Specific redox PTMs can be analyzed by selectively choosing reducing agents for a given type of redox PTM (e.g., ascorbate for SNO and glutaredoxin for SSG). An overview of a biological system's redox state (i.e., thiol oxidation) can be attained using strong reducing agents for general reduction of oxidized thiols. Following reduction, various thiol-reactive tags have been developed to facilitate capture and enrichment of thiol-containing proteins or peptides (Figure 2A). In this section, we will mainly discuss methods for quantifying thiol oxidation as an overall measure of protein cysteine redox states.
The BST [20,21] was the first reported method utilizing this strategy and employs a thiol-reactive biotin tag. This concept was later (free) thiols at the same Cys sites. This allows stoichiometric measurement of thiol oxidation, which is defined as the occupancy of oxidized thiol at a given Cys residue. The OxICAT method and its modifications have been applied to study thiol oxidation in many biological systems, including bacteria [44], yeast [45], mammalian cell lines [46], and tissues [47]. In addition, alternative thiol-reactive tags and enrichment techniques have been developed to facilitate quantitative multiplexing. These include cysTMT [48] and iodoTMT [49] reagents, whereby labeled Cys-containing peptides are enriched using an anti-TMT affinity resin. Nevertheless, both the biotin-based and antibody-based affinity enrichment workflows often suffer from non-specific binding, which in turn limits the coverage of the redox proteome (see Besides affinity tag-based methods, resin-assisted capture (RAC) uses thiolated Sepharose to covalently capture thiol-containing proteins or peptides via thiol-disulfide exchange [31,50]. Pioneered by our CysPAT-TMT-IMAC enable stoichiometric measurement of thiol oxidation at site-specific level. Note that in the selective reduction step, the general reducing reagent (DTT or TCEP) can be replaced by PTM-specific reducing reagents (e.g., Grx cocktail for SSG, and ascorbate/Cu 2+ for SNO) to profile individual types of PTMs. group, this approach was developed as a general tool for multiplexed profiling of redox PTMs by coupling RAC with tandem mass tags (TMT) [31,[51][52][53]. In RAC-TMT ( Figure 2C), free thiols are first blocked with NEM. All reversible redox PTMs are then reduced to free thiols by DTT which are covalently bound to resin forming disulfide bond, thereby permitting intense washing steps to mitigate non-specific binding.
Importantly, RAC-TMT enables stoichiometric quantification by easily including one or several total thiol TMT channels. For these channels, samples are first reduced without NEM blocking. The occupancies of thiol oxidation at site-specific level are calculated by comparing TMT intensities between thiol oxidation channels and total thiol channels for individual Cys sites. RAC-TMT has been applied to cyanobacteria [54], metazoan cell lines [52,55,56], and mammalian tissues [19,38,57]. Besides TMT-based quantification, RAC has been employed for label-free quantification to characterize thiol oxidation in various organisms [58][59][60][61].
Recently, Huang et al. reported a cysteine-specific phosphonate adaptable tag (CysPAT) [62] consisting of a thiol-reactive group, a linker, and a phosphate group. This tag allows one to "trap" thiolcontaining peptides using IMAC (immobilized metal-ion affinity chromatography), which is routinely employed in phosphoproteomics.
Xiao, et al. evaluated CysPATs of different linker length (2C, 3C, 6C) and noted that the 6C-CysPAT (Figure 2A) performs the best [17]. In this work, the CysPAT tagging method was coupled with TMT labeling and IMAC enrichment ( Figure 2D). The CysPAT-TMT-IMAC workflow was applied for a site occupancy measurements of thiol oxidation in mouse tissues from young and old ages [17]. The authors observed significant age-related changes in the redox proteome of multiple tissues. Moreover, the high enrichment efficiency by IMAC, coupled with offline fractionation, led to a high coverage of the thiol redox proteome.

Methods for specific types of PTMs
The selective reduction and tagging strategy has also been applied to study specific types of redox PTMs including SNO and SSG. However, different types of PTMs have their own unique challenges due to differences in stability and abundance as well as the specificity of reducing reagents. In particular, most selective reducing reagents such as ascorbate for SNO and glutaredoxin (Grx) for SSG do not exclusively target a specific redox PTM [31].

