Introduction to approaches and tools for the evaluation of protein cysteine oxidation

Abstract

Oxidative modifications of cysteine thiols in cellular proteins are pivotal to the way signal-stimulated reactive oxygen species are sensed and elicit appropriate or sometimes pathological responses, but the dynamic and often transitory nature of these modifications offer a challenge to the investigator trying to identify such sites and the responses they elicit. A number of reagents and workflows have been developed to identify proteins undergoing oxidation and to query the timing, extent and location of such modifications, as described in this minireview. While no approach is perfect to capture all the redox information in a functioning cell, best practices described herein can enable considerable insights into the “redox world” of cells and organisms.

Introduction

Oxidatively modified cysteinyl residues in proteins play an important part role in the redox regulation of cell signaling and metabolism, and the identification of the sites and modifications that underlie these regulatory events is of great importance. Dynamic post-translational modifications (PTMs, see Box 1 for Glossary) regulate biological processes in many ways, with the most established being the phosphorylation cascades that result from the transmission of extracellular signals to trigger intracellular responses. Oxidation and reduction are in fact part of the regulatory network that includes phosphorylation PTMs as many proteins, including both phosphatases and kinases, are subject to redox control. Because of the labile nature of many of the oxidative PTMs, special attention must be given to the way in which samples are collected and analyzed in order to minimize artifacts and enhance reproducibility. This minireview focuses on introducing the non-specialist to important aspects of this research area, with inclusion of references that provide more detailed descriptions than can be included here.

Box 1. Glossary.

4-TP	4-Thiopyridone
ABD-F	4-(Aminosulfonyl)-7-fluoro-2,1,3-benzoxadiazole
AMS	4-Acetamido-4′-maleimidylstilbene-2,2′-disulfonic acid
BCN	Bicyclo[6.1.0]nonyne
DCP	2,4-(Dioxocyclohexyl)propoxy unit
DIGE	Difference gel electrophoresis
DNFB	1-Fluoro-2,4-dinitrobenzene
DNTP	2,4-Dinitrothiophenol
DPS	4,4′-Dithiodipyridine
DTNB	5,5′-Dithiobis(2-nitrobenzoic acid)
DTPA	Diethylenetriaminepentaacetic acid
DTT	1,4-Dithiothreitol
EDTA	Ethylenediaminetetraacetic acid
Grx	Glutaredoxin
HPLC	High-performance liquid chromatography
IAA	Iodoacetic acid
IAM	Iodoacetamide
ICAT	Isotope-coded affinity tags
MAL-PEG	Methoxypolyethylene glycol-maleimide
ME	2-mercaptoethanol
MMTS	S-Methyl methanethiosulfonate
MSBT	Methylsulfonyl benzothiazole
MSTP	4-(5-Methanesulfonyl-[1,2,3,4]tetrazol-1-yl)-phenol
NBD-Cl	7-Chloro-4-nitrobenzo-2-oxa-1,3-diazole, also known as 4-chloro-7-nitrobenzofurazan
NEM	N-ethylmaleimide
NTB	2-Nitro-5-thiobenzoic acid
NTSB	2-Nitro-5-thiosulfobenzoate
PAGE	Polyacrylamide gel electrophoresis
PROP	Purification of reversibly oxidized proteins
Prx	Peroxiredoxin
PTEN	Phosphatase and tensin homolog deleted from chromosome 10
PTM	Post-translational modification
RSNO	S-Nitrosothiol
SBD-F	7-Fluorobenzo-2-oxa-1,3-diazole-4-sulfonate
SDS	Sodium dodecylsulfate
TCEP	Tris(2-carboxyethyl)phosphine
Trx	Thioredoxin

While oxidative modifications of cysteine are the principal topic of this minireview, it is important to note that cysteine residues also exhibit reactivity toward electrophiles, yielding acylated or alkylated sidechains, and are involved in strong interactions with metals, which can either promote or suppress their sensitivity toward oxidation [1-3]. Indeed, more than 35 modifications on cysteine have been recognized to occur, thus these residues in proteins can be a “hotspot” for modification including oxidative modifications. We address here the detection and identification of oxidative modifications that are most observed and relevant to biology.

