Regulation of Protein Function and Signaling by Reversible Cysteine S-Nitrosylation*

NO is a versatile free radical that mediates numerous biological functions within every major organ system. A molecular pathway by which NO accomplishes functional diversity is the selective modification of protein cysteine residues to form S-nitrosocysteine. This post-translational modification, S-nitrosylation, impacts protein function, stability, and location. Despite considerable advances with individual proteins, the in vivo biological chemistry, the structural elements that govern the selective S-nitrosylation of cysteine residues, and the potential overlap with other redox modifications are unknown. In this minireview, we explore the functional features of S-nitrosylation at the proteome level and the structural diversity of endogenously modified residues, and we discuss the potential overlap and complementation that may exist with other cysteine modifications.

NO is a versatile free radical that mediates numerous biological functions within every major organ system. A molecular pathway by which NO accomplishes functional diversity is the selective modification of protein cysteine residues to form S-nitrosocysteine. This post-translational modification, S-nitrosylation, impacts protein function, stability, and location. Despite considerable advances with individual proteins, the in vivo biological chemistry, the structural elements that govern the selective S-nitrosylation of cysteine residues, and the potential overlap with other redox modifications are unknown. In this minireview, we explore the functional features of S-nitrosylation at the proteome level and the structural diversity of endogenously modified residues, and we discuss the potential overlap and complementation that may exist with other cysteine modifications.
Endogenous production of NO is accomplished principally through the enzymatic action of NOS isoforms (1)(2)(3)(4)(5)(6). These enzymes (eNOS, iNOS, and nNOS) are expressed throughout the body and orchestrate the production of biologically active NO (4 -6). Transcriptional regulation, availability of calcium and various cofactors, and post-translational modifications regulate NOS levels, localization, and functional dimerization, resulting in controlled and timely production of NO (4 -6). Binding of NO to the heme of soluble guanylate cyclase activates this enzyme, leading to the production of cGMP (1)(2)(3). The ensuing activation of cGMP-dependent kinases transduces a multitude of signaling events through protein phosphorylation (3). This canonical signaling cascade is regulated at several steps, including controlling NO production, reversible binding to soluble guanylate cyclase, degradation of cGMP by phosphodiesterases, and dephosphorylation of downstream targets by phosphatases (3). Similar to other signaling pathways, noncanonical NO signaling has been also documented and is achieved principally by the covalent modification of protein cysteine residues to form S-nitrosocysteine (7)(8)(9).
Cysteine residues in proteins can be broadly classified into four functional groups: structural, metal binding, catalytic, and regulatory (10 -13). By forming disulfide bonds, cysteine residues serve a principal structural function during protein folding (14,15). Enzymatically controlled modifications of cysteine residues, such as S-acylation (primarily S-palmitoylation), regulate protein location in membranes, function, and stability (16,17). Cysteine residues coordinate metal binding, which impacts protein activity and provides structural stability (13). Cysteine residues are also used for catalysis in diverse protein families, such as oxidoreductases, proteases, and acyltransferases (10 -13). Last but not least, reversible oxidation states of cysteine residues, such as sulfenic acid, as well covalent adductions, such as S-nitrosylation, S-glutathionylation (addition of glutathione, forming a mixed disulfide), and S-sulfhydrylation (addition of hydrogen sulfide), facilitate redox-dependent signaling and regulation of protein function (18 -28). Overall, the diverse chemical reactivity of the sulfur permits unique modifications of cysteine residues that significantly expand the biological chemistry by which homeostatic regulation of redox sensing, signaling, protein function, stability, and trafficking is achieved and regulated.