S-nitrosylation (SNO)
SNO is difficult to characterize because of its labile nature. BST was the first technique developed to characterize SNO, typically using cells treated with a NO donor [20]. Since its conception, many approaches have adapted the reduction and tagging strategy. For instance, following selective SNO reduction, RAC coupled with iTRAQ has been used for profiling SNOs in samples treated with NO donors [51,63]. The SNOxICAT (S-nitrosothiol redox ICAT) method was recently reported as a modification of oxICAT, where ascorbate and Cu 2+ were used to selectively reduce SNOs, instead of the universal reducing reagent (TCEP) in oxICAT [64]. SNOxICAT was applied to identify potential SNO-modified Cys-sites in myocardium with NO 2 exposure in conjunction with ischemia (insufficient blood supply), which resulted in significant protein SNO modifications. The SNO site occupancy for each Cys residue was calculated in a similar manner to OxICAT. Another recent method, Cys-BOOST, was developed using the clickable thiol-reactive tag IAA-alkyne following ascorbate reduction of SNOs [65]. Subsequently, a cleavable Dde biotin-azide linker was introduced to enrich the tagged peptides prior to LC-MS/MS analyses. Using this Cys-BOOST approach, cells or cell extracts were treated with NO donors such as S-Nitrosoglutathione (GSNO) or cellpermeable NO donor S-Nitroso-N-Acetyl-D,L-Penicillamine (SNAP) to boost the number of SNO identifications [65]. One major limitation of selective reduction and tagging based strategies that use ascorbate is the lack of specificity given that disulfides and mixed disulfide bonds can also be reduced [66]. This limitation leads to ambiguity about whether the observed SNO sites are indeed modified in situ or false positives.

S-glutathionylation (SSG)
The selective reduction and tagging strategy can also be adapted for SSG profiling using the aforementioned workflows (BST, RAC-TMT, and OxICAT). To specifically reverse SSG back to SH, Grx enzymes are used in conjunction with GSH, NAPDH, and glutathione reductase (GR) [31,67]. The specificity of such workflows is improved by employing Grx mutants (e.g., E. coli Grx1 C14S and Grx3 C14S, C65Y ), where the active sites responsible for disulfide reduction are mutated, thus resulting an increased specificity for reducing protein-SSG [68,69]. The utility of the mutant enzymes for selective reduction of protein-SSG has been well demonstrated [67,70]. The RAC-TMT method is amenable to quantitative profiling of SSG-modified Cys sites. It had been applied to quantify SSG, total thiol oxidation, and total thiol in one single experiment [52]. Under basal conditions, the average SSG and total oxidation occupancies were 4.0% and 11.9%, respectively, in macrophages cells.
This workflow had also been applied to several other studies, including muscle and lung tissues [19,38,55].
A modified OxICAT method to quantify SSG at specific Cys residues, termed GluICAT, was also reported by replacing TCEP with Grx1 C14S [ 71]. In this study, SSG site occupancy was determined using heavy-to-light ratios for individual Cys sites. This method was successfully applied to profile the SSG proteome of fibroblasts from healthy subjects and patients with Leber's hereditary optic neuropathy [72].

S-persulfidation (SSH)
Compared to free thiols, SSH has a lower pKa and enhanced nucleophilicity [73]. It reacts with electrophilic thiol-reactive tags (and alkylating reagents) to form a persulfide adduct (Protein-SS-Alk). A common strategy to profile protein-SSH is to tag and capture both SH and SSH in the first step followed by selective reduction and release of SSH-modified protein/peptides via disulfide bond cleavage [74][75][76][77]. One successful method is the biotin thiol assay (BTA), whereby biotinylated maleimide is used to tag both SH and SSH followed by pull-down of tagged proteins using streptavidin-agarose resin [75].