Oxidized cysteine species in proteins

As illustrated by the examples in Table 1, more than 15 oxidatively modified forms of cysteine are of relevance in biological systems and the chemical approaches used to query them, encompassing a range of oxidation states of sulfur (−1 to +4) and in some cases including nitrogen atoms. Disulfide bonds between cysteines are the most common oxidative modification in proteins, particularly when those proteins emerge through the secretory pathway of cells to become exported to the extracellular space. Generation of disulfide bonds in proteins can result from some of the intermediary cysteine modifications (Table 1) interacting with a cysteine thiol (Figure 1), perhaps representing a primary way in which disulfide bonds form in reversible redox switches in proteins. Thiol-disulfide interchange (also referred to as “relay”), discussed more below, leads to no net change in the oxidation states of the sulfurs involved, but a physical transfer of the disulfide bond between species.

Table 1.

Oxidative modifications of cysteine in proteins

Modification	Structure	Oxidation state	Modification	Structure	Oxidation state
None		−2	Sulfenyl halidea		0
Thiyl radical		−1	S-Nitrosothiol		0
Intramolecular disulfide		−1	Thiosulfinate		+1
Mixed (or intermolecular) disulfide		−1	Sulfinamide		+2
Persulfide (perthiol)		−1	Sulfinic acid		+2
Thiosulfate		−1	Thiosulfonate		+3
Sulfenic acid		0	Sulfonic acid		+4
Sulfenamide		0	Sulfonamide		+4

^aX represents chloride, bromide or thiocyanate.

Figure 1.

Thiol reactivity with modified cysteines to generate disulfides

The oxyacids of sulfur, sulfenic, sulfinic and sulfonic acids with one, two and three oxygens attached to the sulfur, respectively, represent sequentially more oxidized species (Table 1). Related modifications with nitrogen atoms are sulfenamide, sulfinamide and sulfonamide species. With sulfur at the same “0” oxidation state, sulfenic acids can readily interconvert to and from sulfenamides through condensation/hydrolysis reactions, and sulfenyl halides like sulfenyl chloride and sulfenyl bromide undergo hydrolysis to sulfenic acids (a reaction shared by sulfenyl thiocyanate, a related “pseudo” sulfenyl halide species) [1,4,5]. In the case of several of these modifications like nitrosothiols and persulfides, the moiety linked to the cysteine sulfur is transferrable, e.g. as NO⁺ from the R-SNO species (useful when S-nitrosoglutathione is used as an NO⁺ donor to proteins). Thus, many of these oxidative modifications represent dynamic PTMs within cells, and their reactivity must be taken into consideration with respect to our ability to observe, trap and/or analyze these oxidative modifications within cellular proteins. Moreover, reductive pathways in cells exist that can reverse most of the oxidative modifications below the +2 oxidation state of the sulfur, a reactivity that implies roles for such modifications as cysteine switches in metabolic, signaling and other proteins. Chemical reductants [particularly dithiothreitol (DTT) and tris(2-carboxyethyl)phosphine (TCEP)] are also potent agents that reverse oxidative modifications, but with little to no selectivity for the specific context of the modification, and are often used in workflows designed to query protein oxidation. As oxidative modifications of cysteine in proteins are highly dynamic, both in vivo and in vitro approaches to investigate the nature of modifications in proteins must be tailored to preserve as much of the native state information as possible.

