Oxidative Cysteine Modification of Thiol Isomerases in Thrombotic Disease: A Hypothesis

Abstract

Significance: Oxidative stress is a characteristic of many systemic diseases associated with thrombosis. Thiol isomerases are a family of oxidoreductases important in protein folding and are exquisitely sensitive to the redox environment. They are essential for thrombus formation and represent a previously unrecognized layer of control of the thrombotic process. Yet, the mechanisms by which thiol isomerases function in thrombus formation are unknown.

Recent Advances: The oxidoreductase activity of thiol isomerases in thrombus formation is controlled by the redox environment via oxidative changes to active site cysteines. Specific alterations can now be detected owing to advances in the chemical biology of oxidative cysteine modifications.

Critical Issues: Understanding of the role of thiol isomerases in thrombus formation has focused largely on identifying single disulfide bond modifications in isolated proteins (e.g., α_IIbβ₃, tissue factor, vitronectin, or glycoprotein Ibα [GPIbα]). An alternative approach is to conceptualize thiol isomerases as effectors in redox signaling pathways that control thrombotic potential by modifying substrate networks.

Future Directions: Cysteine-based chemical biology will be employed to study thiol-dependent dynamics mediated by the redox state of thiol isomerases at the systems level. This approach could identify thiol isomerase-dependent modifications of the disulfide landscape that are prothrombotic.

Introduction

Oxidative stress promotes the generation of occlusive thrombi in many systemic diseases. Dyslipidemia and diabetes resulting in arterial thrombosis, inflammation contributing to disseminated intravascular coagulopathy, sickle cell disease causing vaso-occlusion, and aging promoting a procoagulant state are all examples of conditions in which oxidative stress contributes to stimulation of coagulation and platelet activation. Oxidative stress results in the generation of reactive oxygen species (Fig. 1), which are small molecules that participate in oxidizing, nitrating, nitrosating, or halogenating reactions, including modifications of thiols. These reactive species can modify proteins, lipids, and carbohydrates. Yet, the mechanisms that transduce specific redox reactions resulting from oxidative stress into signals that can be sensed by the thrombotic machinery are largely unknown.

FIG. 1.

Reactive oxygen and nitrogen species encountered in the vasculature. Reactive oxygen species are drawn on the left. Singlet oxygen is derived from one-electron oxidation (−1e⁻) of diatomic oxygen, whereas sequential one-electron reductions of diatomic oxygen yield superoxide radical anion (+1e⁻), hydrogen peroxide (+2e⁻), hydroxyl radical (+3e⁻), and water (+4e⁻). Reactive nitrogen species are drawn on the right with the one-electron reduction from the guanidino nitrogen of l-arginine. Sequential reduction yields nitric oxide radical (+1e⁻), nitrite (+2e⁻), nitrogen dioxide radical (+3e⁻), and nitrate (+4e⁻). Radical–radical recombination of superoxide anion and nitric oxide yields peroxynitrite, which is drawn in the middle of the figure. Protonation of peroxynitrite to the acid generates a highly reactive oxidant that decomposes to the hydroxyl and nitrogen dioxide radical.

Vascular oxidants modify a myriad of plasma and cellular constituents, and in some cases the effects of these modifications on platelets and blood coagulation have been characterized. Oxidation of low-density lipoprotein (LDL) particles to oxidized-LDL (OxLDL) in dyslipidemia is a well-known example of the generation of a prothrombotic entity from an oxidized plasma component. OxLDL not only promotes atherosclerosis (105), but it also induces endothelial dysfunction and directly activates platelets through stimulation of the scavenger receptor CD36 (85, 86, 172). Oxidation of von Willebrand factor inhibits its cleavage by ADAMTS13, enhancing its thrombotic potential (19, 53). Oxidation of thrombomodulin at methionine slows the rate at which the thrombomodulin–thrombin complex generates activated protein C (37, 151). Furthermore, methionine oxidation of activated protein C reduces its enzymatic efficacy (110). Redox-active species, however, are not always prothrombotic. Reactive nitrogen species (Fig. 1), such as nitric oxide radical (NO), bind to the heme group of soluble guanylyl cyclase and trigger cyclic guanosine monophosphate (cGMP) formation (73), inhibiting platelet activation (42). These examples underscore how redox-sensitive plasma and cellular constituents are important targets of oxidants and have direct implications in thrombotic disorders.

The sulfhydryl group on free cysteines is highly susceptible to oxidative modification. Modification of cysteines by two-electron oxidants can be prothrombotic. Oxidative cysteine modification of Src family kinases, for example, promotes tyrosine phosphorylation that is essential for signal transduction and maintenance in platelets (143). Platelet CD36 signaling in dyslipidemia promotes hydrogen peroxide generation to oxidize Src family kinases to maintain kinase activity (66, 150, 171). Oxidation of protein tyrosine phosphatases at an active site cysteine inactivates the enzyme, thus indirectly promoting kinase activity (75). There is also evidence that disulfide bonds form during fibrin polymer formation from fibrinogen (15). These examples demonstrate the important role that sulfhydryl groups serve in sensing the oxidative environment and transducing this information in a manner that impacts thrombosis.

A prominent family of redox-sensitive enzymes that rely on catalytic cysteines are the thiol isomerases. Thiol isomerases participate in a previously unrecognized layer of redox control of the thrombotic process by tightly regulating allosteric disulfide bonds (9, 30, 50, 93, 152) and by participating as a conduit for oxidative signaling (8, 48, 156). In this review, we explore the hypothesis that redox environment controls the catalytic activity and conformation of thiol isomerases, which in turn sort and process different redox signals converting oxidants into prothrombotic modifications (e.g., allosteric disulfide bonds). New evidence demonstrating that oxidative cysteine modifications of the active site cysteines of thiol isomerases regulate their function will be presented. Implications of these modifications for thrombus formation will be discussed. We also discuss advances that have been made in the detection of oxidant-mediated cysteine modifications and the implication of such advances for understanding thiol isomerase activity. Finally, we address potential therapeutic avenues with cysteine-sensitive electrophiles and nucleophiles as strategies to interfere with arterial and venous thrombosis.

Thiol Isomerases Are Redox-Sensitive Enzymes

Thiol isomerases are a family of 21 oxidoreductases classically known to catalyze proper protein folding in the endoplasmic reticulum (145, 168, 170). Thiol isomerases catalyze allosteric and structural disulfide bond oxidation, reduction, and isomerization for nascent proteins (26). These proteins also function as chaperones to assist in protein folding in the absence of catalytic activity (82). Although predominantly residing in the endoplasmic reticulum, thiol isomerases can escape this compartment and be transported extracellularly during cell activation.

Redox-active elements of thiol isomerases

Thiol isomerases have multiple thioredoxin-like domains with cysteine configurations in a CXXC motif in their catalytic (a, a′, or a⁰) domains, where the X denotes intervening amino acids. The thioredoxin-like domains are shown in Figure 2A. These thioredoxin-like domains consist of four-stranded antiparallel β sheet within three α helices (162). The different amino acids within the CXXC motif in the thioredoxin-like domains and the type of domains for the different vascular thiol isomerases are shown in Table 1. These catalytic CXXC motifs within the thioredoxin-like domains are required for protein folding and oxidoreductase activity (50, 139). The intervening amino acids within the flanking cysteines act as rheostats for its oxidoreductase activity by perturbing the spatial geometry and interfering with the local electrostatic environment (152). In this regard, the pKa of the nucleophilic cysteine is changed, impacting the protonation status, and thus reactivity, of the cysteine (28).

FIG. 2.

Thioredoxin structure and the domain organization of PDI. (A) The structure of the thioredoxin and thioredoxin-like domains of PDI. Left, thioredoxin (PDB: 3trx). Middle, the thioredoxin-like a′ domain of PDI (PDB: 1x5c). Right, the thioredoxin-like b′ domain of PDI (PDB: 3bj5). α helices are colored green, and β sheets are colored orange. The active site cysteines of the CXXC motif are colored magenta in the thioredoxin domain and the a′ domain. The cysteines shown to be nitrosated and sulfenylated are colored magenta in the b′ domain. In (B), PDI contains four thioredoxin-like domains in an a-b-b′-a′ configuration. The a and a′ domains are the active site domains that contain the catalytic CXXC motif for its oxidoreductase activity. The a and a′ domains have been shown to be sites of oxidative cysteine modifications, including disulfide formation (–SS–), cysteine nitrosation (–SNO), glutathionylation (–SSG), and succination (–S2S). Unlike the a domain, the a′ domain was shown to be mercurated (–SHg). The b and b′ are not catalytically active sites and do not contain the CXXC motif. These domains provide sites for interaction with its substrate. The b′ domain contains two cysteines that have been shown to be cysteine nitrosated (–SNO) and sulfenylated (–SOH). PDI, protein disulfide isomerase.

Table 1.

