Cell Physiology

Redox signaling: thiol chemistry defines which reactive oxygen and nitrogen species can act as second messengers

Henry Jay Forman, Jon M. Fukuto, Martine Torres


Except for the role of NO in the activation of guanylate cyclase, which is well established, the involvement of reactive oxygen species (ROS) and reactive nitrogen species (RNS) in signal transduction remains controversial, despite a large body of evidence suggestive of their participation in a variety of signaling pathways. Several problems have limited their acceptance as signaling molecules, with the major one being the difficulty in identifying the specific targets for each pathway and the chemical reactions supporting reversible oxidation of these signaling components, consistent with a second messenger role for ROS and RNS. Nevertheless, it has become clear that cysteine residues in the thiolate (i.e., ionized) form that are found in some proteins can be specific targets for reaction with H2O2 and RNS. This review focuses on the chemistry of the reversible oxidation of those thiolates, with a particular emphasis on the critical thiolate found in protein tyrosine phosphatases as an example.

  • hydrogen peroxide
  • thiolate
  • nitrosothiol
  • nitric oxide
  • signal transduction

although the involvement of free radicals in biology was assumed for many years to be restricted to damaging reactions, the discovery of the endogenous generation of NO in mammalian systems and the finding that this small, freely diffusing, chemically unique species participates in specific signal transduction pathways represented an important new paradigm and expanded views of the possible nature of cell communication and/or signaling processes. This novel role in signal transduction for ·NO and other reactive nitrogen species (RNS) now extends to reactive oxygen species (ROS) such as H2O2 and is gaining greater acceptance. The skepticism that still exists about these molecules acting as second messengers in various signaling pathways may vanish with better understanding of their chemistry, particularly regarding the differences in reactivity at high concentrations (mainly associated with pathology and toxicology) from those at low concentrations generated under physiological conditions in response to stimuli. Several excellent reviews have been published regarding evidence supporting a role for ROS and RNS in signaling (26, 32, 75, 82, 87, 117, 119), and we (35, 36) previously described the general properties that define a second messenger and showed how ROS and RNS fit into this role. In this review, the main focus is on the chemistry that may provide specificity, a necessary property of second messengers that has remained the most elusive in the study of ROS and RNS. 


Second messengers are generated at the time of receptor activation, are short-lived, and act specifically on effectors to transiently alter their activity. Indeed, ROS and RNS can be generated at the time of receptor activation and are short-lived, as are other second messengers, but the specificity of their action has been difficult to assess, except for that of ·NO, which binds specifically to the heme of the regulatory domain of soluble guanylate cyclase, resulting in its activation (53). Clearly, a reactive species such as ·OH cannot have any specificity, as it reacts at nearly the rate of diffusion with almost any molecule. Other “reactive species,” such as H2O2 and superoxide (O2·), are not as reactive as their collective name implies. H2O2, a highly diffusible molecule that can be produced endogenously via receptor-mediated mechanisms in many cell types (5, 61) and is enzymatically metabolized, has gained credence as a second messenger, although its targets have yet to be fully defined.


Exposure to a range of nonphysiologically significant concentrations of ROS or RNS that induce oxidative stress but do not kill cells can stimulate responses such as repair, adaptation, or transformation. The chemistry involved may include oxidation of lipids, proteins, DNA bases, and the sugar backbones of DNA and RNA, reactions that are mainly irreversible and induce repair mechanisms. Nonetheless, responses similar to those seen during oxidative stress can be triggered by other stresses, such as heat, ultraviolet radiation, and osmotic shock, as it is the damaged molecule rather than the causative agent that brings about these responses. We argue that in contrast to oxidative stress, redox signaling always involves responses that are specific to oxidation reduction reactions.

Both reversible and irreversible modifications of second messengers and proteins participate in signal transduction. Reversible reactions include phosphorylation of inositol and proteins, while irreversible reactions often involve the degradation of proteins such as that which occurs with an inhibitor of nuclear factor-κβ (Iκβ) during NF-κβ activation or with cyclin proteins during the cell cycle. Irreversible oxidation of the target molecule is common during oxidative stress. In contrast, redox signaling entails at least one reaction, which is usually reversible and involves the oxidation of a signaling molecule by a reactive species. In other words, in redox signaling, the reaction of the ROS or the RNS with the target is more reminiscent of on-off signaling associated with phosphorylation than it is of nonenzymatic lipid peroxidation. An exception to the reversibility rule is the situation in which irreversible oxidation is enzymatically catalyzed, such as that which occurs in the cyclooxygenase reaction. Thus, according to our definition, redox signaling occurs when at least one step in a signaling pathway involves one of its components being specifically modified by a reactive species through a reaction that is chemically reversible under physiological conditions and/or enzymatically catalyzed.