Physicochemical properties of thiols

While an extensive discussion of the chemical and physical properties of cysteine that govern its reactivity are outside the scope of this minireview, it is important to mention the influence that distinct properties like pK_a and reduction potential (typically referred to as redox potential) have on cysteine reactivity (more extensive discussions of these subjects geared to the non-specialist is provided elsewhere [6]). The loss of a proton (H⁺) from cysteine thiols yields the negatively charged thiolate species that is far more nucleophilic and dominates the chemistry of these residues [7]. The parameter that informs on the propensity of a given cysteine to lose its proton is the pK_a, the pH value at which the populations of neutral thiol and negatively charged thiolate species are equal (i.e. where the RSH/RS⁻ ratio is 1). As an “unperturbed” cysteine pK_a is typically at or above pH 8, a relatively small proportion of these residues would be in their thiolate state at neutral pH (1/11 of the molecules at a pH one unit lower than the pK_a). On the other hand, if the pK_a of the cysteine is lower than the pH of the surrounding medium, the fraction of cysteine molecules in the thiolate form will be greater than 50%. In practical terms, equilibrium mixtures of these species can be calculated from known pK_a values and actual pH conditions using the Henderson–Hasselbalch equation [6,7]. It should be noted that biological thiols exhibit a wide range of pK_a values from about 3 to 11 [3,7], and that the thiols in protein cysteine persulfide species (-SSH) have lower pK_a values than the cysteine thiol [8-10].

The other parameter of great significance to cysteine reactivity is the redox potential (Eᵒ), a value (relative to the standard hydrogen cell, but typically adjusted to pH 7 for biochemical conditions, Eᵒ′) that allows calculation of the equilibrium mixture (K_eq) between two species (one oxidized and one reduced), related to the free energy of conversion between them (ΔG). Considering dithiol/disulfide equilibria that reflect 2 electron transfers between species, molecules that typically serve as reductants in the cell like thioredoxin tend to exhibit redox potentials around −230 mV or below [2], whereas molecules that tend to serve as oxidants have higher redox potentials. Important to this discussion of equilibrium-related parameters, however, is an understanding that cellular systems include many components that are rarely found in equilibrium since much of redox chemistry (and chemistry in general) in the cell is driven by reaction kinetics (i.e. just because the net direction of a reaction is governed by its equilibrium constant, there must still exist a facile pathway for conversion between two species for them equilibrate, and there are many competing reactions in cells).

Assessment of cysteine modifications in cellular and isolated proteins

Methods to assess the presence of free thiols (sulfhydryl groups) in proteins as well as their modified forms require special handling of the samples in a way that preserves the native redox status, minimizes rearrangements of thiols and disulfides, and limits adventitious oxidation during sample workup. In addition, denaturation of proteins may be necessary to fully expose the species of interest to enhance their reactivity toward the reagents being used. A detailed examination of the approaches and potential pitfalls encountered with the various reagents and treatments is outside the scope of this brief essay, but available elsewhere [1,11-14].

With respect to thiol-disulfide rearrangements, sulfhydryl groups in their more nucleophilic thiolate forms can attack one of the two sulfurs within the disulfide bond to initiate thiol-disulfide exchange, which is reversible (Figure 1, last reaction). Such reactions are inhibited by lowering the pH. Unless approaches are used to minimize such reactions, this “scrambling” of the disulfide partners can lead to mis-identification of the cysteine residues involved in the disulfide bonds of interest. This concept is also relevant when the disulfides are in per/poly-sulfide PTMs (sulfane sulfur modifications). Treatment or composition of buffers can also be an important way to minimize undesirable thiol oxidation occurring during sample handling, including the bubbling of buffers with nitrogen or argon before their use, and the addition of chelating agents like ethylenediaminetetraacetic acid (EDTA) or diethylenetriaminepentaacetic acid (DTPA) to minimize the metal-dependent oxidation of thiols by oxygen [11].