Thiol Isomerase Family Members

Thiol isomerase	Gene name	Amino acids	Molecular mass (kD)	Domain structure	CXXC motif
PDI	P4HB	508	57.1	a-b-b′-a′	CGHC
PDIp	PDIA2	525	58.2	a-b-b′-′	CGHC; CTHC
PDIr	PDIA5	519	59.5	a0-a-b-b′	CSMC; CGHC; CPHC
PDILT	PDILT	584	66.6	a0-a-b-b′	SKQS; CGHC; SKKC
ERp5	PDIA6	440	48.1	a-a-b	CGHC; CGHC
ERp18	TXNDC12	172	19.2	a	CGAC
ERp27	ERP27	273	30.5	b-b′	—
ERp29	ERP29	261	29	b-b′	—
ERp44	ERP44	406	46.9	a-b-b′	CGHC
ERp46	TXNDC5	432	47.6	a-b-b′	CGHC; CGHC; CGHC
ERp57	PDIA3	505	56.8	a-b-b′-a′	CGHC; CGHC
ERp72	PDIA4	645	72.9	a0-a-b-b-a	CGHC; CGHC; CGHC
ERp90	TXNDC16	825	93.5	a-a-a-b-b′	CX8C; CX9C; CX6C
ERdj5	DNAJC10	793	91.1	a0-b-a-a-a	CSHC; CPPC; CHPC; CGPC
TMX1	TMX1	280	31.7	a	CPAC
TMX2	TMX2	296	34	b	SNDC
TMX3	TMX3	454	51.8	a-b-b′	CGHC
TMX4	TMX4	349	38.9	a	CPSC
TMX5	TMX5	360	39.6	b′	CRFS
AGR2	AGR2	175	19.9	a	CPHS
AGR3	AGR3	166	19.1	a	CQYS

PDI, protein disulfide isomerase.

While members of thiol isomerase family are built from homologous structural elements, they differ in the organization of these elements. Here, we focus on the structure of protein disulfide isomerase (PDI), the archetypical thiol isomerase, as an example. PDI has an a-b-b′-a′-c domain structure (Fig. 2B). The noncatalytic (b and b′) domains are substrate binding and mediate protein–substrate interactions for its oxidoreductase activity as well as its chaperone function (11, 39). In most thiol isomerases, the b′ domain harbors a hydrophobic binding pocket that forms the primary site for substrate interactions. Both the b and b′ domains are flanked by the catalytic a and a′ domain. A flexible x-linker region consisting of 19 amino acids is flanked by the b′ and a′ domain (Fig. 2B). In the unligated state, the x-linker interacts with the hydrophobic pocket of the b′ domain. Extension of the protein supports substrate accessibility (112, 161). The x-linker region is also a hypothesized site for proteolytic cleavage to control thiol isomerase activity (161). Finally, the acidic c-terminal region containing the KDEL endoplasmic reticulum-retaining sequence completes the structure, and may play a role in its secretion from cells or surface membrane association (6) as well as contributing to chaperone activities (127).

Disulfide bond formation

Thiol isomerases participate in allosteric disulfide bond formation in the context of its oxidase function. The oxidase function is promoted when the active site cysteines of the catalytic sites in the a or a′ domain of thiol isomerases are oxidized. A free thiol in the target substrate protein, when deprotonated to a thiolate anion, is nucleophilic and attacks the N-terminal cysteine of the CXXC motif in either the a or a′ domain of thiol isomerases through a nucleophilic substitution (SN₂)-dependent mechanism, forming a mixed disulfide (56). Resolution of the mixed disulfide from an adjacent nucleophilic thiolate anion of the target substrate completes the transfer of a disulfide to the target substrate, yielding a reduced thiol isomerase and a disulfided substrate (Fig. 3A).

FIG. 3.

Disulfide formation and cleavage by PDI. The a′ domain is the major redox-active site for the prothrombotic activity of PDI. The gray arrows indicate attack by the thiolate anion for an SN₂-dependent nucleophilic substitution mechanism. (A) Disulfide formation on a substrate is initiated by nucleophilic attack on oxidized (disulfided) PDI forming a mixed disulfide intermediate. Resolution of the mixed disulfide by nucleophilic attack from a proximal thiolate on the substrate completes the formation of a disulfide. (B) Disulfide cleavage proceeds through the reverse reaction to disulfide formation. Cleavage is initiated by nucleophilic attack of a disulfide on a substrate by the thiolate anion of reduced PDI. Resolution of the mixed disulfide is mediated by nucleophilic attack from the proximal thiolate of the CXXC motif, which completes the transfer of the disulfide to PDI yielding a reduced substrate.

Disulfide bond cleavage

In the reverse reaction to its oxidase function, reduced thiol isomerase can cleave allosteric disulfide bonds. The disulfide bond types most susceptible to reduction by thiol isomerases have been described in the literature (26, 138). The reduced thiol isomerase aligns with a substrate that contains a disulfide bond, so that the free thiolate forms a 180° angle with the vector of the disulfide (14). This mechanism of action proceeds through deprotonation of the N-terminal cysteine of the CXXC motif in the a or a′ domain to a thiolate anion. This thiolate anion acts as the nucleophile to attack allosteric disulfide bonds on target substrates. This forms a mixed disulfide between the thiol isomerase and the substrate, thereby breaking the allosteric disulfide. Resolution of this mixed disulfide between the substrate and thiol isomerase by the proximal C-terminal thiolate anion of the CXXC motif transfers the disulfide from the substrate to thiol isomerases and completes the cycle (Fig. 3B).

Disulfide bond shuffling

As disulfides are formed and broken on substrate proteins by thiol isomerases, disulfide bond shuffling is required for correct introduction of the disulfide into the protein to achieve proper protein folding for optimal function. Disulfide bond shuffling proceeds through the same mechanism described above where electrons are shuffled between the thiol isomerase and the oxidized substrate until the disulfide is introduced into the correct pair of cysteines.

Redox cycling of thiol isomerases

Upon formation of structural or allosteric disulfides to proteins, thiol isomerases require catalytic regeneration of its oxidized (disulfided) state or reduced state by enzymatic and nonenzymatic means. The enzymatic regeneration of the oxidized state is through an oxidative relay driven by the delivery of oxidizing equivalents from its native oxidizing enzymes. Endoplasmic reticulum oxireductin 1 (ERO1) is the best characterized native oxidizing enzyme for thiol isomerases. ERO1 promotes thiol isomerase oxidation by delivering the oxidizing equivalent from flavin adenine dinucleotide to the protein. This mechanism generates reduced flavin adenine dinucleotide while consuming a molecule of diatomic oxygen to generate hydrogen peroxide as a by-product. α and β isoforms exist for the protein, and ERO1α is found on the surface of platelets after agonist stimulation where it supports platelet proaggregatory functions (155).

Several enzymes other than ERO1 can also reoxidize thiol isomerases. First, in an elegant biochemical approach by Chen et al., NADPH oxidase was shown to promote extracellular ERp72 oxidation by superoxide radical and hydrogen peroxide that are generated from the cell (22). In these studies, ERp72 functionally associates with NADPH oxidase at the plasma membrane, and oxidation of ERp72 inhibits its reductase activity (22). Second, thiol isomerase family members that are more electropositive than the member being oxidized could potentially reoxidize the enzymes. This oxidative cycling between thiol isomerase is intimately linked to reductive cycling. And finally, inside the cell, natural oxidases reside with thiol isomerases in the endoplasmic reticulum—including peroxiredoxins, glutathione peroxidases, and vitamin K epoxide reductase—which could oxidize selective thiol isomerase members (47, 116). Future studies will determine whether any of these oxidases are exposed extracellularly upon endothelial cell or platelet activation to cycle thiol isomerases to an oxidized state similarly to ERO1.

Nonenzymatic mechanisms also exist to reoxidize thiol isomerases chemically. The flux of available reactive oxygen species relative to the levels of thiol isomerases influences its reoxidation. This flux is directly linked to the activity of the sources of oxidants during oxidative stress. Several reviews describe the potential sources of vascular oxidants (142, 158). In addition to these enzymatic sources, ERO1 in the endoplasmic reticulum could promote reoxidation of thiol isomerases chemically by generating hydrogen peroxide as a by-product. Hydrogen peroxide is the major reactive oxygen species in the endoplasmic reticulum that oxidizes redox-sensitive proteins, including thiol isomerases (16, 157).

Analogous to the need to regenerate the oxidized state for oxidase function, reduced thiol isomerases require regeneration of the free thiol for its reductase function. However, unlike the oxidized state in which a natural oxidase has been discovered for thiol isomerase, a reductive cycling enzyme for thiol isomerase has not been identified. The reductase for thiol isomerases may be the more electronegative family members. Conversely, the transfer of the oxidative modification to a target substrate yields a free thiol. Finally, given that the oxidized state is driven by oxidants, antioxidative defense mechanisms (e.g., catalase, peroxiredoxins, glutathione peroxidases) prevent oxidation of the protein and thus indirectly maintain thiol isomerases in a reduced form.