This minireview is focused on the biological roles of protein S-nitrosocysteine in health and disease, as well as the potential overlap and complementation that may exist with other cysteine modifications. Before we embark on these discussions, a few disclosures are required. (i) The formation of S-nitrosocysteine requires the addition or transfer of a nitroso group to the reduced cysteine. The term that defines this chemical reaction is S-nitrosation. The term that describes the biological function of this modification is S-nitrosylation, which, although not chemically accurate, is consistent with the commonly used "-ylation" suffix for post-translational modifications. We will use the term S-nitrosylation throughout. (ii) The biological chemistry that forms S-nitrosocysteine in vivo remains uncertain. Although the source of NO is primarily enzymatic, the formation of S-nitrosocysteine includes cysteine oxidation; metal catalysis; exchange reactions with low molecular weight thiols, such as S-nitrosoglutathione (GSNO) 2 ; or transnitrosation reactions between proteins, principally by S-nitrosothioredoxin (29,30). The mechanisms of protein S-nitrosylation mediated via transfer of the nitroso group (GSNO and S-nitrosothioredoxin) designate regulated processes. Comprehensive reviews on the chemistry of S-nitrosocysteine formation have been recently published (31)(32)(33)(34). The presence of GSNO reductase, an enzyme that regulates GSNO levels, indirectly regulates protein S-nitrosylation (35,36). Moreover, evidence indicates that thioredoxin facilitates not only protein-to-protein transfer of the nitroso group but also its removal, providing a complete cycle for protein-mediated regulation of S-nitrosylation (37). Overall, existing data indicate multiple mechanisms for the formation and removal of protein S-nitrosocysteine in vivo, some of which require protein-protein interactions.

S-Nitrosylation in Health and Disease
A literature review was performed to assemble a list of endogenously S-nitrosylated proteins. The list includes cellular models in which protein S-nitrosylation is derived from NOS activity. Studies that exclusively used NO donors or S-nitrosating agents were excluded. The list does not include our recent work in mouse tissues, which will be discussed separately. The data are organized to include the following: protein name, protein accession number, species, organ or cell type of origin, site of S-nitrosylation, and protein location and function (when available) (supplemental Table 1).
The majority of the proteins were identified by the biotin switch method (9). The original biotin switch method and several variations thereafter detect S-nitrosylated proteins by performing the following five steps: 1) blocking of reduced cysteine residues; 2) reductive elimination of NO from S-nitrosocysteine, generating a reduced cysteine; 3) labeling of the resulting reduced cysteine with biotin; 4) enrichment for labeled proteins through biotin-avidin interaction; and 5) protein-or site-specific identification by Western blotting or mass spectrometry. Proteins and sites of modification detected by alternative methods, such as antibody-mediated enrichment, followed by secondary detection methods are indicated as well. For comprehensive appraisal of the methodologies for detection and site-specific identification of S-nitrosocysteine in biological samples, please refer to recent reviews (38 -41).
The literature review identified a total of 233 S-nitrosylated proteins, 171 of which were identified under physiological conditions, 28 under various pathological conditions, and another 34 that may be ascribed to either physiological or pathological conditions. Ontological analysis of this assembled S-nitrosocysteine proteome revealed a significant overrepresentation of mitochondrial proteins within both the intermembrane space and the inner mitochondrial membrane. In terms of molecular function, a significant proportion of proteins are involved in the generation of precursor metabolites and energy, including the electron transport chain, glycolysis, glucose metabolism, and oxidative phosphorylation. The pathological conditions are often associated with overproduction of NO, a condition called nitrosative stress, and result in inappropriate S-nitrosylation of proteins, leading to dysfunction and pathological phenotypes.
