The Redox Biochemistry of Protein Sulfenylation and Sulfinylation^*

Sulfenic Acid as a Post-translational Modification

Disulfide and sulfenamide formations protect Cys-SOH from further oxidation and lay the foundation for redox signaling. In fact, these post-translational modifications (PTMs) can generate conformational changes in protein structure and subsequent modulation of protein activity. In addition, as a result of its intrinsic nucleophilicity, Cys is present in the active site of many enzymes. Transient oxidation of these Cys residues is a well established process through which proteins can be spatially and temporally inhibited (1).

From the first evidence of its existence reported in 1976 (15) to the present, RSOH has been identified in a relatively small number of proteins. In fact, the identification of this elusive modification remains difficult. In 2008, Fetrow and co-workers (16) published a review that included a list of 47 proteins in which Cys-SOH was identified by crystal structure analysis. Because identification of the crystal structure of protein-SOH is problematic, their list represents only the tip of the iceberg. Direct mass analysis shows similar issues, making the use of chemical probes the only suitable technique to monitor RSOH formation (17).

Table 1 provides a list of the principal proteins in which Cys-SOH has been identified using chemical-trapping reagents. In addition to 4-chloro-7-nitrobenzofurazan, which can be employed only in vitro, dimedone-based probes (DBPs) are emerging as the most promising tool for RSOH trapping. These reagents are capable of crossing the cell membrane, capturing RSOHs directly in the cell (17). We concentrated our attention exclusively on those proteins in which the formation of Cys-SOH has been experimentally substantiated and shown to play a regulatory role. Proteomic studies based on the employment of DBPs have identified many other proteins that are able to generate an RSOH transient species in cells (18, 19), although the biological relevance of these oxidations remains unclear. Table 1 clearly demonstrates that many redox-regulated proteins are directly involved in cell signaling.

Protein-tyrosine Phosphatases

Tyrosine phosphorylation levels are maintained by the balanced action of protein-tyrosine kinases (PTKs) and phosphatases (PTPs). Sulfenylation of the catalytic Cys residue (pK_a = 4–6) of PTPs has emerged as a dynamic mechanism for inactivation of this protein family (20). The half-life of RSOH is generally quite low in PTPs. In fact, a neighboring cysteine residue (e.g. in PTEN) or the backbone amide nitrogen (PTP1B) readily reacts with RSOH to yield an intramolecular disulfide (21) or a cyclic sulfenamide species (14), respectively.

Recently, an alternative mechanism of inactivation emerged in SH2 domain-containing PTPs (SHP-1 and SHP-2) (22), which possess two highly conserved distal cysteines, both of which can generate a disulfide with the oxidized catalytic Cys residue. This intermediate disulfide typically rearranges into the more stable disulfide formed by the two backdoor cysteines to regenerate the free catalytic Cys residue. Surprisingly, the conformational change produced by the backdoor disulfide leads to an increased catalytic Cys pK_a value (∼9) with resultant inhibition.

Although SHP-1 and SHP-2 have structural similarities, they are regulated by different cell signaling pathways. For example, SHP-2 exhibits selective oxidation in response to PDGF in association with a PDGF receptor (23). However, these regulatory differences can be influenced by the method used to analyze their oxidation state. Employing an indirect RSOH detection method, T-cell activation, which induces H₂O₂ production, has shown transient oxidation of SHP-2 but not of SHP-1 (24). In a later work in which a DBP was used, both SHP-1 and SHP-2 showed Cys oxidation after T-cell activation, although with a different response time (25). These results could be rationalized by the greater sensitivity of direct RSOH analysis.