Chemically, thiol isomerases are reduced by the reducing capacity of small molecular weight thiol redox buffers of the environment. Cysteines and glutathione provide the reductive buffering capacity. Nucleophilic attack of the oxidized thiol on thiol isomerase would generate a mixed disulfide with the cysteine or glutathione until resolved with a nearby small molecular weight thiol, yielding a cystine or oxidized glutathione. For this reason, it is common in the thiol isomerase field to add a reducing agent for continuous enzymatic assay for its reductive function.

An understanding of the redox cycling mechanism for thiol isomerases is critical to understand its mechanism as a prothrombotic agent. Given the limited amount of thiol isomerases secreted from or surface bound to the cell, it would seem highly inefficient for thiol isomerases to mediate platelet activation and blood clotting stoichiometrically with single turnover events. The identity of oxidases and reductases involved in redox cycling of vascular thiol isomerase in the context of thrombosis remains an area ripe for discovery.

Oxidative Cysteine Modification of Thiol Isomerases by Oxidants

Thiol isomerases are very susceptible to oxidative modification. In particular, the free thiols of the cysteines in the CXXC motif of the a and a′ domain on thiol isomerases are prone to oxidation by redox-reactive species generated in oxidant stress (106, 123). In select cases, these oxidant modifications mediate disulfide bond formation (123). Oxidation of the active site cysteines by select oxidants enables the thiol isomerase to introduce disulfide bonds into substrate proteins (140, 169). In other cases, oxidant-mediated modifications of active site cysteines within thiols can be transferred to substrate proteins (e.g., nitrosation, glutathionylation). Thiol isomerases can also be irreversibly modified by oxidants. In this manner, thiol isomerases sort oxidants and participate in redox signaling pathways. While many potential oxidative modifications of the catalytic cysteines of thiol isomerases are possible, modifications that have been studied include nitrosation, glutathionylation, and sulfur oxoforms.

Cysteine nitrosation of thiol isomerases

The nitric oxide radical is predominantly endothelial derived. Nitric oxide radical inhibits platelet activation potently not only by the well-defined binding of the heme group of soluble guanylyl cyclase but also by oxidizing cysteines and promoting cysteine nitrosation (65, 149). In the context of thiol isomerases, Bekendam et al. showed that PDI is nitrosated in the presence of S-nitrosocysteine, a transnitrosating agent, at the active site cysteine, predominantly in the a′ domain (8). In complementary experiments, decreasing nitric oxide generation of endothelial cells by exposing endothelial cells to a nitric oxide synthase inhibitor promoted the reductase activity of PDI and thrombin generation in response to plasma (8). Recent studies reported that PDI can also be nitrosated at Cys312 and Cys343 in the substrate binding domain (114, 164).

Cysteine glutathionylation of thiol isomerases

Like cysteine nitrosation, thiol isomerases are also regulated by cysteine glutathionylation. Glutathionylation occurs at the active site cysteines of thiol isomerases, driven in part by oxidative and nitrosative stress. Glutathionylation of thiol isomerases has been studied in cancer (160) and neurodegenerative disease (62), but has not yet been linked to coagulation mechanisms. Nonetheless, cysteine glutathionylation regulates platelet activation and aggregation (134). Glutathionylation in platelets was accompanied by a decrease in free glutathione levels when platelets were exposed to the chemical oxidant disulfiram (35). Glutathionylation in this context inhibited platelet aggregation.

Cysteine sulfur oxoforms of thiol isomerases

Although thiol isomerases are known to cycle through a reduced and oxidized state, the fate of the oxidized cysteine modified by two-electron oxidants (e.g., peroxides such as hydrogen peroxide, peroxynitrite, and lipid hydroperoxides) is less clear. In the presence of two-electron oxidants, the thiolate anion is susceptible to oxidation to the first sulfur oxoform, the transient sulfenic acid (33, 59). This modification is reversible and is at the central hub of many other oxidative cysteine modifications, including disulfides, glutathionylates, sulfenamides, thiolsulfenates, and thiolsulfonates, and further cysteine oxoforms including sulfinic and sulfonic acids (122) (Fig. 4). The sulfenic acid oxoform is prothrombotic in conditions of greatly elevated oxidant generation present in diseased conditions (171). Furthermore, a mass spectrometry screen of the platelet sulfenylome in the context of pathogen inactivation suggested that many proteins are sulfenylated; several of these proteins belong to pathways that promote integrin activation and cytoskeletal rearrangement (150).

FIG. 4.

Oxidative cysteine modifications of thiol isomerases. The sulfhydryl group of thiol isomerases (Cys-SH) is sensitive to cysteine modifications by enzymatic- and nonenzymatic mechanisms. Upon oxidation of the thiolate anion with peroxides, the sulfenic acid formation (Cys-SOH) is a labile and transient modification that is at the central hub of additional oxidative modifications. Further oxidation promotes sulfinylation (Cys-SO₂H) and sulfonylation (Cys-SO₃H). The sulfenic state could act as an electrophile to a thiol, and form a disulfide (Cys-SSR) either with itself or with other proteins. Condensation of sulfenic acid with glutathione will generate a gluthionylated cysteine (Cys-S-SG). Reaction of a sulfenic acid with an amine will generate sulfenamides or a cyclic sulfenamide. The presence of a nitrosothiol or nitric oxide will generate a nitrosated cysteine (Cys-S-NO). Sulfenic acid reactivity with a nearby cysteine sulfenic acid will yield a thiolsulfenate. Finally, potential reactivity with a selective alkylation group will generate an alkylated cysteine (Cys-S-R). The sulfhydryl group could be succinated by Michael addition of the nucleophile (the thiolate anion) with the carbonyl carbon of the Krebs cycle-derived fumarate. Another type of chemical modification is the reaction of the sulfhydryl group with methylmercury, a chemical pollutant, to generate a mercurated cysteine. Oxidative cysteine modifications that have been detected on thiol isomerases include cysteine nitrosation, glutathionylation, disulfide formation, sulfenylation, mercuration, and succination. The white thiol isomerase represents the reduced free thiol cysteine. Light gray thiol isomerases represent transient oxidative cysteine modifications. Medium gray thiol isomerases represent terminal oxidative cysteine modifications that are potentially reducible. Dark gray thiol isomerases represent terminal oxidative cysteine modifications that are likely irreversible.

Relating to thiol isomerases, the active site cysteines of PDI are very sensitive to oxidation by peroxides. Peixoto et al. found that peroxynitrite oxidizes PDI at a much faster rate than hydrogen peroxide (∼7 × 10⁴ M⁻¹ s⁻¹ compared with 17 M⁻¹ s⁻¹, respectively) (123). In this study, the authors utilized the intrinsic 4 nm blue shift fluorescence of PDI when excited at 295 nm as a readout of its redox state, and found that hydrogen peroxide promoted a slight dithiothreitol-reversible fluorescence decay. The kinetic studies suggested a much faster reactivity of peroxynitrite with free thiols compared with hydrogen peroxide for oxidative cysteine modification (59). These experiments were verified with elegant mass spectrometry approaches to confirm the oxidation of the active site cysteines. Importantly, however, in these experimental conditions, the authors were not able to identify further sulfur oxoforms (e.g., sulfenic, sulfinic, and sulfonic acid) even when using sulfenic acid-selective dimedones (123). It is possible that overoxidation of the active site cysteines by peroxynitrite eliminated the transient sulfenic state. Oxidation of surface bound PDI by urate hydroperoxide, a two-electron oxidant that targets sulfur-containing biomolecules, was also shown to decrease endothelial cell adherence (106). Using a combination of mass spectrometry approaches, kinetic studies, and in vitro endothelial cell culture assays, Mineiro et al. showed that the active site cysteines in the a and a′ domain are very sensitive to urate hydroperoxide-mediated alkylation with carbamidomethyl and disulfide formation (106). Stopped flow analysis to determine the second-order rate constant of reactivity suggested that this reactivity of urate hydroperoxide with the active site cysteines of PDI is quite fast (6 × 10³ M⁻¹ s⁻¹) (106). Endothelial cell studies with sulfhydryl-reactive agents indicate that cell surface proteins are predominantly oxidized by urate hydroperoxide or the oxidant diamide, that surface bound PDI could be oxidized by hydroperoxide, and that oxidation of the cells or inhibition of PDI decreased cellular adhesion (106). These studies suggest that oxidized PDI has a functional role in endothelial cell homeostasis, which may require further investigation in vivo. Identifying specific sulfur oxoforms of thiol isomerases by physiologic oxidants will be important to further understand its oxidoreductase activity.

Other cysteine modifications

Cysteine alkylation is another type of cysteine modification that is mediated by enzymatic and nonenzymatic mechanisms. Cysteine palmitoylation is the best characterized type of acylation in thrombosis and is an important regulatory modification for platelet function (40, 147). Although secreted thiol isomerases are not known to be regulated by cysteine fatty acylation, palmitoylation of TMX at residues adjacent to the membrane-spanning domain modulates TMX subcellular localization (98). In nonenzymatic cysteine alkylation, Michael addition reaction between the Krebs cycle intermediate fumarate and the thiolate anion of cysteine generates S-(2-succinyl)cysteine (also known as cysteine succination) (95). In a study by Manuel et al., cysteine succination was reported as an alkylated modification that links PDI dysfunction to metabolic and endoplasmic reticulum stress in diabetes (100). The authors showed that endoplasmic reticulum stress driven by mitochondrial dysfunction in high-glucose conditions promotes accumulation of the Krebs cycle intermediates—fumarate and malate (100). The backup of electrons to fumarate promoted cysteine succination of many different types of proteins within adipocytes, including the active site cysteines of PDI. Using mass spectrometry-based approaches, Manuel et al. (100) found that the C-terminal cysteines in the CXXC motifs are the targets of succination, and that this modification inhibited the enzyme's reductase activity.