A number of approaches require the blocking (technically known as alkylation) of free thiols as an initial step to prevent thiol-disulfide interactions; where proteins are denatured as part of the process, it is best to include the alkylating agents before denaturation, as well as during the incubation with denaturant, as the thiol-disulfide reaction may be efficient enough under some conditions to outcompete the thiol alkylation (Box 2). For accurate detection of thiols and their oxidation products, the alkylation step(s) must be specific, efficient and irreversible. Several reagents are typically used in such protocols. Iodoacetamide (IAM) and iodoacetic acid (IAA) are both reactive with deprotonated cysteine residues (i.e. thiolates), thus reacting more quickly with most thiols under slightly alkaline conditions (typically at pH 8), although the negative charge of IAA can in some cases impede its reaction at specific sites. Maleimides including N-ethylmaleimide (NEM) are also commonly used alkylation reagents that react preferentially with thiolates through addition across the double bond of the maleimide, often more efficiently than IAM, especially when conducted at neutral or slightly acidic pH [7]. It should be noted, however, that alkylation of thiols by NEM is not in fact irreversible (as is IAA- or IAM-based alkylation, where iodide is displaced), leading to the possibility that the maleimide may migrate to other thiol groups in the sample [7]. Both NEM and IAM are uncharged and thus cell membrane permeable, whereas IAA cannot pass through membranes. Users should be aware of the potential for reactivity of these agents or their products with other amino acid sidechains, as well [7,15]. Another reagent that has been used to rapidly trap free thiols is methyl methanethiosulfonate (MMTS), but the product of this reaction on the protein is –SS-CH₃, a mixed disulfide. As noted above, such species are prone to thiol-disulfide exchange reactions and in fact the use of MMTS can promote rather than block the formation of intra- and inter-molecular disulfide bonds [16]. In addition, reaction of MMTS with a thiol to give the –SSCH₃ disulfide also produces a sulfinic acid, CH₃SO₂⁻, which may cause problems depending on the application. In particular, sulfinic acids are known to react with S-nitrosothiols [17]. This reagent should therefore be avoided in most applications. It should also be noted that reagents that react with thiols also react with the terminal sulfur of persulfides [10,18].

Summary.

• More than 15 oxidative modifications on protein cysteine residues have been detected experimentally in biological settings.

• Oxidative modifications are key to cellular signaling, but this information is challenging to capture without provoking artifacts that result from sample preparation and detection methods.

• Described herein are best practices, reagents and workflows that are designed to minimize post-lysis oxidation and thiol-disulfide “scrambling” and maximize the quality of the information on relevant redox modifications important to cellular processes.

An important point to note is that, while many in the field have long regarded IAA, IAM, NEM and MMTS as being thiol-specific reagents, they do in fact exhibit cross-reactivity with sulfenic acids, sometimes with reaction rates that rival their rates with thiols [19]. A further complication is that the “–SOR” products of sulfenic acid alkylation may be partially or fully reversed by reductants, further challenging the quantitative accuracy of the analytical methods that employ these chemicals. In an effort to avoid the artifacts of cross-reactivity with sulfenic acids, new workflows call for lower concentrations of IAM (100 μM) rather than the typically recommended mM concentrations based on earlier work by Weerapana et al. [20]. However, under these conditions the blocking of thiols takes considerably longer, which allows for introduction of artifacts from thiol-disulfide exchange reactions or post-lysis oxidation. Fortunately, recent studies with thiol-reactive heteroaromatic sulfones [21] have shown that these reagents do exhibit selectivity for sulfhydryl groups and are non-reactive with sulfenic acids [22], providing a better way to alkylate thiols without the sulfenic acid cross-reactivity exhibited by the more commonly used IAM, NEM and MMTS reagents. Among these, methylsulfonyl benzothiazole (MSBT) and 4-(5-methanesulfonyl-[1,2,3,4]tetrazol-1-yl)-phenol (MSTP) show great promise as replacements for IAM and NEM in such analytical procedures.