These findings were significantly expanded by the site-specific identification of S-nitrosylated proteins in different organs of the wild-type mouse. Using a biochemical method for the enrichment of S-nitrosylated proteins and peptides, along with negative controls and the use high mass accuracy tandem mass spectrometry, we reported 1011 S-nitrosocysteine residues in 647 proteins in wild-type mouse brain, heart, kidney, liver, lung, and thymus (42). The chemical enrichment is based on the principles of the Saville reaction (43), in which the nitroso group of S-nitrosocysteine is displaced by phenylmercury, and a new covalent bond between the cysteine sulfur and organic mercury is formed. Bioinformatic analysis of this newly acquired data set indicated the following. (i) Two-thirds of the proteins identified are shared among the different organs; (ii) mitochondrial proteins are significantly enriched compared with the entire mouse proteome; (iii) the majority of the shared mitochondrial proteins are functionally involved in metabolic pathways; (iv) proteins with catalytic and metal-binding cysteine residues are present, but not cysteine residues participating in structural disulfide bonds; and (v) nearly half of the identified S-nitrosylated proteins require the expression of eNOS. Analysis of the same organs from eNOS null mice revealed an organ-specific dependence on the expression of eNOS (ranging from 47 to 87%) and the enzymatic production of NO for the formation of protein S-nitrosylation. Collectively, the data indicated that protein S-nitrosylation may provide a molecular and biochemical mechanism for the previously recognized regulation of these metabolic processes by enzymatically produced NO (44 -47). The data may also serve as a resource to investigate the principles that guide the apparent selective targeting of particular cysteine residues for S-nitrosylation in cellular proteomes.

S-Nitrosylation
Although significant progress has been made in recognizing the biological roles of protein S-nitrosylation in health and disease, several critical questions regarding the selectivity and abundance of S-nitrosylation remain. Concerning selectivity, what defines and determines the endogenous selectivity of S-nitrosocysteine formation in proteins? Attempts to provide answers to this question have been made by analyzing the biochemical, biophysical, and structural properties of S-nitrosocysteine residues. Although these previous attempts did not provide explicit biochemical or structural requirements (48,49), the presence of acid-base residues in the vicinity of modified cysteine residues may offer some selectivity (31). Alternatively, the presence of charged residues located distally to the modified cysteine residues could facilitate catalysis or provide sites for protein-protein interfaces to direct site-specific S-nitrosylation (48). We attempted a similar structural analysis using only in vivo identified S-nitrosylated proteins (50). Using structures of proteins identified in wild-type mouse liver, we compared the biochemical and biophysical properties of S-nitrosylated cysteine residues with those of unmodified cysteine residues on the same proteins. This analysis revealed some differences that may distinguish cysteine residues targeted for S-nitrosylation. Specifically, sites of S-nitrosylation were overrepresented in ␣-helices, were located in larger surface-accessible areas, and were surrounded by charged amino acids within a 6-Å distance. However, not every S-nitrosocysteine residue conforms to these parameters (50). This was apparent when we performed the same analysis using the data from all mouse organs (42). This analysis reinforced the proposition that groups of cysteine residues that share similar biochemical (pK a and solvent exposure) or structural (location in helices and proximal to positively charged residues) properties constitute diverse molecular targets that can accommodate different pathways for S-nitrosocysteine formation. In the mouse S-nitrosocysteine proteome, we have also identified cysteine residues located near metal centers that can catalyze site-specific modification (49,51,52). We propose that the multiplicity in the biological chemistry of S-nitrosocysteine formation is matched by the structural diversity of protein cysteine residues, which is principally responsible for the selectivity of this posttranslational modification. The existence of these diverse protein microenvironments guides site-specific S-nitrosylation and may also explain the apparent absence of significant overlap with other cysteine modifications that will be discussed below. However, this proposal deviates from other know mechanisms of post-translational modification, such as phosphorylation. The specificity of phosphorylation is derived primarily by the selectivity of kinases for protein targets and the presence of certain motifs in the client proteins. Diversity is achieved by the presence of different kinase families and by regulation of kinase activity. A parallel to the function of kinases in phosphorylation is a protein that can transfer the nitroso moiety to a target cysteine residue (S-nitrosylase activity). Thioredoxin and GAPDH are two proteins with documented protein-assisted transnitrosylation or S-nitrosylase activity (29,53). Protein-catalyzed transnitrosylation reactions could have an inherent selectivity due to the required protein-protein interactions between the S-nitrosylase and the target protein. Half of the 93 proteins with resolved three-dimensional structures in the mouse S-nitrosoproteome have S-nitrosylated cysteine residues localized in solvent-exposed areas that could be targeted for protein-assisted S-nitrosation. Additional research delineating the molecular targets of protein-assisted trans-S-nitrosylation would vastly improve our understanding of the selectivity of in vivo S-nitrosylation and may provide firm criteria that will distinguish or predict S-nitrosylated cysteine residues in proteins.