Finally, cysteine mercuration occurs during endoplasmic reticulum stress by methylmercury, an environmental pollutant important in neuronal cell dysfunction. PDI cysteine mercuration was observed in neuroblastoma cells when treated with methylmercury (67, 99). Interestingly, it is the active site cysteines in the a′ domain of PDI that is S-mercurated (99). Functionally, cysteine mercuration prevents nitrosation of PDI by S-nitrosocysteine (99).

As shown in Figure 4 and discussed above, the free thiol of a cysteine could be oxidized to transient, reducible modifications (light gray) or terminal modifications (medium gray or dark gray). The modifications that are terminal and potentially reducible are in medium gray. While chemical reduction is more typical, in some instances enzymatic reduction of the modification may be needed, such as the case for glutathionylation and sulfinylation with glutaredoxin and sulfiredoxins, respectively. In the case of disulfide formation by two-electron oxidants, augmentation of thiol isomerase oxidase function could potentially be achieved. However, peroxides react with varying second-order rate constants to the thiolate anion to generate sulfenic acids (59). This means that milder oxidants, such as hydrogen peroxide, may shift sulfenic acid to disulfides (functional for activity and still reducible) based on the physical proximity between the active site thiols of the CXXC motif. This is in contrast to stronger oxidants, such as peroxynitrite, which still generate sulfenic acid but may potentiate further sulfur oxoforms, including sulfinic and sulfonic acids, which are generally irreversible. The irreversible nature of the modification would render the protein inactive. Notably, the reactivity rates between hydrogen peroxide and thiol isomerase cysteines are slower than their reactivity rates with the oxyferryl porphyrin in catalase or other heme peroxidases (113, 123). Based on these possible mechanisms, we hypothesize that thiol isomerases sense oxidants, which they sort, process, and transmit, to achieve selective oxidative cysteine modifications, thereby creating redox signaling pathways. This hypothesis is already validated for transnitrosation events by thiol isomerases where cysteine nitrosation is passed to a substrate and prevents proaggregatory and procoagulant functions (8).

Redox Control of Thiol Isomerase Conformation

Just as the fate of distinct oxidants is affected by their interaction with thiol isomerases, so too are thiol isomerase conformations modified by the redox environment. Structural studies of PDI reveal distinct structural configurations, for example, in the reduced versus oxidized state. X-ray crystallographic experiments evaluating the reduced and oxidized states of the protein show that the a and a′ domains of PDI are ∼27 Å apart in reduced PDI compared with 40 Å apart in oxidized PDI (162) (Fig. 5A). The X-ray crystallographic studies, however, are limited by the constraint of the specific conformation in the crystal environment. The structure of PDI in solution is substantially more dynamic. A recent study by Chinnaraj et al. used single-molecule fluorescence resonance energy transfer (smFRET) to evaluate the dynamic conformation changes of PDI in solution (23). PDI conformation was probed by site-specific incorporation of the unnatural amino acid Prk at lysine 42, 308, and 407 followed by attachment of fluorescent probes using copper-mediated azide alkyne cycloaddition reactions (click chemistry). The smFRET approach also showed that PDI in solution samples multiple open and closed structural configurations in a redox-dependent manner (24). In addition to smFRET data, small-angle X-ray scattering data of reduced and oxidized PDI showed that PDI is in a more extended conformation in solution (96, 117), suggesting a more complex redox regulation of the structure of thiol isomerases, which could impact its activity.

FIG. 5.

Redox-induced conformational changes of thiol isomerases. (A) Redox environment influences the structural configuration of PDI as demonstrated by its X-ray crystal structure. A reduced environment promotes a compact PDI structure compared with the more open oxidized PDI. Figure adapted from Wang et al. (162). (B) Schematic structures from atomic force microscopy of reduced and oxidized PDI. These studies suggest that oxidized PDI is in equilibrium between an open and closed conformation, whereas reduced PDI is in a closed/twisted conformation. Figure adapted from Okumura et al. (117).

An important unaddressed question is whether physiologic oxidants (e.g., two-electron oxidants such as hydrogen peroxide, peroxynitrite, and lipid hydroperoxides) influence the structural state of the protein. This is especially relevant in the context of oxidative stress where oxidants readily generated in pathophysiologic conditions could influence the oxidase (and isomerase) activity of the PDI or induce oxidative cysteine modification. Cysteine modification outside of the active site may also impact the structural configuration of the protein as evidenced by altered protein activity of PDI when nitrosated at Cys343 (114). Furthermore, bepristats, which are chemical inhibitors that target the substrate binding b′ domain, paradoxically enhance the catalytic activity of PDI (7). It is likely that oxidative modifications at sites other than the catalytic sites (e.g., in the b and b′ domain) promote change to the overall structure of the protein and influence activity.

Further evidence to support redox regulation of the PDI's structure by oxidants was derived from elegant time-resolved single-molecule observations using atomic force microscopy. Atomic force microscopy allows direct visualization of the protein immobilized on a surface. Okumura et al. showed that cysteine oxidation by the chemical oxidant diamide promoted multiple configurations of PDI (117). Reducing the protein using dithiothreitol promoted compaction of the protein, which is in agreement with the structural studies that reduced PDI is more compact (162, 173) (Fig. 5B). Surprisingly the oxidized form of PDI also becomes more compact compared with the reduced form and dimerizes in the presence of multiple different types of substrates (117). This study suggested potential multimerized conformation states of the protein to regulate oxidative folding.

Dynamic changes to the active site cysteine residues of thiol isomerases influence their structural configuration. Chinnaraj et al. performed smFRET using a PDI construct in which fluorescent probes were introduced at Ser88 and Lys467 (24). They showed that PDI toggles dynamically between open and closed configurations in the millisecond timescale, and that the redox environment dictates conformation. In addition, the authors used a PDI variant in which the active site cysteines in the CXXC motif of the a and a′ domains were mutated to alanine. Unexpectedly, the authors found that loss of free thiols, and not disulfide formation as was previously thought, was responsible for directing the closed conformation (24). Although the specifics of how modification at these domains influences the change in the overall structure of the protein remain unknown, this observation and additional mutational analyses suggest that these thiols may participate in a network of allosteric interactions that control the conformation of PDI. The mutant protein showed similar structural configuration to the oxidized protein and was also insensitive to the sulfhydryl reducing agent dithiolthreitol. Notably, the use of irreversible small molecule inhibitors targeting the active site cysteines shifted the configurations toward a closed ensemble, also suggesting that it is the availability of the sulfhydryl group that controls the structure of PDI. Redox modification of the CXXC motif as described in Figure 4 would therefore influence the equilibrium of the protein between the open and closed ensemble. The structural understanding of thiol isomerases by these modifications will allow for a detailed mechanism of their oxidoreductase and chaperone activities in addition to identifying potential therapeutic approaches to target such redox states of the protein.

Effects of Redox Environment on Thiol Isomerase Function

The redox state of secreted thiol isomerases is not known. As a result, we do not know whether thiol isomerase-mediated formation or cleavage of disulfide bonds, or both, functions in thrombus formation. It is likely that the pool of secreted PDI is in a mixture of both the reduced and oxidized states and not exclusively one or the other. Data in support of this are obtained from studies looking at the redox status of PDI within and outside of cells (72). PDI could be in a reduced or oxidized state depending on the type of cells, the stimulant of the cells, the local redox state of the environment, or the method used to determine the redox status (72). Burgess et al. (12) evaluated the dependency of PDI redox status on platelet activation. They found that the number of free thiols in surface-bound PDI increased in activated platelets compared with resting platelets (12). This observation could be due to the release of more PDI, to activation-induced reduction, or a combination. In a similar study by Jiang et al., PDI secretion from fibroblasts promoted a decrease in net disulfides on the cell surface (78). Other reports suggest that PDI is predominantly reduced (77, 104, 136, 159), raising the possibility that PDI is secreted in a reduced form. However, in HEK293 cells (5) and microsome isolated from the rat liver (111), PDI was shown to be either predominantly oxidized or in a partially oxidized state (72). Based on the different redox states of PDI, it is useful to think of the redox state as a ratio of reduced to oxidized in a manner similar to how small molecular weight thiols are described (e.g., reduced glutathione and oxidized glutathione; cysteine and cystine). This ratio is dependent on the local redox environment.