During oxidative stress, the ensuing stress response is likely to involve both redox signaling as defined above and nonenzymatic irreversible reactions, although the extent to which each of these processes contributes to the ultimate outcome varies with the concentration and chemical nature of the species. The rest of this review focuses largely on those reversible redox reactions.


Schafer and Buettner (93, 94) showed that the glutathione couple glutathione disulfide (GSSG)/reduced glutathione (GSH) dominates when describing the global state of reduced and oxidized species in the cell. GSSG/2GSH is regulated in part by the action of glutathione peroxidase, in which ROOH is a reactive hydroperoxide Math

Four seleno-glutathione peroxidases have been identified that have differential specificity toward H2O2 or lipid and other hydroperoxides (for review of glutathione peroxidases and a different perspective on redox state, see Ref. 33).

Glutathione disulfide reductase can restore GSH Math(2)

The GSSG-2GSH ratio, however, is also partially determined by export of GSH and GSSG, GSH synthesis, glutathione S-transferase catalyzed conjugation (Reaction 3), and disulfide exchange with proteins (Reaction 4) Math(3) Math(4) where GS is the dissociated form of GSH. It is unlikely for a thiol such as GSH to exchange with a disulfide without catalysis. Thus, in a slight variation of Reaction 4, the enzymes, protein disulfide isomerases (PDI), and glutaredoxin catalyze a reaction in which GSH substitutes for GS. The active sites of these two enzymes contain two cysteines, one of which is in the dissociated thiolate form (S). This type of active site, referred to as the thioredoxin (Trx) motif (CXXC), plays a prominent role in redox signaling (4, 52) and is discussed further below.

The global GSSG-2GSH ratio is thought to play a major role in maintaining the reduced state of most cellular molecules. Nonetheless, as signal transduction reactions are not global but localized processes, we argue that there is not a direct connection between the global GSSG-2GSH ratio and any specific redox signaling reaction. The importance of localized events in signaling is well illustrated by the sudden increase in cAMP that can occur after stimulation and activate a small portion of the cellular protein kinase A (PKA), even though enough phosphodiesterase is present in the cell to rapidly decrease cAMP to its prestimulation concentration. Thus cAMP activates a PKA molecule only within the distance allowed by the phosphodiesterase. Similarly, it seems that for H2O2 to play a role in signaling, its targets must be localized near its site of production because of the high cellular activity of glutathione peroxidase, catalase, and other enzymes that rapidly eliminate H2O2. Just as phosphatase activity is one factor in defining cAMP signaling, GSSG/2GSH is just one factor in determining H2O2 signaling. Therefore, the local concentration of H2O2 is as important as the global GSSG-2GSH ratio, if not more so.


What are the signaling reactions that can compete with glutathione peroxidase and other H2O2-eliminating enzymes? A reasonable assumption would be that they involve targets that share chemical traits with the enzymes that eliminate H2O2. Catalase and several other peroxidases use heme iron. The only described example of a heme iron-H2O2 interaction in signaling is the mimicking by H2O2 and catalase (presumably complex I) of the activation of guanylate cyclase by ·NO, which is not clearly of physiological relevance (120). Glutathione peroxidases are selenoproteins that catalyze Reaction 1 through a selenenic acid intermediate. Selenium is in the same series as sulfur and oxygen in the periodic table and shares similar chemistry with these elements. Although selenoproteins, glutathione peroxidases, and Trx reductases may be involved in regulating redox signaling, no selenoprotein has thus far been identified as a target component in a signaling pathway.

Before giving up on the idea of similarity between the chemistry of redox signaling and H2O2 elimination, we next briefly consider the chemistry of thiols on the basis of current knowledge about thiols and disulfides (33, 38, 52, 118). We also describe the mechanisms by which thiols may provide the specificity required for signaling as we discuss the extensive work of the Rhee and Poole laboratories with members of the peroxiredoxin (Prx) family, which are nonselenium peroxidases that use either Trx or GSH as a substrate (for review, see Refs. 83, 84, 88, 123).