In analyses of cell-based systems, the goal of stabilizing or “freezing” the redox state of all cysteine-containing proteins is critical to capturing the information about native modifications, yet this is quite difficult to fully achieve. Flash freezing can be used to rapidly quench samples and temporarily preserve chemical species, but a second approach must ultimately be included to stabilize or trap redox modifications once samples are thawed. Typical approaches include rapid acidification of samples with trichloroacetic acid (TCA) or brief pretreatment of intact cells with high concentrations of cell permeable alkylating agents like NEM before lysis. TCA added to samples at 10–15% has the added advantage of serving as a potent denaturant and is more reliable for preserving cysteine redox state than some other acids [11]. The use of NEM pretreatment is considered by many researchers in the field to be the ideal approach given its effectiveness as an alkylating agent combined with its ability to alkylate free thiol groups under conditions where cellular compartments are not mixed [23,24]. It should be noted, however, that the effectiveness of NEM blocking across all cellular molecules and organelles, including the kinetics of its penetration into cells, is not well investigated, and the influence of the high intracellular concentrations of reduced glutathione during this trapping process would be expected to delay and perhaps even bias the in situ equilibria between species that are being investigated [7].

Workflows for the characterization of redox-sensitive or -modified sites in proteins may also involve the denaturation and/or reduction of proteins as part of the analytical procedure. Denaturants in particular must be chosen to be compatible with subsequent steps. Typical denaturing conditions include the use of 6 M guanidine hydrochloride, 8 M urea (which must be made fresh) or 5% sodium dodecyl sulfate (SDS). Organic solvents like acetone are also denaturing but also lead to protein precipitation, sometimes forming pellets that are challenging to resolubilize.

In the case of reduction, the most widely used reagents are 2-mercaptoethanol (ME), DTT and TCEP. ME is often used at high concentrations in SDS-polyacrylamide gel electrophoresis (PAGE), but is a relatively weak reductant and can form mixed-disulfide bonds with proteins, so may not be adequate for most procedures. DTT is a much stronger reductant that at even slight excesses can efficiently reduce many oxidized protein sites, although when used in excess it must typically be removed before subsequent alkylation or capture steps since it also possesses reactive thiols. TCEP relies on phosphine chemistry to reduce disulfides and other reversibly oxidized cysteine sites, although it exhibits some unexplained instability in neutral pH phosphate buffers [11]. Its addition to solutions also makes them quite acidic if not strongly buffered or the pH carefully adjusted. While sulfhydryl groups are not present in TCEP, it does still cross-react with alkylating reagents like NEM and IAM and must also typically be removed prior to subsequent labeling steps. One commercially available form of TCEP linked to agarose beads (Pierce™ Immobilized TCEP) can be helpful for the short-term storage of pre-reduced proteins from which the soluble DTT or TCEP has already been removed using gel filtration. Another strongly reducing reagent that can be used to reduce disulfides is sodium borohydride, although this reagent can also reduce other groups in proteins (like keto groups). A major advantage of its use, however, is that it can be destroyed by acid treatment rather than requiring removal by gel filtration or precipitation like the other reductants described [11]. Other reductants including low redox potential reductase proteins like thioredoxin (Trx) and glutaredoxin (Grx) can also be used with care to take into account the selectivity toward specific sites or modifications that they impart.

Spectroscopic methods for protein thiol, disulfide, sulfenic acid and persulfide detection

Quantification of thiol oxidation in pure proteins, particularly when there are only a few cysteine residues within the protein, often relies on a reagent, 5,5′-dithiobis(2-nitrobenzoic acid) (DTNB), which was introduced by Ellman in the 1950s as a sensitive reagent for measuring the presence of thiols in proteins [25]. This reagent reacts with accessible thiols to undergo thiol-disulfide exchange (Figure 2A) that results in half of the molecule forming a disulfide bond with the thiol, and the other half being released into solution. This reaction yields a highly stabilized leaving group, 2-nitro-5-thiobenzoate (abbreviated NTB or TNB) with a strong yellow color at pH 7.3 or higher where the dianion is stabilized [Figure 2A; ε₄₁₂ (NTB) = 14,150 M⁻¹ cm⁻¹] [26]. Another reagent that can be used in this way is 4,4′-dithiodipyridine (DPS), which reacts with thiols in a parallel manner, forming 4-thiopyridone (4-TP) (Figure 2B) [11,27]. Besides its higher extinction coefficient [ε₃₂₄ ( 4-TP) = 21,400 M⁻¹ cm⁻¹], an advantage of using this reagent is its reactivity at low pH even down to pH 3. As one of the issues with using DTNB at pH values above 7 (needed to fully ionize NTB to its yellow dianion) is its base-mediated hydrolysis to slowly release NTB in the absence of thiols, this can offer a significant advantage of DPS over the use of DTNB. A potential disadvantage of using DPS is the shorter wavelength maximum of 4-TP (324 nm), which may be subject to spectral interference from other sample components. Note that both of these reagents are also reactive with thiol-containing reductants, therefore pre-treatment with DTT or thiol-based biological reductants to ensure that a protein is present in its fully reduced state will necessitate removal of that reductant prior to the spectral experiment. With both DTNB and DPS, the labeled protein product can also be isolated, and subsequent treatment with DTT to release the NTB or linked DPS product can provide another opportunity for quantification (Figure 2A).