Redox potential of the plasma in relation to thiol isomerase activity

In the context of platelets and endothelial cells, the plasma is the proposed extracellular environment where secreted thiol isomerases participate in thrombus formation. The plasma is buffered by two major thiol-dependent redox buffers: (i) cysteines and cystines, and (ii) reduced glutathione and oxidized glutathione (58). Elegant work looking at redox potential of the plasma in both human and nonhuman primates suggests that the ratios of cysteines to cystines and reduced glutathione to oxidized glutathione decrease with aging (132). These studies suggest that the free thiols that buffer plasma are oxidized in older individuals, which supports the theory that oxidative stress contributes to adverse events in many diseases. The mechanisms of oxidative stress during aging are multifactorial and are actively being researched. Age-related oxidative stress is also implicated in the pathophysiology of thrombosis (38); yet the redox potential of these small molecular weight thiols at the site of thrombus formation has not been studied. It is likely that the redox status of PDI is directly related to the ratio of reduced and oxidized small molecular weight thiols, which may shift toward an oxidized state in diseased condition. As the plasma is also an oxidizing environment (redox potential of ∼140 mV more negative than the intracellular environment) (79), it is likely the redox potential of the plasma in combination with oxidant generation that govern whether the ratio of reduced to oxidized thiol isomerases is shifted toward the oxidized state.

Redox potential between thiol isomerases might govern a complex electron relay

In addition to the plasma redox potential as a potential modulator of thiol isomerase activity, the redox potential of thiol isomerase family members may influence a complex chain of electron relay between family members. This hypothesis is supported by experimental evidence that electron shuffling occurs between the thiol isomerases (115). This electron relay could be governed by the redox potential as outlined by Chiu and Hogg (26); the redox potentials range from the more electronegative active site cysteines of thioredoxin (−270 mV) to the more electropositive redox potential of the active site cysteines of ERp57 (a and a′ domains at −167 mV and −156 mV, respectively) (26) (Fig. 6A). As electrons transfer from the more electronegative to the more electropositive redox potential (more reducing toward electropositive and more oxidizing toward electronegative potentials), a tightly coordinated redox relay may be present extracellularly to regulate thrombosis by modifying substrates involved in thrombus formation. It is also possible that during oxidative stress where reactive oxygen species influence the thiol states of proteins, this electron relay may be broken or shifted toward an oxidative relay to the more electronegative members (Fig. 6B). On the contrary, in the conditions of reductive stress, the electron relay may be shifted to the more electropositive members. The observation that inhibition or genetic knockdown of a single thiol isomerase inhibits thrombosis is consistent with the premise that the thiol isomerase family members act in a series of electron transfers. As a proof of concept, using the PDI kinetic trapping mutant, the identification of thioredoxin as biased toward oxidation and ERp57 showing bias toward reduction (152) is consistent with redox potentials between the members and the proposed electron transfer between electronegative and electropositive species. These studies also suggest that some thiol isomerases physically interact with one another. However, whether these interactions occur for all thiol isomerases or involve additional substrates requires investigation as physical interaction—as well as redox potential—is necessary to promote electron transfer.

FIG. 6.

Hypothesized influence of oxidative stress on an extracellular electron relay between thiol isomerases. We propose that thiol isomerases sense the reductive or oxidizing environment and transfer electrons accordingly so as to maintain redox homeostasis. (A) Electrons generally flow from the more negative redox potential (mV, E₀) to the more positive redox potential (reducing condition). Reduction of a disulfide on a thiol isomerase by the more electronegative member only occurs when the energetics between the two thiol isomerases are matched (26, 120). In this case, a lower redox potential thiol isomerase will promote electron donation to the higher redox potential thiol isomerase, thus cleaving the disulfide bond. (B) In physiologic oxidative stress, we propose that the electrons participate in transitioning between thiol isomerases and between substrates until an end-stage disulfide is introduced into a protein as the stable oxidative cysteine modification. This oxidative condition is influenced by many factors, including the relative oxidizing potential of the oxidants, their stoichiometry and proximity relative to the thiol isomerases, and their reactivity rates to redox-sensitive cysteines.

Deprotonation status modulates thiol isomerase activity

The reactivity of the active site cysteines of thiol isomerases is controlled by their ability to be deprotonated. Several factors influence deprotonation, including the pKa of the cysteines and the pH of the local environment. Cysteines have pKas ranging from 4 to 10. If the pKa of the cysteine is greater than the pH, the cysteine is shifted toward a more protonated state. Given the physiologic pH of ∼7.3 and that the formation of a thiolate anion promotes its reactivity to oxidants, lower pKa cysteines have decreased nucleophilicity of the thiolate anion (101, 122). The pKa of a cysteine on a protein is governed by solvent accessibility, proximal polar or positive amino acids, and the factors in the microenvironment that stabilize the hydrogen leaving group by decreasing the energy barrier of the transition state (133). Indeed, the pKa of the active site cysteines of PDI was shown to modulate its oxidoreductase activity (83). In addition, the conserved Arg120 of thiol isomerases impacts the pKa value of the active site cysteine and thus its catalysis (92). Although PDI was studied as the model protein in this context, this is likely to be the case for the other thiol isomerase family members as the pKas of their active site cysteines are significantly lower than the reported pKa of cysteine of 8.3 (36, 64, 81). The pH of the environment is also known to influence the activity of thiol isomerases as suggested by earlier studies utilizing a peptide substrate as an indicator of thiol isomerase activity (2, 135). A 10-mer peptide was constructed based on the fluorescence of a tyrosine residue in the N-terminus, which could be quenched with an arginine residue in the C-terminus upon disulfide formation. PDI-mediated disulfide formation within the peptide could be achieved only when an appropriate redox buffer was present. The rate of disulfide formation increased with increasing pH (about fourfold increase in the rate from pH 4.5 to the maximal rate observed ∼pH 7) (2, 135). These studies indicate that pH of the local environment impacts thiol isomerase activity. Novel chemical biology approaches to detect and image pH changes in real-time will facilitate the evaluation of pH influence on thiol isomerase activity (174).

Proximity to oxidant sources influences thiol isomerase redox status

The proximity to sources of oxidants, including the prominent plasma membrane NADPH oxidase, influences the redox state of nearby proteins. Thiol isomerase redox states, for example, are influenced by proximity to NADPH oxidase. Chen et al. showed in Chinese hamster ovary cells that ERp72 functionally associates with and is oxidized by oxidants from NADPH oxidase at the plasma membrane (22). This association may be independent of oxidation status of ERp72 and may require an extracellular stimulus. In a related study, Gimenez et al. showed that inter- and intramolecular disulfides between NADPH oxidase subunit p47phox and PDI promoted oxidant generation and smooth muscle cell migration during atherosclerotic plaques formation (57). This study showed that Cys400 of PDI and Cys196 of p47phox were required for the association between the proteins. The absence of all four active site cysteines of PDI prevented superoxide anion generation.

Detection and Targeting of Oxidative Cysteine Modifications

Advances in chemical biology allow for reactive cysteine profiling and detection of specific oxidative cysteine modifications that could be employed to study not only thiol-dependent dynamics in thrombosis and hemostasis but also redox regulatory mechanisms of thiol isomerases. As the sulfur cysteine exists in different oxidation states, oxidative cysteine modifications of the free thiol will yield different electrophilic and nucleophilic species (59). In the following section, we discuss some of the electrophilic and nucleophilic probes that have been or could be employed to study thiol-dependent dynamics and oxidative cysteine modifications that occur during blood clotting.

Sulfhydryl-reactive compounds

The free thiol of a cysteine is nucleophilic and has a low redox potential (−0.27 to −0.125 V) (59). This property is commonly utilized for efficient alkylation of the sulfhydryl group with electrophilic or Michael acceptor-based probes. Examples of these probes are shown in Figure 7A, which includes alkylating agents, N-ethylmaleimide (121) and iodoacetamide (108), and the mixed disulfide-forming methyl methanethiolsulfonates (MMTS) (84), pyridyldithiol-based probes (63), and dithiol nitrobenzoic acid [DTNB; also known as Ellman's reagent (41)]. These probes covalently tag the sulfhydryl group through Michael addition or nucleophilic substitution mechanisms (122). It should be noted that maleimides are kinetically faster than IAMs for the labeling of free thiols and have a broader range of pH for its utility; yet, maleimides could potentially modify the amine group of amino acid side chains, such as lysines and arginines (10, 122). Fluorogenic derivatives of these alkylating probes have been described (124). Other halogen-based probes, such as the commonly utilized monobromobimane, have also been employed based on their fluorogenic properties. In a study to look at efficient blocking of the sulfhydryl group by forming disulfides, pyridyldithiols were shown to label sulfhydryl groups much faster than MMTS (34), but this could be related to the efficiency of the probes at target pH ranges. Forming a mixed disulfide with DTNB or pyridyldithiols (2,2′-dithiodipyridine or 4,4′-dithiodipyridine; 2-DPS and 4-DPS, respectively) is also utilized for quantification of sulfhydryls. The mixed disulfide products formed by these compounds can be quantitated spectrophotometrically (412 and 324 nm in the case of nitrobenzoic acid and 2- or 4-thiopyridone, respectively). Efficient labeling of the sulfhydryl group with these compounds allows for direct detection of reduced cysteines and indirect detection of oxidative cysteine modification using measurable handles such as fluorophores, biotin, or isotopes. The compounds depicted in Figure 7 are commonly used sulfhydryl-based probes; probes for specific applications have been described (124).