SSH-modified proteins were then collected by elution with DTT and finally analyzed by LC-MS/MS. The method was further modified by performing streptavidin resin-based enrichment after protein digestion, enabling peptide-level enrichment. After TCEP elution of formerly SSH-modified peptides, the resulting nascent free thiols can be labeled by iodoTMT for multiplexed quantification [78]. Recently, concerns about SSH detection methods were raised because common alkylating reagents including NEM and IAM can rapidly convert protein-SS-Alk (or polysulfide adducts) to protein-S-Alk by transferring sulfur to ambient nucleophiles [27,79,80], which likely impacts detection and overall interpretation of SSH data.

Enhancing the coverage of the redox proteome
One of the general goals of PTM profiling is to achieve a deep coverage of the functional proteome. For redox proteomics, coverage can be defined as the number of unique cysteine sites and cysteinecontaining proteins identified and quantified in each experiment. A major reason for generally low coverage of the redox proteome can be attributed to the natural low abundance of most redox PTMs. To enhance coverage, the first and arguably most important consideration is how to selectively enrich the PTM of interest. Highly specific enrichment approaches can greatly reduce sample complexity by removing non-cysteine-containing peptides and peptides not bearing the PTM of interest, thus allowing more sensitive detection of Cys-containing peptides by MS. For instance, we reported that the RAC approach can enrich Cys-containing peptides with >95% specificity [31,57]. Another commonly applied strategy is to introduce a fractionation step prior to final LC-MS/MS analyses (a.k.a. offline fractionation). Such fractionation can be performed either prior to or after enrichment. For example, in the workflow reported by Xiao et al., IMAC-enriched peptides were separated into six fractions using a commercially available high-pH reversed-phase peptide fractionation kit [17]. The fractionation of one peptide sample into six fractions for final LC-MS/MS resulted in greater coverage of the redox proteome (∼5000 and ∼15,000 unique Cys sites in mouse muscle and brain tissues, respectively). Our lab has also reported the incorporation of a microscale high-pH reversed phase LC-fractionation for enriched Cys-containing peptides, which resulted in considerable improvement of the redox proteome coverage [53].
When applied to mouse muscle tissue, ∼18,000 unique Cys sites were quantifiable from 24 fractions compared to ∼3800 Cys sites in unfractionated samples. Another benefit of fractionation is to alleviate the ratio compression issue of TMT-based quantification, which is a welldocumented phenomena stemming from co-eluting peptides entering MS/MS [81].
Additional strategies have proven to be promising for improving coverage including the use of multiple enzymes for protein digestion [82,83]. Moreover, advances in MS instrumentation will continue to provide an improved proteome coverage. For instance, the coupling of field asymmetric waveform ion mobility spectrometry (FAIMS), a gasphase separation technique, has resulted in higher proteome coverage in shotgun proteomics [84]. Yan et al. recently also reported the optimization of the compensation voltages of FAIMS to achieve enhanced redox proteome coverage [83].

CHEMOSELECTIVE PROBE-BASED METHODS
Besides the selective reduction and tagging strategy, there has been appreciable interest in advancing chemical biology tools to target specific redox PTMs. Chemoselective probes typically consist of a selective tagging group and a reporter group for affinity-based enrichment and/or fluorescence-based detection. Isotope-coded chemoselective probes are also available for quantifying redox PTMs using MS. In gen-eral, chemoselective probes react with specific redox PTMs via unique chemistries. Besides specificity and stability, chemical probes should ideally be membrane permeable, so that they can be applied to in situ cell culture profiling. Figure 3A illustrates the general workflow for the chemoselective probe-based strategy, which typically consists of chemical probe-based tagging, downstream alkylation to block free thiols, enrichment (facilitated either by biotin or an affinity tag introduced via click chemistry), and LC-MS/MS analysis. In the following sections, we summarize methods targeting specific types of redox PTMs using chemoselective probes.

S-sulfenylation (SOH)
SOH is often regarded as the most important PTM in redox signaling [1,85]. However, SOH is perhaps also the most difficult to detect due to its labile and transient nature [39]. For this reason, developing chemical probes for SOH has drawn significant attention over the last decade with efforts focusing on two important aspects: (1) the extremely short halftime of SOH requires small molecule probes with high reactivity [39], and (2)  cyclic C-nucleophiles for their reactivity toward SOH [87] and further developed several alkyne-tagged probes for protein SOH labeling [88].