In vitro experiments have demonstrated that H₂O₂ deactivates SHP-1 and SHP-2 with second-order rate constants of 2.0 and 2.4 m⁻¹ s⁻¹, respectively (22). These values are similar to those observed with other PTPs and are apparently too low to justify their oxidation within the cellular context. Recently, our group has observed that, following EGF stimulation, SHP-2 forms a complex with the EGF receptor (EGFR) and Nox2 (6), which could provide an explanation for its high propensity to oxidize. In a similar manner, PTP1B, which is localized exclusively on the cytoplasmic face of the endoplasmic reticulum, appears to be oxidized through the H₂O₂ generated by Nox4, an NADPH oxidase highly abundant in the endoplasmic reticulum (5). These two examples highlight the importance of the proximity of the target protein to the ROS source in explaining PTP oxidation.

Kinases

Increasing research has highlighted the key role of H₂O₂ in the modulation of PTK activity. In comparison with PTPs, which are always inhibited by ROS, the oxidation of PTKs can lead to both enhancement and inhibition of kinase activity (26, 27).

The central role of Cys oxidation in PTK activity is exemplified by the redox control of EGFR signaling. EGFR is a receptor tyrosine kinase, the activation of which is involved in the regulation of cell proliferation, differentiation, and survival. In addition to promoting the tyrosine phosphorylation of protein targets, EGFR stimulation triggers the production of endogenous H₂O₂ by Nox activation. This localized increase in H₂O₂ concentration leads to the sulfenylation of a conserved Cys residue located within the intracellular kinase domain of EGFR (Cys-797), which enhances its tyrosine kinase activity (6). The redox regulation of EGFR could represent a more general mechanism for the modulation of other receptor tyrosine kinase activity. In fact, nine additional members of this family show a Cys residue structurally analogous to EGFR Cys-797, although further studies are needed in this direction.

Recently, Akt (a serine/threonine protein kinase) was also identified as a redox target. PDGF stimulation of fibroblasts induced H₂O₂ production, which led to isoform-specific regulation of Akt2 (28). A cysteine (Cys-124) positioned in the linker region connecting the PH (pleckstrin homology) domain to the kinase domain was found to be susceptible to sulfenylation. Cys-124-SOH can generate a disulfide with two distinct Cys residues (Cys-297 and Cys-311) located in the kinase domain. This modification negatively modulates Akt2, although in vitro experiments showed that disulfide formation has no direct effect on kinase activity. The inhibition mechanism remains unclear, but a previous work showed that Akt oxidation enhances its association with PP2A (protein phosphatase 2A), which could promote dephosphorylation of Akt (29).

Transcription Factors

In addition to the redox switch of PTK and PTP activities, which indirectly regulate transcription factors (TFs), H₂O₂ can directly modulate several TFs through the formation of intra- and intermolecular disulfide bonds (30). The first evidence of a redox-sensitive TF was identified in the bacterial TF OxyR, in which Cys-SOH mediates disulfide bond formation between Cys-199 and Cys-208 (31). Many other TFs are redox-regulated in prokaryotes (32–36), but relatively few cases have been identified in eukaryotes. In yeast, the activation of Yap1 represents an interesting case of TF redox regulation in which Gpx3-SOH mediates the oxidation of Yap1 through the formation of a Gpx3-Yap1 intermolecular disulfide (37, 38).

The anti-apoptotic NF-κB remains the only mammalian TF in which formation of Cys-SOH has been verified experimentally; however, this modification may also have a role in other peroxide-sensitive pathways of gene activation, such as the Nrf2/Keap1 system (39). H₂O₂ negatively switches the DNA affinity of NF-κB: directly through the oxidation of its p50 subunit at Cys-62 (40) and indirectly via Cys-179 sulfenylation of the β-subunit of the IκB kinase complex (IκB kinase β), the kinase that is responsible for the NF-κB activation along the canonical pathway (41).

Cysteine Proteases

Protein ubiquitination has emerged as a central PTM whereby lysine residues are conjugated to ubiquitin (Ub), a 76-amino acid polypeptide (42). Deubiquitinating enzymes (DUBs) cleave Ub or Ub-like proteins from the target, contributing to the balance of the Ub system. Four of the five different families of DUBs are cysteine proteases, which share in common a low-pK_a Cys residue essential for the catalytic mechanism.