FIG. 7.

Probes for chemoselective targeting of sulfhydryls on cysteines and oxidative cysteine modifications. Sulfhydryl-reactive compounds are drawn in (A). The sulfhydryl of cysteine is nucleophilic and can be alkylated with electrophilic probes, including NEM and IAM. Probes that form a mixed disulfide with the sulfhydryl group include MMTS, pyridyldithiol-based DPS, DTNB, and mBrB. (B) The cysteine sulfinic state is also nucleophilic and can be tagged with electrophilic probes. Shown is NO-Bio, which contains a biotin handle, a linker, and an aryl nitroso group for trapping of sulfinic acid. Also drawn is DiaAlk, which utilizes the electrophilic nitrogen species to trap the sulfinic acid. The sulfenic state of cysteine oxidation holds both electrophilic and nucleophilic properties. (C) Nucleophilic probes have been designed to trap the sulfenic state (60, 61). These include the benzothiazine-based probe (BTD), 1,3-cyclohexanedione-based probe (or also known as dimedone; CHD), a pyrrolidine-based probe (PYD), and a piperidine-based probe (PRD). Of these, the BTD probe has the fastest reactivity to the sulfenic state. The others show differential reactivity to buried or exposed thiols. Not drawn with these nucleophiles is the alkyne handle for azide–alkyne cycloaddition reactions. (D) Two examples of strain-based alkyne (BCN) and alkene (norbornene) electrophilic probes for the selective detection of cysteine sulfenic acids are shown. BCN, bicyclo[6.1.0]nonyne; BTD, benzothiazine-based carbon nucleophile; DPS, dithiodipyridine; DTNB, 5′5′-dithiol-bis-(2-nitrobenzoic acid); IAM, iodoacetamide; mBrB, monobromobimane; MMTS, methyl methanethiosulfinate; NEM, N-ethylmaleimide; PYD, pyrrolidine-based carbon nucleophile.

The so-called biotin-switch method has been well described in the literature as a standard approach for indirectly detecting reversible oxidative cysteine modifications (74), including nitrosation (74), glutathionylation (97, 131), sulfhydration (109), and sulfenylation (137). In this approach, the free sulfhydryl groups on the target protein are blocked by alkylation with maleimide or other sulfhydryl-reactive agents as described above. The modified cysteine is then reduced, and the newly available sulfhydryl group labeled with a maleimide-biotin, -fluorophore, or -isotope (74). It should be noted that with this approach, the blocking step has to be 100% efficient, or false positives will occur in the second alkylation step. Furthermore, the reduction for the modification has to be selective and efficient, or the second alkylation step will not occur. Some potent reductants for oxidative cysteine modification include dithiothreitol (31, 55) and tris(2-carboxyethyl)phosphine (13, 55), ascorbate for the selective reduction of nitrosated cysteines (70, 90) and arsenite for sulfenylated cysteines (89, 128, 137). In the case of glutathionylation, enzymatic reduction of the modified thiol may be required [e.g., with glutaredoxin (97, 131)]. Successful use of the biotin-switch assay indicated that PDI undergoes cysteine nitrosation as a regulatory mechanism for vascular thrombotic complications (8). A second method to detect cysteine modifications is loss of labeling by sulfhydryl-selective compounds. This approach was utilized to show that oxidative cysteine modification occurs on src family kinases when platelets are activated through the prothrombotic CD36 receptor (171). The redox status of surface proteins on activated platelets was also probed using sulfhydryl-reactive agents. These studies showed that the platelet surface contains ∼26% free thiols and increases almost fourfold to ∼81% free thiols with activation (12, 103). However, it could be possible that upon activation, there is an increase in the surface expression of proteins (e.g., due to granule secretion), which could cause the apparent increase in available free thiols.

Sulfhydryl-selective electrophilic compounds are utilized in mass spectrometry-based experiments to study thiol dynamics and to characterize allosteric disulfide bond modifications by thiol isomerases (25). In addition, high molecular weight derivatives of maleimide could be utilized to observe a shift in the protein's molecular weight when the cysteines are modified, which was employed to show that secreted PDI, but not ERp57, reduces vitronectin to support thrombosis (9). These approaches are all complementary, and useful for tracking electron transfer between thiol isomerases and their substrates.

Electrophilic probes for the selective detection of cysteine sulfur oxoforms

Electrophilic probes have also been designed for the direct detection of cysteine sulfenylation and sulfinylation, the sequential sulfur oxoforms generated by two-electron oxidation of the cysteine sulfhydryl groups (Fig. 4). The cysteine sulfinic acid oxoform generated from sulfenic acid was originally thought to be irreversible until the identification of sulfiredoxin reductase (17). Like free sulfhydryl groups, sulfinic acid is nucleophilic and is generated by attack of the peroxide species when sulfenic acid is deprotonated (33). Using chemical biology approaches, elegant work by the Carroll laboratory showed that aryl nitroso compounds are chemoselective probes for the detection of sulfinic acids (32). They further developed the probe NO-Bio (Fig. 7B) for sulfinic acid nitroso ligation and found it to be selective for sulfinic acid as observed by oxidation of the model protein DJ-1 and detection of sulfinylation in cells oxidized by hydrogen peroxide (34). In addition, they found that electrophilic diazenes react with sulfinic acids and have developed the DiaAlk probe (Fig. 7B) for chemoproteomic approaches (1). Using A549 and HeLa cells treated with hydrogen peroxide, the authors identified a number of sulfinylated cysteines in multiple proteins. In particular, with their HeLa cell dataset, they showed that ERp57 was a cysteine sulfinylated protein detected by DiaAlk labeling. Sulfinylation occurs on Cys85 and Cys92, which were not previously recognized as redox-active sites.

Carbon nucleophiles for the selective detection of cysteine sulfenic acid

As mentioned above, sulfenic acid has both nucleophilic and electrophilic properties (59). The nonselective labeling of nucleophilic free sulfhydryls by 4-chloro-7-nitrobenzofurazan (NBD-Cl) prompted a need for more selective probes for the sulfenic state. As such, the Carroll laboratory synthesized and characterized nucleophilic probes that target the electrophilic property of sulfenic acids (146). Nucleophilic 1,3-cyclohexanedione (or dimedones) is now commonly utilized for detecting this modification and has been utilized to detect sulfenic acids in platelets (150, 171). However, dimedones are slow in capturing the transient sulfenic state. The Carroll laboratory profiled many carbon nucleophiles that more efficiently tag the sulfenic state (60). They found that the benzothiazine-based carbon nucleophiles (BTD) (Fig. 7C) react with the sulfenic state 170 times faster than dimedones (61, 146), which may be important in conditions where low levels of oxidants are generated to augment thrombosis. Using HeLa cells treated with hydrogen peroxide, the Carroll laboratory found in their screen that PDI is sulfenylated on cysteines 312 and 343 as detected by chemoproteomics with BTD (1). These cysteine sites are known to be sensitive to oxidative cysteine modification as described above for nitrosation (114, 164). BTD will enable the determination of whether the active site cysteines in the thioredoxin-like domains are sensitive to sulfenic acid formation, which might shift the pool of extracellular thiol isomerases toward the oxidized state as sulfenic acids are at the central node of oxidative modifications described in Figure 4.

We recently utilized BTD in platelets to profile platelet CD36-mediated hydrogen peroxide signaling in dyslipidemia, a condition associated with greatly elevated oxidative stress (171). We found that Src family kinases, which are highly prone to oxidation and become activated in the sulfenylated state (66, 150), are a target of hydrogen peroxide by CD36 signaling (171). We also utilized the carbon nucleophiles that were profiled against the sulfenic state as a way to prevent reduction of sulfenic acid back to a sulfhydryl group or further oxidation for sulfinic and sulfonic acids (171). Within this framework, we found that cysteine sulfenylation is important for the proaggregatory and prothrombotic nature of platelets when utilizing a panel of carbon nucleophiles shown in Figure 7C. In line with determining the functional relevance of sulfenic acids in vivo with these nucleophiles, the detection of in vivo sulfenic acids is challenging, given the requirement of click chemistry adaptation and the toxic effects of copper in the reaction. These challenges prompted the development of copper-free methods for detecting cysteine sulfenic acids, which might be feasible for ex vivo and in vivo thrombosis experiments.

Electrophilic strain-promoted cysteine sulfenic acid labeling.