A thiol (SH) does not react at physiologically significant rates with a hydroperoxide, such as H2O2, unless the reaction is catalyzed. Thiolates (S), however, react with hydroperoxides at rates varying from ∼10–105 M−1·s−1, depending on their local environment. In the active sites of some proteins, such as Prx and Trx, one cysteine is in the thiolate form, which can potentially react with H2O2. The general reaction series below defines thiolate-H2O2 chemistry where RSH is a thiol, RS is a thiolate, and RSO is a sulfenate [the ionized form of a sulfenic acid (SOH)] Math(5) Math(6)

The original thiolate can then be restored by exchange with another thiolate Math(7)

All Prx except one (Prx VI, also known as 1-cys Prx) contain two cysteines in their active sites, with one being a thiolate (S) residue. They demonstrate the following chemistry that is analogous to Reaction 5 and in which Prx-(SH)(S) represents a 2-cys Prx Math(8)

Evidence for the formation of the sulfenate intermediate was clearly provided by work on a bacterial Prx (AhpC) in which a second cysteine was mutated, allowing the formation of a stable SOH (27). Similar studies with mammalian Prx have not been conducted yet.

In the next step, the second thiol acts in a manner analogous to that of R′SH in Reaction 6 but results in the formation of a Prx intramolecular disulfide [Prx-(S)2] Math(9)

In the final reaction, all Prx except the mammalian Prx VI use Trx [formally written as Trx-(SH)(S), while the disulfide form is written as Trx(S)2] to restore the original thiolate Math(10)

An intermediate disulfide between Prx and Trx is likely to be formed, but because of the high propensity of Trx to form an intramolecular disulfide, the reaction appears to occur in one step. Trx disulfide does not readily exchange with other thiols or thiolates and must be reduced by Trx reductase with the addition of NADPH Math(11)

TrxS2 also may be formed by direct reaction of Trx with H2O2 (15). Nonetheless, the nonenzymatic reaction of Trx with H2O2 may be too slow to account for Trx oxidation under physiological conditions. It was demonstrated that Prx profoundly enhance the rate of reaction between Trx and H2O2 (15). Considering the ubiquitous and abundant presence of Prx in cells (16, 89), it is possible that all Trx(S2) may be formed through catalysis by the Trx-specific Prx isoforms.

As mentioned earlier, Prx VI/1-cys Prx has been found to use other thiols, most likely GSH under physiological conditions (72). The initial reaction of 1-cys Prx with H2O2 is similar to that of the other Prx, except that it forms a mixed disulfide instead of the intramolecular disulfide Math(12) Math(13) Math(14)

Because disulfide exchange is relatively slow for a biological reaction, even when GSH is dissociated to the thiolate, we propose that reduction of PrxVI-SSG requires an enzyme-catalyzed reaction with GSH. Recently, Fisher and coworkers (72) showed that Prx VI activity requires the presence of glutathione S-transferase-π. Hence it is possible that this enzyme functions as a catalyst in Reaction 14 under physiological conditions.


As mentioned above, ROS and RNS have been shown to alter numerous signaling pathways (for review, see Refs. 26, 32, 75, 82, 87, 117, 119). As an example, all of the mitogen-activated protein kinases (MAPK) (i.e., ERK, JNK, p38MAPK), which are activated through specific and separate kinase cascades, have been shown to be activated in various cell types by exogenous H2O2 or by receptor-stimulated production of H2O2 (105). While they do not appear to be direct targets, the upstream component targeted in each pathway and the nature of its alteration by ROS and RNS have not been clearly identified.

Nevertheless, recent data indicate that signaling proteins containing critical cysteines that must be retained for activity are potential targets for ROS and RNS in various pathways. For some of these signaling proteins, the cysteine has been clearly shown to be a thiolate [e.g., protein tyrosine phosphatase (PTP) (24), Trx (51)]. For others, the relative ease of their oxidation and rereduction strongly suggests that their critical thiols are in the thiolate form [e.g., the bacterial transcription factor OxyR (6), the eukaryotic transcription factors AP-1 (59) and NF-κβ (80), caspases (12)]. Similarly, the cysteines in the regulatory site of some protein kinase C isoforms are bound to zinc in such a way that the negative character of their sulfur atoms renders them susceptible to H2O2 oxidation, thereby altering the regulation of these important signaling proteins (40). Oxidation of the thiolate or substitution by site-directed mutagenesis suppresses the activity of these proteins. Nonetheless, although thiolates rather than thiols react with H2O2, the rate constants for some thiolates are still relatively slow and may require catalysis to form the sulfenate. This issue is addressed in the next section.