Among them, a benzothiazine-based BTD probe ( Figure 3B) had the highest SOH reactivity (2 orders of magnitude higher than DYn-2) and has been successfully applied to detect SOH sites in a colon cancer cell line [88]. More recently, Shi et al. harnessed the reactivity of a new class of highly nucleophilic and selective triphenylphosphonium (TPP) ylides, known as Wittig reagents, for enabling SOH detection under biocompatible reaction conditions [25]. The TPP-based WYneN probe ( Figure 3B) enabled the profiling of SOH sites in A549 cells [25]. An interesting observation is that the TPP-cleaved SOH was identical to the product of alkyne-tagged iodoacetamide probe (IPM) with free thiols, enabling stoichiometric estimation of SOH at Cys site level.
In this workflow, cells were first treated with 13
Phenylmercury-based reagents, including organomercury-conjugated resin (MRC) and phenylmercury-polyethyleneglycol-biotin (mPEGb) ( Figure 3B) have been developed based on the reactivity of phenylmercury to SNO, which forms a covalent thiol-mercury bond (S-Hg) [90]. The direct covalent capture of SNO alleviates the specificity limitations observed for selective reduction and tagging approaches [91]. Using MRC, Raju et al. observed decreased S-nitrosylation of proteins involved in glutamate uptake and metabolism in mice lacking the neuronal isoform of nitric oxide synthase [92]. Wang et al. reported another tagging concept to generate stable sulfenamide analogs with phosphine (PH 3 ) esters [93]. The reaction of triphenylphosphine esters with protein-SNO occurs without detectable interference from protein disulfides (e.g., SSG and S-S) or SOH [94]. Since then, several phosphine-based chemical probes linked to fluorophores, biotin derivatives, and clickable alkynes have emerged for detecting SNO via gel-and MS-based proteomics approaches [95][96][97][98]. In the SNOTRAP method, triphenylphosphine thioester ( Figure 3B) is linked to biotin through a polyethyleneglycol (PEG) group [96]. This probe enabled direct tagging of ∼300 endogenous SNO sites in mouse brain [96]. More recently, Clements et al. reported a clickable alkyne probe called "PBZyn," which contains a o-phosphino-benzoyl group warhead that conjugates to small molecule SNOs (e.g., GSNO, SNAP) and protein-SNOs to form a disulfide linker [98]. The PBZyn probe has a potential versatile utility in imaging, protein blotting, and proteomics; however, its utility in SNO profiling by MS has not been demonstrated [98].

S-persulfidation (SSH)
Due to similarity in SH and SSH reactivities, it is challenging to specifically tag SSH using chemical probes. Zhang et al. reported the first selective tag-switch method to target protein-SSH [99]. In this approach, SH and SSH are first blocked by a SH-blocking reagent (e.g., methylsulfonyl benzothiazole, MSBT, Figure 3B), and then a nucleophilic tag (biotin-linked cyanoacetate, Figure 3B) was introduced to attack the disulfide of blocked SSH, thus labeling only the persulfide adducts for further enrichment and detection [99].
More recently, Zivanovic et al. reported an improved tag-switch method where SH, SSH, and SOH are all blocked with 4-chloro-7nitrobenzofuranzan (NBF-Cl, Figure 3B), and then persulfide adducts can be switched to a dimedone biotin tag [100]. The dimedone-based biotinylated probe (DCP-Bio1, Figure 3B) was used to selectively react with the persulfide adduct -SS-NBF and converted it to S-DCP-Bio1. Thus far, this method only reported the identification of SSH-modified protein targets but holds potential for site-specific SSH quantification.  Figure 3B), which consists of a C-nitroso reactive group and a biotin handle [101]. Although electrophilic nitrogen species also react with free thiols and SOH, the resulting product (sulfonamide) from SO 2 H reaction with NO-Bio is stable, whereas the other modifications are readily reduced by DTT [101]. Biotin-GSNO was used to label SO 2 H as well, but there is a concern that a possible byproduct nitroxyl (HNO) can convert thiols to sulfinamides and then sulfinic acids [102,103]. New probes were recently developed by screening the reactivity of electron-deficient diazonium salts and diazenes (both new classes of electrophilic nitrogen species) with yeast thiol peroxidase Gpx3
The authors found that di-t-butyl azodicarboxylate (DBAD) completely converted Gpx3-SO 2 H. Akter et al. then synthesized DBAD-based probes with clickable alkyne (DiaAlk, Figure 3B) and fluorescein (DiaFluo) functional groups, which showed good specificity and sensitivity for detection of exposed and barried SO 2 H within human DJ-1 and E. coli rhodanase PspE proteins [8]. By coupling click chemistry and affinity enrichments, the chemoproteomic workflow effectively identified >200 SO 2 H sites.