Recently, three distinct works have shown that Cys oxidation can modulate DUB activity. Cotto-Rios et al. (43) reported transient sulfenylation of catalytic Cys for several members of the Ub-specific protease (USP) family and for UCH-L1. In particular, the authors established that USP1, a DUB involved in DNA damage response pathways, is reversibly inactivated following the induction of oxidative stress in cells. Additionally, Komander, in collaboration with our group (44), demonstrated that many members of the ovarian tumor DUBs also undergo Cys oxidation upon H₂O₂ treatment, including the tumor suppressor A20. Crystal structure analysis of oxidized A20 showed that transient RSOH can be stabilized by the formation of hydrogen bonds with the highly conserved residues located in the loop preceding catalytic Cys. Both works noted that each DUB family member exhibits a distinct level of sensitivity to oxidation. Differences in behavior can reflect various ranges of catalytic activation in which the conformational inactive enzyme could be less susceptible to oxidation. Lee et al. (45) confirmed this hypothesis by showing that preincubation of USP7 with Ub, which behaves as an allosteric activator, increases USP7 sensitivity to ROS.

An analogous inhibition has been found in small Ub-like modifier (SUMO) proteases. H₂O₂ treatment induces RSOH-mediated formation of an intermolecular disulfide in the yeast SUMO protease Ulp1 and in its human equivalent, SENP1 (46). Interestingly, SUMOylation also appears to be redox-regulated by reversible oxidation of the catalytic Cys residue of SUMO-conjugating enzymes (47), although no clear evidence of Cys-SOH formation has been provided.

Ion Channels

It is well established that ROS plays a regulatory role for some ion channels (48), but little is known about the molecular mechanism through which this modulation is explicated. For example, human T-helper lymphocyte ORAI1 channels, a member of the CRAC (Ca²⁺ release-activated Ca²⁺) channel family, are inhibited by oxidation of the extracellular Cys-195 residue (49), although the nature of this Cys oxidation remains unknown.

One exception is represented by the redox regulation of Kv1.5, a potassium voltage-gated channel expressed in the heart and pulmonary vasculature. Several studies have highlighted the fact that increased ROS concentration in cells is correlated with a reduction in Kv1.5 expression but have not provided a clear relationship between the two events. In collaboration with the Martens laboratory, we were recently able to elucidate the specific mechanism for Kv1.5 channel redox regulation (50). Labeling studies with DBPs have shown that a single Cys residue located in the extracellular C-terminal domain of Kv1.5 (Cys-581) forms Cys-SOH after H₂O₂ exposure. This modification triggers channel internalization, blocking Kv1.5 recycling to the cell membrane, and promotes its degradation.

Cellular Lifetime of Sulfenic Acid

Although limited solvent access and nearby hydrogen bond acceptors would contribute to RSOH stabilization, the absence of proximal thiols capable of generating an intramolecular disulfide is considered a major stabilizing factor. In the absence of neighboring Cys residues, RSOH can be directly reduced to RSH by thioredoxin (Trx) (Fig. 2A, Cycle 1) or may react with GSH to generate a mixed disulfide, which is later reduced by glutaredoxin (Fig. 2A, Cycle 2). For example, human serum albumin (HSA) has only one free cysteine residue (Cys-34), which is susceptible to H₂O₂ oxidation (rate constant of 2.5 m⁻¹ s⁻¹). We can estimate the half-life of HSA-SOH based on its reaction with GSH. Using the known second-order rate constant for this reaction (∼3 m⁻¹ s⁻¹) (51) and estimating the GSH concentration at 1 mm, the first-order rate constant would be 0.003 s⁻¹. Substituting this value in the equation t½ = ln 2/k, the estimated half-life of HSA-SOH would be ∼4 min.