Table 1 lists the signaling proteins that have been clearly shown to be modified by ROS and RNS via thiol chemistry. Ras, a member of a family of small GTPases that plays a critical role in the activation of several signaling pathways (47), was one of the first such well-defined signaling proteins to be identified as a target for ·NO and H2O2, resulting in its activation (25, 62, 63, 115). However, the first evidence of reversibility of thiol oxidation by ROS and RNS was demonstrated with PTPs at the time these were being recognized as important players in signaling. The redox-mediated regulation of PTPs is supported by several in vivo studies (see below); thus our focus in this article is on the reversible inactivation of PTPs to illustrate both the chemical principles and the unanswered questions that will be the source material for investigations into redox signaling that are currently underway in many laboratories.

View this table:
Table 1.

Known reactive oxygen species/reactive nitrogen species targets in redox signaling


Phosphorylation of tyrosine residues is a major mechanism for posttranslational modification of proteins that results in a change in their function. For example, receptors for growth factors have intracellular protein tyrosine kinase (PTK) domains that are activated by binding of the growth factor and dimerization of the receptor, leading to transphosphorylation and the creation of binding sites for other proteins in the signaling pathway. It is important to note that under physiological conditions, tyrosine phosphorylation is markedly elevated only after stimulation. This indicates that the activity of PTPs predominates over that of PTKs. Further evidence for the dominance of basal PTP activity is demonstrated by the increase in tyrosine phosphorylated proteins in the presence of vanadate, a general PTP inhibitor (41). This suggests that the regulated inactivation of PTPs is a critical signaling mechanism. Indeed, redox signaling by H2O2 has been clearly demonstrated through the inactivation of several PTPs, as depicted in Fig. 1 (14, 19, 24, 39, 64, 65, 76, 100, 112).

Fig. 1.

Role of protein tyrosine phosphatase (PTP) oxidation in signaling. Step 1: a constitutively active PTP removes phosphate from protein X. Step 2: H2O2 is generated by an NADPH oxidase (NOX) protein in response to a stimulant. Step 3: H2O2 inactivates the PTP. Step 4: this can then result in the increased tyrosine phosphorylation of protein X if a protein tyrosine kinase (PTK) is already active or is activated in Step 5. PY, phosphorylated tyrosine residue.

Evidence of increased tyrosine phosphorylation caused by H2O2 was presented more than 15 years ago (48), although whether this resulted from increased PTK activity and/or decreased PTP activity is not clear. The first hint that PTPs may participate in redox signaling was based on the observation that nontoxic concentrations of oxidants caused an increase in protein tyrosine phosphorylation that correlated with PTP inactivation (24, 100, 112). Evidence of greater physiological relevance was provided by the observations that PTP1B could be reversibly inactivated by H2O2 produced endogenously in cells stimulated by epithelial growth factor (EGF) (65). At Rhee's laboratory, Kim et al. (57) then showed that PTP1B inactivation was due to a reversible oxidation of cysteine residues by H2O2, and others (76) showed that reversible oxidation of Src homology phosphatase-2 (SHP-2) by H2O2 is an essential step in PDGF signaling, confirming the initial in vitro work by Denu and Tanner (24). Nevertheless, several questions remain open regarding the mechanisms involved in reversible inhibition of PTPs by H2O2.

As mentioned above, all PTPs have a critical thiolate cysteine in a CX5R motif that participates in the dephosphorylation reaction (24, 31, 65). Therefore, the oxidation of the thiolate to a sulfenate, which cannot function in the catalytic process, would likely account for inhibition of PTP activity by H2O2. Although several studies provided some evidence suggesting the formation of an intermediate sulfenate in the active sites of PTPs, the rate of reaction of a PTP with H2O2 is ∼105 times slower than the Prx reaction (24, 33, 118) or the rate of oxidation of a critical cysteine in the bacterial transcription factor OxyR (6) (i.e., ∼10 M−1·s−1), which is quite slow. Thus the question is raised whether a nonenzymatic reaction can account for the formation of the sulfenate. In Fig. 2, two possible mechanisms are depicted: 1) H2O2 is generated by a NADPH oxidase closely neighboring the PTP so that the concentration of H2O2 is high enough to make the rate of the nonenzymatic reaction physiologically significant, and/or 2) another enzyme, essentially a cysteine oxidase, catalyzes the reaction. Supporting evidence for these mechanisms awaits better knowledge of compartmentalization and the identification of a cysteine oxidase, which would be of great importance in advancing the understanding of redox signaling. The PTP-sulfenate intermediate (analogous to the Prx VI reaction displayed in Reaction 12) can react with GSH to form a mixed disulfide (analogous to the Prx VI reaction displayed in Reaction 13), which also is catalytically inactive.

Fig. 2.