Figure 2. Key reactions in spectral analysis of cysteinyl species.

(A and B) Free thiols in proteins can be detected by use of DTNB (A) or 4-DPS (B), which release one strongly absorbing molecule, NTB or 4-TP, respectively, per thiol. (C) An assay for disulfide bonds using a derivitive of DTNB, NTSB, is conducted concomitant with sulfitolysis of the disulfide bond, releasing one NTB per disulfide bond. (D) Solutions of NTB mixed with cysteine sulfenic acid-containing proteins form mixed disulfides, decreasing the A₄₁₂ of the solution. (E) Both thiol and sulfenic acid groups react with NBD-Cl to yield the respective thioether or sulfoxide products with distinct spectral features and masses. (F) Persulfides (and polysulfides) react with DNFB to form the disulfide-linked product (or trisulfide, tetrasulfide, etc for polysulfides); products that are reducible by DTT to release DNTP that is spectroscopically detectable under strongly alkaline conditions [28]. See text for additional references and details.

As alluded to above, the application of DTNB for quantification of thiols requires attention to some of the competing chemical reactions and sensitivities of the reagent and the NTB product, and proper controls must be included. For example, in addition to the tendency of DTNB to be hydrolyzed at high pH (above 7), the released NTB can also air oxidize back to DTNB, particularly when the reaction requires many minutes to be completed due to only partial exposure of the thiols (fully inaccessible thiols do not react with DTNB). Denaturation of the protein of interest with guanidine hydrochloride is typically used to expose all thiols on a protein, but the extinction coefficient of the released NTB changes modestly in its presence. Finally, not all proteins (e.g. those with CXXC-containing reactive centers) become fully modified by NTB when treated with DTNB; instead, they may mimic the catalytic process by first reacting with DTNB at the most nucleophilic Cys to yield the NTB mixed disulfide, then undergoing attack by the second (i.e. “resolving”) Cys thiol to release the protein-bound NTB and form the normal, oxidized disulfide. In these cases, the stoichiometry of NTB release remains the same, one per thiol group, but no NTB is released from the isolated protein upon DTT treatment. With this in mind, DTNB can be used as a tool for revealing intramolecular and intermolecular pathways of electron transfers through and between protein redox centers using strategically designed mutants, kinetic profiles of the release of NTB and characterization of the products formed [29,30].

An important caveat to mention in the use of DTNB to quantify thiols is that the NTB released from these reagents will rapidly react with sulfenic acids (Figure 2D), thus decreasing the absorbance at 412 nm and obfuscating the actual thiol content if both thiols and sulfenic acids are present; one way to address this is to pretreat such proteins with dimedone to block the sulfenic acids, or to fully reduce the protein with DTT or TCEP prior to the DTNB assay in cases where the total reduced and reversibly oxidized cysteine content is being measured.

Disulfide bonds in proteins can be quantitatively assessed in an approach that is similar to the DTNB assay, except that the reagent, 2-nitro-5-thiosulfobenzoate (NTSB), is prepared in advance by treatment of the DTNB with sodium sulfite, and the protein is subjected to sulfitolysis of the disulfide bonds during the assay [31] (Figure 2C). This assay can be highly quantitative for disulfide bonds but is complicated by the need to prepare the reagent in advance through oxygen-bubbling of the DTNB-sulfite mixture, as well as the sensitivity of the assay mixture to room light [32].