THIOL REACTIVITY-BASED APPROACHES
As an alternative to directly targeting redox PTMs with chemical probes, thiol-reactive probes are often employed for measuring protein thiol abundances at the proteome level, which can serve as an indirect means for measuring thiol reactivity, pinpointing redox changes, and identifying highly reactive Cys sites in responses to treatment. One such approach is called stable isotope cysteine labeling with iodoacetamide (SICyLIA), where the levels of protein free thiols are quantified as an indirect readout of cysteine oxidation by LC-MS/MS without the need for enrichment [104]. With this approach, reduced protein thiols from two experimental conditions are first differentially labeled with either stable light or heavy isotope-coded IAM. Samples are then mixed, and oxidized thiols are reduced and subsequently blocked with NEM. The mixed proteomes are then digested, followed by offline fractionation and LC-MS/MS. Although no enrichment is used, this approach targets the more abundant free thiols and uses extensive offline fractionation to achieve a good coverage of the redox proteome. However, its low multiplexing capacity limits the application of this approach for a large cohort study.
Weerapana et al. first reported the concept of proteomewide quantitative thiol reactivity profiling to predict functional cysteines using isoTOP-ABPP (isotopic Tandem Orthogonal Proteolysis-Activity-Based Protein Profiling) [105]. In this work, a thiol reactive iodoacetamide (IA)-alkyne probe was used to react with free thiols followed by click chemistry and avidin enrichment for MS based redox proteome profiling [105]. Since its inception, chemoproteomic workflows utilizing thiol-reactive probes with a clickable alkyne functional group and cleavable azido biotin reagents have been developed to improve the probe reactivity and accessibility to cysteine sites in natively folded proteins while expanding proteome profiling applications [106,107]. Kuljanin  workflow was developed to simultaneously map protein-SSH and -SH [109]. In this workflow, both protein-SH and -SSH were tagged with IPM at pH 5, followed by a second alkylation with iodoacetamide and trypsin digestion. All IPM-tagged peptides were tagged with light or heavy isotope-coded, UV-cleavable azido-biotin (Az-UV-biotin), then combined (1:1 mixed) and subjected to affinity enrichment. Finally, peptides were released by UV exposure prior to LC-MS analysis. It should be noted that the pK a of SSH is 4.3 while the pK a of SH is 8.29.
Thus, the low pH alkylation only allowed labeling of hyperreactive -SH and -SSH with low pK a values. The low-pH QTRP workflow was successfully applied to cell lines and tissues, yielding 988 unique SSH sites in HEK293 cell treated with NaHS [109].  [110].

DIRECT DETECTION-BASED APPROACHES
Recently, we explored the utility of a mild alkylation reagent, β-(4-hydroxyphenyl) ethyl iodoacetamide (HPE-IAM), for preserving endogenous redox PTMs and applied a global deep proteome profiling workflow (TMT coupled with offline high pH LC fractionation) to identify multiple types of redox PTMs in mouse pancreatic β-cells [27].