Inactivation and reactivation of a PTP by H2O2. Two mechanisms for the formation of an inactive sulfenate are shown. In one, a proximate NOX generates H2O2 close enough to the PTP active site that the concentration of H2O2 is high enough to make the rate of this nonenzymatic reaction physiologically significant. In the second, a putative cysteine oxidase catalyzes the reaction. In addition, there is a question of whether the removal of the glutathione in the final step requires catalysis that could be provided by glutaredoxin. SSG, glutathione-protein disulfide conjugate; GSH, reduced glutathione; GSSG, glutathione couple oxidized glutathione.

Other mechanisms have been proposed for the redox regulation of PTP activity, among which is the formation of a PTP-mixed disulfide intermediate through thiol/disulfide exchange with GSSG, which is increasingly referred to as glutathionylation (reversal of the reaction displayed in Reaction 4) (10). This mechanism presents two particular challenges, however. First, as stated above, the rate of exchange for GSSG with the thiolate in a protein is slow and probably would need catalysis. Second, because the reaction would produce GSH, which is 1–10 mM in cells, it would be unfavorable both thermodynamically and kinetically (23). Regardless, glutathionylation through thiol/disulfide exchange remains a likely mechanism in oxidative stress in which significant transient increases in GSSG occur. Formation of mixed disulfide by reaction of a sulfenate with GSH, as shown in Reactions 6 and 13, is more feasible both kinetically and thermodynamically.

Others (9) suggested that reversible inhibition of PTPs occurs via the reaction of the active site cysteine with O2·. The argument presented was that the relative rate constants for cysteine in its thiol form with O2· and H2O2 favored O2·. Nonetheless, this reaction is rather unlikely, as the active site cysteine has clearly been shown to be a thiolate created by a high pH environment in the PTP active site and superoxide radical is an anion with a pKa of 4.7. Furthermore, the reaction of O2· with a thiol, while significantly faster than that of H2O2, is 106 times slower than the rate of dismutation of O2· by superoxide dismutase, which is abundant in the cytosol.

The reversibility of inactivation of most PTPs occurs via reduction of the mixed disulfide to the thiolate (Fig. 2). This again raises the question of a role for an enzymatically catalyzed reaction, as nonenzymatic thiol/disulfide exchange with GSH is slow. The enzyme glutaredoxin, which is found in the cytosol and contains a Trx motif in its active site, may be involved in this reaction (Fig. 2). Glutaredoxin is related to PDI, most of which are found in the lumen of the endoplasmic reticulum, where they are involved in protein folding (107), although some PDI activity is associated with the plasma membrane (60) (for review of glutaredoxin, see Refs. 50, 52). Interestingly, in the case of low-molecular-weight PTP, which has a second cysteine in proximity to the active site, an intramolecular disulfide rather than a mixed disulfide is formed upon reaction with H2O2 (14). The reduction of an intramolecular disulfide is unlikely to occur without catalysis that may be accomplished by a PDI, glutaredoxin, or perhaps an NADPH-requiring enzyme such as Trx reductase. Thus we suggest that the recovery of the phosphatase activity may in all cases necessitate an enzymatic reaction under physiological conditions.

Further complexity in the reversible inactivation of PTP activity by H2O2 has been suggested by recent studies with isolated PTP1B, demonstrating the formation of an intermediate sulfenamide from reaction of the SOH intermediate with an amide in the PTP backbone in proximity to the active site (91, 109). This sulfenamide is reactive with GSH and could be an intermediate in the formation of the mixed disulfide. Nonetheless, whether the reaction of the sulfenate with the amide occurs faster than the reaction of the sulfenate with GSH under in vivo conditions of high GSH concentration is uncertain. Thus the mechanisms of formation of sulfenates and mixed disulfides are still far from well understood. The readers are referred to reviews by Poole et al. (84) and Claiborne et al. (21), which provide a more in-depth analysis of the current understanding of sulfenate formation in biology than that provided herein. To further complicate the picture, Woo et al. (122) recently showed that Prx I can be reversibly oxidized to sulfinic acid (SO2H), although oxidation to that state had previously been viewed as irreversible. A recent publication (11) described an ATP- and thiol-dependent reduction of the sulfinic acid form of yeast 2-cys Prx (Tsa1) catalyzed by a newly discovered yeast enzyme called sulfiredoxin. In the same article, the authors reported that sequences homologous to sulfiredoxin can be found in higher eukaryotes. While a role for reversible inactivation of PTP by H2O2 in redox signaling is gaining support, it is clear that further investigations are needed to fully understand the underlying mechanisms.