Electrophilic alkyne- and alkene-based strain-promoted probes have been developed for the copper-free detection of sulfenic acids in cells and organisms (87). Figure 7D shows two examples of alkyne- and alkene-based probes that target sulfenic acids (3, 102, 125). These probes utilize the nucleophilic properties of sulfenic acids for covalent adduction. In the case of the alkene-based norbornene probe, Alcock et al. (3) found in their screen that the thiol isomerase member ERp5 was covalently adducted with norbornene when HeLa cells were treated with hydrogen peroxide; the control conditions without hydrogen peroxide showed less ERp5 enrichment. Although not one of their major enrichment targets, the detection of ERp5 by norbornenes as a sulfenylated target of hydrogen peroxide, which has not been previously found with dimedones, suggests that selective sulfenic acid probes may be required to detect sulfenylation on certain thiol isomerase members in specific environments within or outside of the cell. This is exemplified by bicyclo[6.1.0]nonyne (BCN), which is an alkene-based probe for sulfenic acids that is cell permeable, has low cellular toxicity, and can be targeted to the mitochondria by incorporation of a mitochondrially targeted triphenylphosphonium motif (94). These strain-promoted copper-free sulfenic acid probes could allow for imaging of the dynamics of sulfenylation in animal models of thrombosis. However, with every probe, a detailed understanding of their mode of action is required to prevent misinterpretation of results (126, 146).

With advances in chemical biology tools to detect further sulfur oxoforms, it will be possible to determine whether further sulfur oxoforms of thiol isomerases regulate clotting and bleeding since the sulfenic state is at the central hub of oxidative cysteine modification and was shown to be prothrombotic (59, 171).

Thiol Isomerases and Cysteine Switches in Thrombosis

Several members of thiol isomerases are found in the vasculature, and some of these have been implicated in blood clotting. Specifically, the secreted thiol isomerase PDI (9, 29, 88, 178), ERp57 (69, 165), ERp5 (43, 80, 119), and ERp72 (177) were shown to support platelet accumulation and fibrin deposition in a variety of in vivo murine models of thrombosis. In contrast, the membrane-bound thiol isomerase TMX1, which is present on platelet surfaces, was found to be antithrombotic (176). The evidence for a role of thiol isomerases in animal models of thrombosis is based on genetic models, antibody inhibition, and small molecule inhibitors (Table 2), and has been described in several reviews (27, 49, 50, 52, 54, 139, 168, 170). In a phase II study, a reduced incidence of thrombosis was observed in cancer patients treated with a PDI inhibitor (179). Despite this evidence, the mechanisms of action of thiol isomerases in thrombosis remain largely unknown. Several in vivo studies have indicated that the catalytic motifs of PDI function in thrombus formation (88, 177, 178). These observations have led to the inference that the oxidoreductive function of thiol isomerases is critical for their prothrombotic properties. A corollary of this idea is that thiol isomerases act to either cleave or introduce disulfide bonds into substrate proteins involved in thrombus formation. Efforts have been made to identify substrates in thrombosis. An additional possibility is that active site cysteines in thiol isomerases facilitate alternative oxidative modifications of thiols such as nitrosation or sulfenylation.

Table 2.

Deletion or Inhibition of Thiol Isomerases Inhibits Thrombus Formation

Mode of inhibition	Thrombosis model	Effect on thrombus formation	Refs.
Genetically engineered mice
Platelet/megakaryocyte-specific ERp57 null	FeCl3-induced carotid artery injury	∼1.9-fold increase in occlusion time	Wang et al. (165)
Platelet/megakaryocyte-specific PDI null	Laser-induced injury cremaster arteriole	∼75% inhibition of peak platelet accumulation	Kim et al. (88)
Platelet/megakaryocyte-specific PDI null	FeCl3-induced carotid artery injury	2-fold increase in occlusion time	Zhou et al. (178)
Mice expressing a mutant PDI (CGHC/SGHS)	FeCl3-induced carotid artery injury	∼1.5-fold increase in occlusion time	Zhou et al. (178)
Platelet/megakaryocyte-specific ERp72 null	Laser-induced injury cremaster arteriole	∼50% reduction in platelet accumulation; no decrease in fibrin deposition	Zhou et al. (177)
Endothelial cell-specific ERp72 null	Laser-induced injury cremaster arteriole	75% reduction in platelet and fibrin accumulation	Zhou et al. (177)
Endothelial cell-specific ERp72 null	FeCl3-induced mesenteric artery injury	50% reduction in platelet accumulation	Zhou et al. (177)
Platelet/megakaryocyte-specific TMX1 null	FeCl3-induced mesenteric artery injury	2-fold increase in platelet accumulation	Zhao et al. (176)
Inhibitory antibodies
Anti-PDI antibody (RL90)	Laser-induced injury cremaster arteriole	Minimal platelet accumulation; no fibrin generation	Cho et al. (29)
Anti-PDI antibody	Ligation Model	∼70% reduction in fibrin formation at 30'	Reinhardt et al. (130)
Anti-ERp5 antibody	Laser-induced injury cremaster arteriole	70% decrease in platelet accumulation; 62% decrease in fibrin generation	Passam et al. (119)
Anti-ERp57 antibody (Mab1)	FeCl3-induced carotid artery injury	∼3-fold increase in occlusion time	Wu et al. (167)
Anti-ERp57 antibody	Laser-induced injury cremaster arteriole	∼80% reduction in platelet accumulation	Holbrook et al. (69)
Small molecules, peptides, and proteins
Quercetin-3-rutinoside, 0.5 mg/kg	Laser-induced injury cremaster arteriole	Near absence of platelet accumulation and fibrin generation	Jasuja et al. (76)
Quercetin-3-rutinoside	FeCl3-induced carotid artery injury	2.2-fold increase in occlusion time	Jasuja et al. (76)
PDI b′ domain-targeting bepristat	Laser-induced injury cremaster arteriole	90% reduction in platelet and fibrin accumulation	Bekendam et al. (7)
Pan-thiol isomerase inhibitor Zafirlukast	Laser-induced injury cremaster arteriole	Modest ∼15% reduction in platelet accumulation	Holbrook et al. (68)
PDI abb′x fragment	Laser-induced injury cremaster arteriole	Reduced platelet accumulation by 60%	Wang et al. (166)
Cysteine nitrosated PDI	Laser-induced injury cremaster arteriole	50% reduction in both platelet and fibrin accumulation	Bekendam et al. (8)

Thiol isomerase substrates in thrombosis

As thiol isomerases participate in making and breaking allosteric disulfide bonds, substantial efforts have been made to identify their substrates. Trapping mutants of thiol isomerases have been constructed and utilized as a tool to identify such substrates in plasma and on platelets (44, 141, 152). Relating to thrombosis, Stopa et al. mutated the intervening glycine and histidine sequences between the flanking cysteines of the active site motif in PDI to allow for a slower resolution of the oxidoreductase cycle (152). Specifically, mutations of this intervening sequence stabilized the oxidoreductase in a reduced or oxidized form depending on perturbation of the spatial geometry or interference with the local electrostatic interactions (44, 152). In coupling this method with mass spectrometry, they found that cathepsin G, glutaredoxin-1, thioredoxin, glycoprotein Ib [GPIb], and fibrinogen are biased toward oxidation (152). This is in contrast to annexin V, heparinase, ERp57, kallikrein-14, serpin B6, tetranectin, and collagen VI, which showed a bias toward reduction (152). These studies revealed that PDI was associated with many different proteins in specific redox states (152).

Another approach is to mutate the more C-terminal cysteine in the CXXC motif to an alanine or serine to enable thiol isomerases to form mixed disulfide with its target substrate without completing the catalytic cycle (9). Using this approach, Bowley et al. identified vitronectin, histidine-rich glycoprotein, protein S, and prothrombin as redox substrates of PDI. They further showed that reduction of the disulfides in vitronectin at Cys137-Cys161 and Cys274-Cys453 mediates binding to β3 integrin (9). The authors used a laser-ablation arterial thrombosis model to demonstrate that reduced vitronectin accumulates at the site of vessel injury in mice. Furthermore, accumulation of vitronectin could be prevented by inhibition of PDI. These studies suggest a functional role for PDI in reducing allosteric disulfide bonds.

It is possible that this method of mutating the intervening sequence of the CXXC motif or using the trapping mutants (CXXA) could be adopted for other vascular thiol isomerases to identify additional substrates and define substrate specificity. In addition, this approach has applications beyond thrombosis and hemostasis (e.g., evaluating protein folding defects in neurodegenerative disease and cancer).