In contrast, many redox-regulated proteins have a second proximal Cys residue that can form an internal disulfide with RSOH (Fig. 2B). In Cdc25c, for example, Cys-377-SOH reacts with the “backdoor” Cys-330 residue at a rate constant of 0.012 s⁻¹ (52). Applying the same calculations as above, the half-life of Cdc25c-SOH would be ∼1 min. Taken together, these estimated protein-SOH half-lives correlate well with the sulfenylation studies published by our group, and they appear similar to the cellular lifetimes of many other PTMs, such as phosphorylation. In A431 cells, we observed a peak of protein sulfenylation at ∼5 min after EGF stimulation, with a subsequent decay over 30 min (6).

Sulfinic Acid as a PTM

Cys-SO₂H was long considered merely an artifact of protein purification. However, increasing evidence indicates that hyperoxidation to RSO₂H is not a rare event. Indeed, quantitative analysis of soluble proteins from rat liver has shown that ∼5% of Cys residues exist as Cys-SO₂H (55). Finally, the discovery of sulfiredoxin (Srx), an ATP-dependent protein that specifically reduces Cys-SO₂H in the peroxiredoxin (Prx) family, has opened the door to an additional layer of redox regulation and increased interest in this specific modification (56).

Table 2 provides a list of proteins in which a biological functional role has emerged for RSO₂H. In comparison with Table 1, the number of reported proteins is decidedly exiguous. This does not necessarily indicate that RSO₂H plays a negligible role in protein redox regulation but rather reflects the lack of robust methods for monitoring the formation of such modifications within proteins. Although RSO₂H shows higher stability in comparison with RSOH, mass and crystal structure analyses can introduce a high percentage of artifacts. In addition, the emerging relevance of persulfide modification (RSSH), which has the same nominal mass shift of 32 Da, makes the use of high-resolution mass spectroscopy essential (57). We believe that the development of chemical probes capable of specifically trapping RSO₂H will push this Cys modification from the minor role to which it has been relegated. In this connection, we recently proposed the use of aryl-nitroso compounds as chemoselective probes for RSO₂H (58).

Prxs and Srx

Prxs are a family of cysteine-based peroxidases that remove H₂O₂ and other peroxides from cells. Being highly abundant and exceptionally efficient (second-order rate constant of 10⁵–10⁷ m⁻¹ s⁻¹), Prxs maintain the cytosolic concentration of H₂O₂ under 100 nm (59). Therefore, regulation of Prx activity is required to trigger H₂O₂-mediated intracellular signaling.

Typical Prxs exist in antiparallel dimeric or decameric forms and possess two Cys residues: “peroxidatic” Cys, which reacts directly with H₂O₂ to generate Cys-SOH, and “resolving” Cys, which forms an intramolecular disulfide with transient sulfenic acid. Finally, Trx reduces disulfide, restoring the catalytic cycle. Eukaryotic 2-Cys Prxs possess two sequence motifs (GGLG and YF) in their C termini that reduce the ability of resolving Cys to approach peroxidatic Cys-SOH (60). The resulting decrease in the disulfide formation rate allows a second molecule of H₂O₂ to react with peroxidatic Cys-SOH (Fig. 2C), generating a Cys-SO₂H. Such overoxidation leads to the deactivation of peroxidase activity and the formation of high-molecular-weight aggregates, which exhibit molecular chaperone activity (61). Although just 0.1% of peroxidatic Cys in human PrxI is oxidized to Cys-SO₂H during each turnover (62) at low concentrations of H₂O₂, recent kinetic studies demonstrate that Prx2 and Prx3 can undergo appreciable hyperoxidation without requiring recycling of the disulfide (63).

The peroxidase activity of 2-Cys Prxs is restored by Srx (64). The first step in the proposed catalytic mechanism involves the oxygen attack of RSO₂⁻ on the γ-phosphate of ATP and the resulting generation of a sulfinic phosphoryl ester (Fig. 2C). This species represents a sort of activated SO₂H, which collapses to a thiosulfinate intermediate (Prx-S(O)-S-Srx) after attack by a conserved Cys residue in Srx (65). Thiosulfinate is subsequently resolved by a third reducing species. Kinetic studies show that Srx is an inefficient enzyme. The rate of Prx-SO₂H reduction is indeed rather low (k₂ > 120 s⁻¹, k₃ ∼ 85 s⁻¹), suggesting that Prx requires a slow reparation process to allow H₂O₂ transient accumulation in response to extracellular signals.