As for H2O2, it has become apparent that the biochemical targets for ·NO and RNS in signaling are metalloproteins and thiol proteins. That is, RNS are capable of chemically modifying critical thiols on numerous proteins and forming S-nitrosothiols (RSNO), leading to disruption and/or regulation of the target protein. Moreover, S-nitrosothiol formation has been regarded as part of the signaling mechanism for many of these proteins. In fact, many consider S-nitrosation (or S-nitrosylation1) to be analogous to phosphorylation as a signaling event (see, e.g., Ref. 73).

Myriad reports allude to the involvement of S-nitrosylation in cell signaling. For example, the N-methyl-d-aspartate (NMDA) subclass of glutamate receptors (69), the calcium channel ryanodine receptor (30), caspase activity (58, 67), metalloproteinase (45), PTPs (68), NF-κβ (74), Trx (46), and the Trx-ASK1 pathway (101) have been reported to be at least partially regulated by S-nitrosylation. In fact, nearly 100 proteins have been reported to be S-nitrosylated (49), making the comprehensive listing of all proteins potentially affected by S-nitrosylation beyond the scope of this review. Nevertheless, the partial listing above is indicative of the diverse array of systems that may be affected by S-nitrosylation. While it is clear that S-nitrosylation is a potentially important signaling process, knowledge of the mechanisms by which protein thiols are nitrosylated and denitrosylated in biological systems is far more tenuous.

The generation of S-nitrosothiols can occur via several mechanisms that are dictated by the cellular environment. In the following sections, we present various possibilities that have been established in purely chemical systems. Most of them, however, have not yet been firmly shown to occur in vivo. The most well-known and chemically accessible pathways involve ·NO-O2 reaction products (113, 116, 117). The reaction of ·NO with O2 generates nitrogen dioxide (NO2) as an initial product Math

Further reaction of ·NO2 with ·NO forms dinitrogen trioxide (N2O3) Math(16)

Both ·NO2 and N2O3 are capable of reacting with thiols. ·NO2 is a fairly potent one-electron oxidant that can oxidize thiols by a single electron to form nitrite and the thiyl radical that can directly react with ·NO to produce an S-nitrosothiol Math(17) Math(18)

This mechanism via initial NO2 oxidation and a thiyl radical intermediate have been proposed to be significant in vivo. Nonphysiological O2 concentration decreases RSNO formation because O2 and ROS can remove both thiyl radicals and ·NO. In contrast, the low O2 tension in tissues allows an increase in steady-state thiyl radical formation that may involve ·NO2 (55). Moreover, others (34) suggested that ·NO2 generated in the cytoplasm is unlikely to react with ·NO to form N2O3, owing to the rapid reaction of ·NO2 with thiols and urate. Thus there is the distinct possibility that nitrosation chemistry in the cytoplasm is dominated by rapid reactions of ·NO2.

An S-nitrosothiol also can be generated by the reaction of a thiol with N2O3 Math(19)

The chemistry outlined above can explain the endogenous generation of nitrosothiols from ·NO in an aerobic environment, although the reaction kinetics of these processes may preclude a role in vivo. The generation of ·NO2 is an overall third-order reaction (first order in O2 and second order in ·NO). Thus significant levels of ·NO2 are generated only in conditions of significant ·NO concentration. As “signaling” levels of ·NO are likely to be submicromolar, the rate of ·NO2 generation is slow. Nevertheless, Liu et al. (71) showed that ·NO and O2 favorably partition and accumulate in lipid membranes, increasing the likelihood of ·NO2-N2O3 chemistry. Although not yet established as the primary mechanism in vivo, the S-nitrosylation of protein thiols residing in membrane or hydrophobic environments may in fact occur via ·NO2-N2O3-mediated chemistry.

Metals have been shown to participate in ·NO-mediated nitrosation chemistry in chemical studies. For example, binding of ·NO to a ferric species generates an intermediate with significant nitrosonium ion (+NO) “character.” This intermediate species can further react with a nucleophile (e.g., a thiol), generating a nitrosated nucleophile and a ferrous species (110) Math(20) Math(21)

Of course, this metal-mediated chemistry requires the existence of an oxidized metal (Fe3+) in proximity to a nucleophilic thiol species to make a nitrosothiol, since other nucleophiles (including water) can also react with the ferrous nitrosonium intermediate. The in vivo relevance of this metal-mediated chemistry, however, remains speculative. It is worth noting that O2-dependent oxidation of ·NO is not the only way to generate ·NO2. Peroxidase-mediated oxidation of nitrite (NO2) can also form ·NO2, which can lead to thiyl radical formation and subsequent S-nitrosothiol generation, as described above (108) Math(22)