As more knowledge is gained for S-nitrosylation as a functional post-translational modification, a common question often asked at different forums relates to the stoichiometry and occupancy of this modification. Specifically, what percentage of a protein needs to be S-nitrosylated to elicit a functional change or initiate a signaling event? Several analytical approaches have been generated to measure the total levels of protein S-nitrosocysteine, and typical levels are in the nanomolar range (54). However, the levels of S-nitrosylation for specific proteins have not always been quantified. For example, thioredoxin exhibits substantial influence on the levels of S-nitrosylation by executing both transnitrosylation and denitrosylation reactions (29,37). Basal S-nitrosylation of thioredoxin is involved in maintaining cellular redox status and exhibits anti-apoptotic effects (30). However, the proportion of modified thioredoxin is presently unknown.
Recently, cysteine-reactive tandem mass tags have been used to quantify S-nitrosylation occupancy in a model of ischemic reperfusion injury (55). Under physiological conditions, 1-10% of any given protein was occupied by S-nitrosocysteine at a specific site (55). These data indicate that one in every 10 (maximal) or one in every 100 (minimal) protein molecules are occupied at a given cysteine residue (55). These levels are consistent with other post-translational modifications of cysteine residues, such as oxidation, which was shown to vary from 14 to 22% in cellular model systems (12).
It is possible, however, that the levels of S-nitrosylation are underestimated. Factors such as (i) denitrosylation (37), (ii) compartmentalization of the S-nitrosylated proteins (53), (iii) protein stability (56), and (iv) detection methodologies will contribute to the reported relative low abundance of S-nitrosylated proteins in vivo. We reported that 25% of very long chain acyl-CoA dehydrogenase molecules were S-nitrosylated on a single cysteine residue in vivo under physiological conditions (42). S-Nitrosylation lowered the K m by nearly 5-fold and improved the catalytic efficiency (K m /K cat ) of very long chain acyl-CoA dehydrogenase by 29-fold, enabling the S-nitrosylated protein to effectively metabolize most of the substrate in mouse liver (42).
Another aspect of S-nitrosylation that has garnered little attention is the potential for proteins to be functionally regulated by poly-S-nitrosylation (57). The ryanodine receptor may represent a prototypical example of functional regulation by poly-S-nitrosylation (58). This receptor was shown to be progressively activated based on the number of modified sites. It is also possible that GAPDH can be S-nitrosylated at least at two different sites and that the site of modification could influence its trafficking, association with other proteins, and its ability to act as an S-nitrosylase. For example, recent studies indicated that GAPDH engages RPL13a in a chaperone-like manner to prevent ubiquitination and proteasomal degradation of newly synthesized RPL13a (59). The GAPDH-RPL13a pair regulates the interferon-␥-activated inhibitor of the translation complex that controls the translation of proteins participating in inflammatory processes. This functional regulation was lost upon S-nitrosylation of GAPDH at Cys 247 (human protein; Cys 245 for the mouse and rat sequences) (59). The biological relevance of poly-S-nitrosylation or differential modification of cysteine residues requires further investigations.