Initially, free thiols, persulfides, and polysulfides are blocked by HPE-IAM, while other redox PTMs were preserved. Interestingly, this study revealed a complex landscape of redox PTMs, with SSH, S-S, and SO 3 H as the most prevalent modifications and many Cys sites bearing ≥2 types of modifications. It is likely that advanced deep proteome pro-filing workflows can be applied to directly detect unstable redox PTMs such as SOH and SNO once the redox PTMs are stabilized either in situ or in vitro using some of the chemoselective probes described above.
One example is the direct detection of SSH following stabilization with a thiol reactive chemical probe [109]. It is also foreseeable that low coverage, which currently stymies application of direct detection approaches, can be mitigated by innovative fractionation methods.

HIGHLIGHTED APPLICATIONS OF REDOX PROTEOMICS
Proteomics analyses provide large, condition-specific datasets that are prime resources for discovery and hypothesis-driven research. learning and molecular dynamics to elucidate relationships between cysteine reactivity and protein structure [58,111], application of redox proteomics datasets for uncovering evolutionary insights [112], and analyses that couple structural proteomics to redox proteomics to investigate extracellular matrix proteins (e.g., the adhesome complex) [18].
OxiMouse is a compendium of ten tissue-specific redox proteomes for which percentage reversible cysteine modifications were quantified in young and old mice [17]. Xiao et al. use this expansive dataset to illustrate redox signaling networks in a tissue-specific fashion in the context of aging. Mapping redox data onto proteome networks using BioPlex 2.0 [113], they pinpointed redox networks that were differentially regulated. Going one step further, they used the DisGeNET [114] platform to coalesce their identified age-dependent redox networks with established disease-linked proteins. These interactomes facilitate deeper investigations into regulatory networks, which they demonstrated for hexokinase and tRNA synthetases using various chemical and siRNA inhibitors to modulate redox states and expression levels.
Overall, Xiao et al. illustrate how redox proteomics and network modeling can be leveraged to advance our understanding of physiological phenomena.
The systematic study of signaling networks is confounded by a paucity of information for interacting networks. Su et al. evaluated the potential crosstalk among phosphorylation and redox signaling pathways using a model of oxidative stress in adipocytes [115]. A multi-omics approach in combination with insulin and drug treatments was employed to characterize dynamic changes to protein expression and PTMs (phosphorylation and thiol oxidation). For insulin signaling, they hypothesized that redox modifications to upstream kinases (Akt, mTOR, and AMPK) cause global phosphorylation changes. Akt transduces growth signals after binding PIP3 at the plasma membrane and subsequent phosphorylation by PDPK1 [116]. The mechanism of Akt redox regulation was investigated using site-directed mutagenesis,  [115,123]. The former was shown to modulate PIP3 binding for plasma membrane recruitment and is essential for Akt activation. The latter has been implicated in kinase inhibition [124] perhaps by protecting Thr309 from phosphatases. In the excerpts, dotted yellow lines were added to highlight the disulfide bonds of the yellow cysteine residues. In the right panel, a phospho-functional group (red ball-and-stick) was added for Thr309. molecular dynamics, modeling, and fluorescence microscopy. With this set of tools, they were able to make two substantial conclusions: that insulin exposure increases disulfide bond formation between residues C60 and C77 of the Akt PH domain, thereby increasing affinity for PIP3 Interestingly, Behring et al. also observed oxidation of Cys297, but upon treatment with epidermal growth factor (EGF) [58]. As with the article discussed above, they provide a prime example of using systemslevel data to drive deeper mechanistic investigations. In their study, though, they focus on spatiotemporal dynamics of EGF-dependent redox PTMs in epidermal cancer cells. Like Su et al., they showcase examples of how phosphorylation and redox PTMs interact. EGF stimulates phosphorylation of mitogen-activated protein kinase 1 (ERK2), thereby altering its structure and making Cys65 solvent-exposed. The functional consequence of this exposure (or lack thereof) is unclear; however, increased oxidation (presumably as a result of exposure) may affect ATP binding and/or ERK inhibition [117].
Another recent study by Cuervo et al. reported the Cys sites for redox-dependent inhibition of human STING [118], which is known for its role as a central adaptor of the interferon (IFN) pathway in innate immune defense when the system detects DNA from pathogen.