Although the peroxidase-catalyzed process outlined above requires NO2, which may come from O2-mediated oxidation of ·NO (and ·NO2 intermediacy), NO2 levels in cells can potentially be much higher than levels of ·NO. Thus, under cellular conditions in which significant NO2 has accumulated in the presence of H2O2 and a peroxidase, even low levels of ·NO can react with a thiyl radical (first order in ·NO) and lead to S-nitrosylation. It has been reported (43) that S-nitrosothiol formation can occur via a direct reaction of ·NO with a thiol followed by oxidation of the intermediate NO-thiol radical adduct. This intriguing report provides a first-order process (in ·NO) for the generation of S-nitrosothiols. However, the validity of this reaction remains to be established. Thus it is clear that S-nitrosothiols are formed in vivo; however, the mechanism by which they are formed requires further investigation, and a variety of factors, such as the proximity of an appropriate metal, the cellular environment of the thiol, the presence of peroxidases, and/or the concentration of the reacting species, dictate which of the described mechanisms are relevant for signaling in vivo.

As most biological signals need to be transient, the involvement of S-nitrosothiol in signaling pathways calls for mechanisms for its degradation. One such possible mechanism involves the attack of the nucleophilic thiol species at the sulfur atom of the S-nitrosothiol, resulting in the generation of nitroxyl (HNO) and the corresponding disulfide (3, 121) Math(23)

Another possible mechanism involves transnitrosation (i.e., transfer) of the equivalent of +NO from one thiol to another (7, 8). This does not directly result in the degradation of the S-nitrosothiol function but merely transfers it to another thiol Math(24)

The relative rates of these two processes are certainly a function of the accessibility of the attacking nucleophilic thiol to the two sites of attack as well as the relative electrophilicity of the sulfur and nitrogen atoms of the S-nitrosothiol. To date, the degree to which these factors dictate the reaction pathway has not been examined rigorously. Significantly, disulfide bond formation is the likely mechanism by which PTPs are inhibited by nitrosothiol species, since only reagents capable of reducing disulfide bonds regenerate activity (68).

S-nitrosothiols are also redox active and, indeed, can be destroyed by redox chemistry, with the most studied of such reactions being that with the cuprous ion (114). Cuprous ion reduces S-nitrosothiols to generate ·NO, thiolate, and cupric ion Math(25)

The prevalence of the chemistry in a biological system is dependent on the availability and accessibility of the reducing copper species. Since “free” levels of copper are extremely low in cells, the likelihood of this process occurring is highly dependent on the biological juxtaposition of the reducing copper with the S-nitrosothiol. O2· has been shown to react with S-nitrosothiols in a reaction that is second order in S-nitrosothiol, forming an oxidant and disulfide (56). As with mechanisms of their formation, the biologically relevant mechanisms by which S-nitrosothiols are degraded remains unclear and are a function of a variety of factors that are strictly defined by the nature of the chemical process (e.g., metal, thiol proximity, ROS generation).

One of the most important questions to ask when speculating about the importance and/or the existence of S-nitrosylation in cell signaling is that of specificity, as RSNO would be formed in a “sea of thiols.” Considering that glutathione can exceed millimolar concentrations in cells, the rate of reaction of the nitrosylating event must be significantly greater than the rate of reaction with other cellular thiols. The rate of thiol reaction with electrophilic and/or oxidizing species such as N2O3 or ·NO2 is determined in part by the thiol protonation state, as thiolates are much more reactive with RNS (and ROS, as mentioned above) than with thiols. Thus proteins with deprotonated regulatory thiols react at a faster rate with relevant RNS, a chemistry reminiscent of that with H2O2. Moreover, metal thiolates may also possess increased reactivity compared with simple thiols. For example, Chen et al. (18) reported that transnitrosation chemistry with the Zn thiolate functions of metallothionein is extremely facile. With respect to this issue, Stamler (99) also proposed that protein thiols in a consensus motif whereby the thiol is adjacent to a basic and acidic residue confer special reactivity conducive to S-nitrosothiol formation via nitrosation chemistry. Indeed, this motif has predicted thiol nitrosation in several systems containing multiple thiol targets (73). The proximity of a redox metal to a target thiol may be an important factor in thiol nitrosylation. Espey et al. (29) hypothesized that nitrosylation chemistry may occur in discrete domains or compartments on the basis of the compartmentalization or local levels of antioxidants such as ascorbate or glutathione.