Overlap and Complementation of S-Nitrosylation and Other Redox-based Protein Cysteine Modifications
The knowledge and recognition that cysteine residues undergo several different redox-dependent modifications other than S-nitrosylation imply that caution must be exercised when ascribing biological roles to S-nitrosylation. One approach to explore the potential overlap among the different redox-dependent cysteine modifications is to compare the sites and proteins that are modified in biological systems. This approach was used by the pioneering study of Leonard et al. (60). In this study, endogenously identified sites of sulfenic acid (S-sulfenylation) were compared with published data of protein disulfides and S-glutathionylated and S-nitrosylated proteins. The comparison revealed an 11, 5, and 18% overlap between S-sulfenylated and disulfide, S-glutathionylated, and S-nitrosylated proteins, respectively (60). Herein, we expanded these findings by using the recently published S-nitrosoproteome and an expanded S-glutathionylated proteome assembled from an updated literature review (Fig. 1). The proteomes consists of 650 S-nitrosylated, 181 S-sulfenylated, and 118 S-glutathionylated proteins. Only five proteins are present in all three proteomes. A small fraction of the S-nitrosocysteine proteome (7 and 1%, respectively) is shared with S-sulfenylated and S-glutathionylated proteomes. Although these comparisons imply that the majority of the redox-dependent cysteine modifications target distinct and separate proteomes, we must consider several caveats. 1) The data were collected from proteomic studies in different tissues, cells, and species. 2) The different methodologies used to acquire these proteomes may have inherent biases. 3) The S-sulfenylated and S-glutathionylated proteomes are relatively small, and as such, they may under-represent the proteins modified.
Optimal comparisons should include rich proteomes of endogenously identified sites and proteins in the same organ or cell. To date, the best comparison we could provide is depicted in Table 1. In a comprehensive and quantitative manner, Cravatt and co-workers (11) reported the reactivity profile of cysteine residues for a small electrophile, alkene-tagged iodoacetamide, in cellular model systems and in wild-type mouse heart. Systematic analysis of the reactivity of the different cysteine residues toward iodoacetamide failed to reveal consensus motifs and indicated a preference for N-terminal cysteine residues in ␣-helices. These conclusions were similar to our analy-sis of the sites of S-nitrosylation, which prompted us to compare the reported 168 iodoacetamide-reactive cysteine residues with the S-nitrosylated residues in wild-type mouse heart (Table 1). Of the 119 S-nitrosylated proteins, only 38 were also iodoacetamide-reactive. Within the 38 proteins, 32 cysteine residues were both iodoacetamide-reactive and S-nitrosylated, whereas 40 and 46 cysteine residues were modified solely by S-nitrosylation or iodoacetamide, respectively. Therefore, only 15% of the S-nitrosylated residues (32 of 211 cysteine residues) in wild-type mouse heart were labeled with iodoacetamide, suggesting that the reactivity of cysteine residues toward different modifying agents is selective and guided by undefined structural and biochemical principles. Protein abundance or other basic biochemical properties of cysteine residues alone, such as pK a , surface exposure (two of the principal properties that govern reactivity), and hydropathy, do not sufficiently explain the observed selectivity for either S-nitrosylation or reactivity toward electrophiles. A trivial explanation is that the occupancy of these sites by S-nitrosylation may have prevented reactivity toward iodoacetamide. However, as discussed above, S-nitrosylation occupancy is not 100%, indicating that differences in reactivity guide this apparent selectivity. A limitation of these comparisons relates to the relative small size of the proteomes, which limits the number of three-dimensional structures available for analysis. A comparison between S-palmitoylation and S-nitrosylation using larger sets of proteins (1302 and 647 proteins, respectively; data collected from Refs. 17, 42, and 61) revealed that only 209 proteins were shared between these two modifications. This overlap represented 16% of the S-palmitoylation and 32% of the S-nitrosylation proteomes. S-Palmitoylation is an enzymatic process catalyzed by a family of zinc finger Asp-His-His-Cys (DHHC domain)-containing protein acyltransferases, creating a unique thioester bond that tethers proteins to membranes and directs localization of proteins to cellular lipid domains (16,17). Therefore, even when larger data sets are compared, irrespective of mechanism of formation (redox-dependent or enzymatic), cysteine modifications appear to target different proteins.