Potential substrates upon which PDI acts during thrombus formation have also been identified via candidate approaches. Thrombospondin-1 was the first extracellular substrate identified for PDI. Secreted PDI isomerizes a disulfide bond in the C-terminal domain of the protein (71). Platelet α_IIbβ₃ has long been considered a substrate of thiol isomerases, and several reviews have addressed the possibility that thiol isomerase-mediated modification of α_IIbβ₃ functions in thrombus formation (45, 46, 107). Evidence from PDI-deficient platelets indicates a role for PDI in α_IIbβ₃ integrin activation (88). ERp5 is also known to interact with platelet α_IIbβ₃ (80). Recent studies also show that ERp5-mediated cleavage of Cys177-Cys184 within the βI domain of α_IIbβ₃ facilitates release of fibrinogen from the integrin (118). Despite these observations, the mechanism by which thiol isomerases modify α_IIbβ₃ during thrombosis remains an area of active investigation. The longstanding proposal that modifications of allosteric disulfide bonds in tissue factor by PDI are important for thrombus formation has also been reviewed (4, 21, 91). Inhibition of PDI impairs tissue factor activity in endothelial cells (8, 20) and peripheral blood mononuclear cells (18, 91). Yet, the relevance of thiol isomerase modification of tissue factor to thrombus formation in vivo remains unknown. Trapping studies also identified platelet factor Va as a PDI substrate, and inhibition of PDI interfered with factor Va generation and platelet-dependent thrombin generation (153). GPIbα has previously been implicated as a PDI substrate (12). More recently, Li et al. showed that PDI cleaves GPIbα Cys4-Cys-17 and Cys209-Cys248, and regulates GPIbα function. They also showed that PDI is essential in GPIbα-mediated thromboinflammation, promoting platelet–neutrophil interactions in vivo (93). A great challenge in determining whether PDI substrate functions in thrombus formation is to show that a single disulfide bond in a candidate substrate functions in thrombus formation in vivo (51). This level of proof has not yet been achieved for any substrate candidate.

Inhibition of thrombosis by nitrosated thiol isomerases

Very few studies have evaluated the possibility that oxidative protein modifications affect the ability of thiol isomerases to modulate thrombus formation. The study of Bekendam et al. showed that endothelial NO was capable of nitrosating endogenous PDI, resulting in the formation of nitrosated PDI (SNO-PDI) (8). This study also showed that SNO-PDI was capable of inhibiting platelet aggregation and granule release (8). SNO-ERp5 and SNO-ERp57 also inhibited platelet aggregation. To assess whether SNO-PDI affected thrombus formation in vivo, SNO-PDI formed from incubating recombinant PDI with S-nitrosocysteine was infused into mice that were subsequently subjected to laser injury to induce thrombus formation. Infusion of SNO-PDI into mice inhibited both maximal platelet accumulation and fibrin generation. The fact that the SNO-PDI preparation, which was devoid of free S-nitrosocysteine after dialysis, inhibited thrombosis raised the possibility that PDI-SNO transfers NO to either platelet surface proteins or soluble guanylyl cyclase (Fig. 8). Previous studies have shown that extracellular PDI can transfer NO to surface proteins or intracellularly (129, 148, 175). Thus, in the resting state, nitrosation of PDI could both inhibit PDI oxidoreductase activity and transfer NO to vascular cells, thereby contributing to vascular quiescence. In the setting of thrombosis, this constitutive inhibitory mechanism could be lost (Fig. 8).

FIG. 8.

Model of redox regulation of vascular PDI function. Resting state: The quiescent endothelium generates nitric oxide, which nitrosates PDI. SNO-PDI can transnitrosate endothelial and platelet surface proteins such as integrins. It can also transfer nitric oxide into platelets, stimulating production of cGMP by sGC. Prothrombotic state. In the setting of thromboinflammation, PDI is released from endothelium and platelets. If released into an oxidizing environment, PDI can oxidize surface proteins on endothelium (such as TF) and on platelets (such as α_IIbβ₃). If released into a reduced environment, PDI can reduce proteins such as TSP1 and VTN. It is not yet clear whether oxidized PDI, reduced PDI, or both are released in the setting of thromboinflammation. Endothelial cells are depicted as beige rectangles. Platelets are depicted as blue ovals. sGC, soluble guanylyl cyclase; TF, tissue factor; TSP1, thrombospondin 1; VTN, vitronectin.

Concluding Remarks

Extracellular thiol isomerases encounter and are capable of reacting with an extensive array of oxidants in the vasculature. From the viewpoint of redox biology, vascular thiol isomerases act as a manager of oxidants, which they sort, process, and transmit to substrate proteins. A subset of oxidants bind thiol isomerases and irreversibly modify them. However, the interaction of other oxidants with thiol isomerases results in the formation of disulfide bonds, which can then be transferred to other binding partners, altering the activity of these substrates. Thiol isomerases can also mediate alternative post-translational modifications by transferring oxidant species, such as nitric oxide or glutathione, to substrates. In this manner, thiol isomerases are capable of initiating redox signaling pathways. Furthermore, oxidant modification of thiol isomerases is not limited to active site cysteines, but also elicits substantial changes in conformation throughout the protein. These redox-mediated conformational changes control the interaction of thiol isomerases with binding partners and can even affect chaperone activity (144, 163). In this regard, oxidoreductase and chaperone activities are both controlled by the effects that the redox environment has on the active site cysteines in the catalytic motif.

There is substantial evidence that redox stress promotes thrombosis, and we propose that extracellular thiol isomerases contribute to this process by promoting prothrombotic signaling in response to changes in redox environment. Thiol isomerases could transmit redox signals to modify disulfide bonding in substrates, thereby promoting thrombus formation. While much progress has been made in identifying vascular thiol isomerases that participate in thrombus formation and in developing thiol isomerase inhibitors that block thrombus formation, when viewed from the lens of redox biology, many fundamental questions about thiol isomerase function in thrombosis persist. What is the redox status (oxidized, reduced, or both) of thiol isomerases when secreted from endothelial cells and platelets during thrombosis? Is disulfide bond formation or cleavage (or both) critical for thrombus formation? Is there a redox-sensitive electron relay between thiol isomerase family members that functions in thrombus formation? Do thiol isomerases influence the thiol- and disulfide landscape of surface proteins on vascular cells in a redox-dependent manner? While there has been a focus on identifying a single substrate (e.g., α_IIbβ₃, TF, vitronectin, GPIbα) substantially responsible for the prothrombotic function of PDI, it is also possible that thiol isomerases modify a multitude of substrates modestly or have additional effects (e.g., oxidant transfer, redox-dependent chaperone functions) so as to render the vasculature more prothrombotic. Evaluation of such diffuse and widespread functions of thiol isomerases will require a systems approach to characterizing thiol isomerase function. The development of new probes capable of distinguishing different sulfur oxoforms could enable such evaluation. In addition, thiol isomerases could have additional functions in controlling post-translational modifications (nitrosation, glutathionylation) in response to changes in redox environment. Understanding the redox state of thiol isomerases during thrombosis and controlling thiol isomerase activity via their cysteine modification may offer a new strategy to regulate thrombotic complications of chronic diseases such as autoimmune inflammatory states, cancer, dyslipidemia, and diabetes mellitus.

Abbreviations Used

α_IIbβ₃

integrin α_IIbβ₃

ADAMTS13

a disintegrin and metalloproteinase with a thrombospondin type 1 motif, member 13

BCN

bicyclo[6.1.0]nonyne

BTD

benzothiazine-based carbon nucleophile

CD36

cluster of differentiation 36

cGMP

cyclic guanosine monophosphate

CHD

1,3-cyclohexanedione; dimedone

CXXA

cysteine-X-X-alanine (where X is any amino acid)

CXXC

cysteine-X-X-cysteine (where X is any amino acid)

DPS

dithiodipyridine

DTNB

dithiol nitrobenzoic acid; Ellman's reagent

E ₀

redox potential

ERO1

endoplasmic reticulum oxireductin 1

ERp5

endoplasmic reticulum resident protein 5; protein disulfide-isomerase A6

ERp57

endoplasmic reticulum resident protein 57; protein disulfide-isomerase A3

ERp72

endoplasmic reticulum resident protein 72; protein disulfide-isomerase A4

GPIbα

glycoprotein Ibα

HEK293

human embryonic kidney 293

IAM

iodoacetamide

KDEL

lysine–aspartate–glutamate–leucine

mBrB

monobromobimane

MMTS

methyl methanethiolsulfonates

NADPH

protonated nicotinamide adenine dinucleotide phosphate

NBD-Cl

4-chloro-7-nitrobenzofurazan

NEM

N-ethylmaleimide

NO

nitric oxide radical

OxLDL

oxidized low-density lipoprotein

PDI

protein disulfide isomerase

PRD

piperidine-based carbon nucleophile

PYD

pyrrolidine-based carbon nucleophile

S2S

cysteine succination

sGC

soluble guanylyl cyclase

SH

cysteine free thiol

SHg

mercurated cysteine

smFRET

single-molecule fluorescence resonance energy transfer

SN₂

nucleophilic substitution

SNO

cysteine nitrosation

SNO-ERp5

nitrosated ERp5

SNO-ERp57

nitrosated ERp57

SNO-PDI

nitrosated PDI

SO₂H

sulfinylation

SO₃H

sulfonylation

SOH

sulfenylation

S-R

cysteine alkylation

SS

disulfided cysteine

SSG

cysteine glutathionylation

TF

tissue factor

TMX

thioredoxin-related transmembrane protein

TSP1

thrombospondin 1

VTN

vitronectin

Authors' Contributions

M.Y. and R.F. wrote and edited the article.

Author Disclosure Statement

R.F. is a founder and consultant for PlateletDiagnostics. The interests of R.F. are reviewed and managed by the Beth Israel Deaconess Medical Center Office of Compliance and Business Conduct.

Funding Information

This work is supported by the National Institute of Health grants R35HL135775, U01HL143365, and T32HL007917 to R.F.