The above discussion of the chemistry of potential signaling processes involving the interaction of RNS with thiol species merely describes possible mechanistic scenarios that have not been firmly established. Nevertheless, in spite of the current mechanistic ambiguities, it is clear that RNS-mediated signaling via thiol modification is a potentially important process that needs to be considered as a fundamental signaling paradigm.


Thiols have long been known to play an important role in regulating cellular function. The discovery that ROS and RNS can act as signaling molecules has spurred further studies of their possible targets, and, as shown in this review, redox thiol chemistry has taken center stage. As only those proteins with cysteines in the thiolate form significantly react with low concentrations of H2O2 or ·NO-derived species, one can postulate that the number of potential targets is limited, thereby providing specificity. In addition, the subcellular compartmentation of the target and the proximity of the source of reactive species are expected to play a role in specificity, although this has not yet been studied extensively. Nevertheless, Meng et al. (76) showed that only the pool of SHP-2 associated with the PDGF receptor was reversibly inhibited by H2O2 produced upon receptor binding, indicating that localization of SHP-2 close to the membrane where a putative oxidase is more likely to be present played a critical role in this signaling pathway. Similarly, protein-protein interactions through the scaffolding proteins PSD95 and CAPON are essential to the regulation of the NMDA class of glutamate receptor via S-nitrosylation (13, 54).

Because the intermediates purportedly formed during the oxidation reaction are highly unstable, a characteristic that is favorable for the reversibility of the reaction, it has been difficult to definitely prove their formation. While several “trapping” methods have allowed the identification of S-nitrosylated proteins in vivo, the formation of sulfenate by H2O2 has been mostly inferred (37). The reduction of sulfenate or S-nitrosothiols to thiolate is likely to involve the formation of mixed disulfide with GSH followed by disulfide exchange, as mentioned above. Although all of these reactions can occur nonenzymatically, enzymatic catalysis may be required to proceed rapidly enough in vivo (17, 37, 98). Other intermediates, such as a sulfenamide, may also be formed, but the general three-step process (see Reactions 5–7) includes the essential reversible steps required to fit the currently available evidence (Fig. 2). Thus the possible involvement of enzymes such as cysteine oxidase or glutaredoxin in this three-step process merits investigation.

Finally, one aspect that has not received careful attention to date is the timing and/or duration of the reversible inhibition. Some studies (76) have implied correlation between reversible inhibition of a target and downstream events; however, the mechanisms by which this process can be tightly controlled have not been put forth. Some (22, 96) have suggested compartmentalization of glutathione, protein S-glutathionylation, and mixed disulfides in cells with the use of fluorescence microscopy. Spatiotemporal recruitment of the target to these areas in the cells could be envisioned as a control mechanism. Better tools allowing in vivo observation of signaling events are forthcoming (106, 125) and will be of great benefit to the area of redox signaling as well as to other areas of signal transduction.


This work was supported in part by National Heart, Lung, and Blood Institute Grant HL-37556 and National Institute of Environmental Health Sciences Grant ES-05511.


We thank Leslie B. Poole (Department of Biochemistry, Wake Forest University School of Medicine, Winston-Salem, NC) for sending a preprint of Poole et al. (84).


  • 1 Some confusion exists regarding the terms “nitrosation” and “nitrosylation.” In the purest chemical sense, nitrosation refers to chemical processes involving the nitrosonium cation (+NO). Thus nitrosation of a thiol with the use of +NO (or a species with +NO “character”) can be termed S-nitrosation. This term is mechanistically defined. The term “nitrosyl” has been used by inorganic chemists to describe an NO function associated with, for example, metals. In the case of metal nitrosyls, this term does not describe the chemical process by which the metal-NO bond is formed. Another term, actually a prefix, often used in the literature is “nitroso-.” Like the term “nitrosyl,” “nitroso-” merely describes an NO function associated with another functional group. For example, organic chemists refer to a phenyl group with covalent attachment to NO as nitrosobenzene. Thus the prefix “nitroso-” is not mechanistically defined but describes only the connectivity of the atoms, and the term S-nitrosothiols is used to describe the S-NO function. The term “S-nitrosylation” has been adopted by many who do not want to be specific regarding how the functional group was chemically formed. Moreover, many researchers use the term “S-nitrosylation” as a means of drawing an analogy to other signaling phenomena such as phosphorylation and prenylation. Thus, for the sake of consistency, the term “nitrosylation” is used in this review to indicate the formation of an S-NO bond without regard to its mechanism. When the term “nitrosation” is used, we are indicating the involvement of +NO (or the equivalent).


View Abstract