We propose that, in cellular proteomes, distinct and separate clusters of proteins that share cysteine residues within similar microenvironments show preferential reactivity toward specific modifiers. The biological selectivity is then a consequence of the enzymatic process or chemical modifier and the presence of strategically located cysteine residues that guide selective post-translational modifications. The best analogy is that of mail delivery in the city of Philadelphia. Different cargos are delivered to protein clusters within different zip codes. However, a zip code is not sufficient for accurate delivery. A house number (cysteine residue) and a street name (biochemical properties and the structural elements in the vicinity of cysteine residues) are also required. Therefore, future studies aimed at elucidating the apparent selectivity of cysteine targeting need to consider identifying only endogenously generated modifications and implement methodologies that pinpoint the sites of modification.
The site-specific identification of the modified cysteine residues is also critical because the same protein can be modified on the same or different cysteine residues in a manner that  (60); the S-glutathionylation data originated from multiple publications and include data from basal and oxidatively stressed cellular model systems and tissues; and the S-nitrosylation proteome is from wild-type mouse tissues (brain, heart, kidney, liver, lung, and thymus) under physiological conditions (42). The size of the circle is proportional to the number of proteins. The degree of overlap is modest; 31% of the S-glutathionylated proteins, 27% of the S-sulfenylated proteins, and 9% of the S-nitrosylated proteins are shared with the other modifications. Because the sites of modification are not available for all three modifications, it is unclear if the same residues are modified in the overlapping proteins.
achieves complementation in function. For example, synaptic targeting of the post-synaptic protein PSD-95 is reciprocally regulated by S-nitrosylation and S-palmitoylation on Cys 3 and Cys 5 (62). Similar co-regulation between S-nitrosylation and S-glutathionylation exists. In the case of thioredoxin, S-nitrosylation or S-glutathionylation at different sites results in opposing effects on activity, whereas isocitrate dehydrogenase is inhibited by either modification (63,64). The number of identified proteins that are co-regulated by several modifications is growing as methods improve, although the effects of redox modifications go beyond strictly activating or inhibiting enzymatic function. For example, S-glutathionylation of eNOS results in an uncoupling of the enzyme, leading to increased superoxide over NO production, which affects cellular redox signaling (65). In this way, the redox status of cysteine residues does not result in loss or gain of function but acts as a molecular switch dictating the products generated. Overall, analysis of the current data indicates that redox/cysteine modifications occur on predominantly distinct and non-overlapping proteomes. As noted previously, the biochemical and structural determinants of this apparent selectivity are not known, and future studies using enriched proteomes with site-specific identification are required.

Concluding Remarks
The expansion of work pertaining to NO and protein S-nitrosylation over recent years has led to some very intriguing and promising areas of research. Although the vast majority of the studies explored functional aspects of S-nitrosylation on specific proteins, the development of mass spectrometry-based approaches (60, 66 -69) has enabled studies that hold excellent promise in providing proteome-wide information for the structural and functional biology of S-nitrosylation. Acquisition of endogenous S-nitrosocysteine proteomes will also facilitate investigations on the issues that still remain uncertain, such as the biochemical mechanisms of S-nitrosocysteine formation, the mechanisms of denitrosylation, the effects on protein stability, the structural elements that may govern selective S-ni-

-nitrosylated and iodoacetamide-reactive cysteine residues and proteins in wild-type mouse heart
The asterisks indicate cysteine residues in the same tryptic fragment. The same cysteine residues that can be modified by S-nitrosylation and iodoacetamide are shown in boldface. Data were extracted from published data from wild-type mouse heart: S-nitrosylation (42) and iodoacetamide reactivity (11). Cys-SNO, S-nitrosylated cysteine. trosylation, and the molecular mechanisms by which protein S-nitrosylation regulates protein function and location.