In this study, a maleimide-based BST approach coupled with MS was applied to identify ∼2700 Cys sites, and reversible oxidation of Cys148 and Cys206 in STING was observed. Compared to Cys148 (which was oxidized in all conditions), oxidation of Cys206 required oxidative stress or stimulation with 2′3'-cGAMP docking to the protein.
Further structural modeling and functional studies (e.g., protein localization, activity, etc.) using mutagenesis (STING-C206S) indicated that oxidation of Cys206 diminished protein activity and prevented Ser366 hyperphosphorylation, suggesting an inhibitory role of Cys206 modification to prevent protein hyperactivation via inhibition of Ser366 phosphorylation.
Despite differences in approach, there are some common themes in these papers that yield additional routes for discovery. Firstly, all the papers highlighted in this section used the general thiol oxidation approach, which does not specify particular types of redox PTMs.
Thus, to draw conclusions like those represented in Figure 4, methods that are complementary to redox proteomics approaches must be employed to learn more about specific PTMs present at highly modified residues. Another prominent theme includes regulatory interactions between phospho-functional groups and cysteine thiols that have functional consequences for signaling pathways and their respective targets. Conceptually, one could use known and predicted (with AlphaFold [119] for example) protein structures to facilitate a deeper dive into the mechanistic interplay between phospho-and redox PTMs, perhaps starting with certain regulatory proteins implicated in signal transduction, protein translation, and ROS metabolism, which were highlighted in the aforementioned papers. Continuing the example introduced in Figure 4, the overall goal is to connect functional and structural changes to redox PTMs at a systems-level, which includes considerations about other PTMs. It is expected that these and other articles implementing expression and redox proteomics, modeling, and reductionist approaches will advance our comprehension of complex biological systems in the years to come.

OUTLOOK AND CONCLUSION
All aspects of life are inextricably linked to redox chemistry. MSbased proteomics approaches have made impressive strides in the field of redox proteomics for the identification and quantification of redox PTMs in diverse biological systems. The development of various tools to tackle the unique chemistries and challenges for each redox PTM type has enabled researchers to appreciate the diverse and dynamic nature of the redox proteome. Redox proteomics has become an indispensable field with unmeasurable potential for advancing the biomedical sciences as well as underexplored areas like synthetic biology (to engineer organisms for sustainable production of fuels, chemicals, and materials).
Most current redox proteomics workflows are labor intensive and low-throughput, preventing their broad applications to large-scale studies. Recent developments in sample processing techniques such as SP3 (single-pot, solid-phase-enhanced sample-preparation) [120] and S-Trap [121], which are amenable to automation [121], may greatly enhance throughput and flexibility of sample processing workflows. In addition to sample preparation, advances in MS instrumentation, data processing algorithms, and bioinformatics tools will be a salient part of the overall formula for advancing redox proteomics, even though they weren't discussed herein. Recent data independent approaches (DIA) used in redox proteomics (for example, oxSWATH [122]) are worth mentioning as they decrease instrument time compared to data dependent approaches and can improve sensitivity.
In the coming years, impactful technological innovations will address persistent challenges that arise due to the complexity of redox PTMs, their transient and unstable nature, and their low abundance levels. These innovations will improve specificity and authenticity in detection (especially for SNO and SOH), overall sensitivity, and proteome coverage to widen the landscape of analyzable redox PTMs.
We also expect advances to improve the sensitivity and throughput of workflows that can process quantity-limited samples (even at the single-cell level)-pushing redox proteomics one step closer to large-scale applications in clinical research.
In addition, functional elucidation of the biological consequences of redox PTMs will be required to fully understand the roles of redox PTMs. Advances in the development of chemoselective probes with high reactivity and specificity will facilitate the in situ capture of even transient and rare PTMs important for redox signaling. Integration with other omics such as phosphoproteomics and metabolomics in addition to computational biology approaches is promising to delineate the crosstalk among PTM types and their implications on cell phenotypes.
With effective labeling strategies and a "universal" sample processing workflow, one can ideally analyze multiple types of PTMs with diverse probes in a quantitative and dynamic fashion to investigate interplay among PTMs at the systems level.