Parkinson Disease Protein DJ-1

DJ-1 is a homodimeric small protein that has been associated with early-onset Parkinson disease (66). Many studies demonstrate that DJ-1 protects cells against oxidative stress-mediated apoptosis; however, the mechanism of its protective function remains largely unknown (54).

A conserved Cys residue, Cys-106, is extremely sensitive to oxidation and tends to form a Cys-SO₂H species, the generation of which appears to be critical for DJ-1 function. The highly conserved Glu-18 residue facilitates the ionization of Cys-106, reduces its pK_a, and helps to stabilize Cys-106-SO₂H through the formation of an unusually short and consequently strong hydrogen bond (67). Wilson and co-workers (68) have shown that small changes in this position can drastically influence the oxidation propensity of Cys-106. For example, in the DJ-1 E18D mutant, the distance between the thiolate and the protonated carboxylic side chain is increased, and Cys-106 is oxidized predominantly to sulfenic acid (68). In fact, Asp-18 tends to stabilize Cys-106-SOH, hampering further oxidation. In contrast, the structurally similar E18N mutant shows an increased propensity to oxidize even in the absence of H₂O₂. More important, E18D mutants fail to protect cells from ROS, whereas E18N mutant show similar levels of cell viability in comparison with the wild type, demonstrating that Cys-106 oxidation to RSO₂H is essential for maintaining protective functions (69).

Considering the high propensity of Cys-106 to oxidize, it has been proposed that DJ-1 acts merely as a direct ROS scavenger. However, an elegant new study reported that the DJ-1 C106DD mutant is still able to protect cells against oxidative stress (70), excluding direct scavenger action by Cys oxidation.

Cysteine Oxidation and Metal Binding Properties

Cys residues are very common in metal-binding motifs and can form coordinative bonds with several metal ions, including zinc, copper, and iron. Many proteins contain a Cys-zinc-Cys complex, for example, which furnishes structural rigidity. Oxidation of these cysteines causes Zn²⁺ release and a subsequent conformational change, which can switch protein function. Although oxidation is usually transient, through the formation of a disulfide bond, in some cases, it can lead to irreversible Cys-SO₂H (71).

Redox zinc switching is also involved in the activation of matrix metalloproteinases (MMPs). Matrilysin (MMP-7) contains a highly conserved cysteine switch sequence, PRCGVPDVA, in its prodomain. The thiolate side chain coordinates the catalytic Zn²⁺, contributing to the maintenance of enzyme inactivity. Fu et al. (72) showed that hypochlorous acid (HOCl), but not H₂O₂, can activate the enzyme through the conversion of Cys to RSO₂H, which disrupts zinc coordination. An analogous redox mechanism also appears to be involved in the activation of other MMPs (73).

The unique active site of nitrile hydratase (NHase) offers a sort of compendium of thiol oxidation states and metal coordinations. Structural analysis reveals that NHase contains an Fe^III or Co^III active site, in which three Cys residues, with three different oxidation states (RSH, RSOH, and RSO₂H), contribute to the coordination of the metal ion (74, 75). The fully reduced enzyme appears inactive, suggesting that Cys sulfenylation and sulfinylation are critical in maintaining the catalytic activity of NHase (76), probably by increasing the Lewis acidity of the metal ion. An analogous motif was more recently found in the catalytic site of thiocyanate hydrolase (SCNase), which incorporates Co^III only after Cys oxidation (77).

The active site of NHase and SCNase suggests that the oxidation state may influence Cys binding properties, switching the affinity from zinc (for RSH) to iron and cobalt (for oxygenated sulfur species). This change could provide additional redox control of protein functions (78).