Two spectroscopy-based assays have been developed for sulfenic acids, with the first, as mentioned above, relying on their reactivity with NTB (Figure 2D). NTB as the assay reagent is prepared by mixing equimolar amounts of DTT and DTNB (ensuring no excess of either remains), then the rapid reaction of sulfenic acids with NTB leads to a decrease in absorbance at 412 nm [33,34]. This has been successfully used in the demonstration of the single sulfenic acid generated by peroxide in bacterial AhpC lacking the resolving cysteine [34,35], and in assessing sulfenic acid generation at the free thiol present in human serum albumin [36]. The second assay relies on reactivity of both thiols and sulfenic acids with the electrophilic reagent 7-chloro-4-nitrobenzo-2-oxa-1,3-diazole (NBD-Cl), where the products formed from each (a thioether in the case of thiols and sulfoxide in the case of sulfenic acids) exhibit distinct spectra, with maximal absorbance at 420 nm for the thioether and ~347 nm for the sulfoxide (Figure 2E) [34,37]. These products can be isolated and confirmed by mass spectrometry to demonstrate the additional 16 atomic mass units for the sulfoxide product. As a word of caution, it should be noted that NBD-Cl can also react with other nucleophiles in proteins including lysine and tyrosine, although fortunately such adducts have spectral properties distinct from the cysteine-based products [34,36].

Reactivity of persulfides (and polysulfides) toward thiol alkylating agents has also been exploited in spectral assays to assess their presence in proteins and small molecules. As the reaction products of reagents like 1-fluoro-2,4-dinitrobenzene (abbreviated DNFB or FDNB) with persulfides are disulfides rather than the thioethers generated from the reaction with protein cysteine thiols, the subsequent release of the detectable modifying group [2,4-dinitrothiophenol (DNTP) in the case of DNFB] upon addition of DTT provides a spectroscopic method to evaluate their presence (Figure 2F) [28]. Additional fluorescent probes to detect these sulfane sulfur species (persulfide, polysulfide and elemental sulfur) are continuing to be developed [38].

Fluorescent reagents are also available for use as thiol reagents, a number of which employ iodoacetamide and maleimide groups in common with IAM and NEM, or other groups with distinct chemistries (e.g. the benzofurazans SBD-F and ABD-F) [11,39]. Popular examples include the maleimide-based CyDyes (e.g. Cy2, Cy3 and Cy5) [40] and the IAM derivative BODIPY FL C1-IA [41] that are used in 2D gel applications, including “redoxDIGE” [42,43]. Another, monobromobimane (mBBr), is only weakly fluorescent until it reacts with a thiol to generate the thioether; it has found utility in 2D gel analyses [44] as well as identification and analysis of low molecular weight thiols from prokaryotes, isolated and quantified via HPLC [45-47]. It should be noted that, like NEM, IAM and NBD-Cl, many of these reagents are likely to exhibit cross-reactivity with sulfenic acids, as well [19], which could be expected to affect the proper interpretation of such analyses. As an alternative to or in conjunction with spectroscopy, redox modifications in recombinant proteins can be identified and quantified by mass spectrometry using either intact proteins or proteolytic peptides. This is particularly important for modifications that are relatively stable and are less amenable to methods like those described above (e.g. thiosulfates, thiosulfinates, thiosulfonates, sulfonic acids, sulfinamides and sulfonamides). For intact protein analysis, special salt- and detergent-free conditions must be used as well as significant amounts of protein. The advantage is that one could maintain the protein in its native folded state, increasing the stability of labile oxidative modifications. Analysis of proteolytic digests allows for further assignment of oxidative modifications to specific cysteines in proteins, with the caveat that in the absence of labeling reagents, this is limited to analysis of stable modifications (e.g. sulfinic acid, disulfides). Mass spectrometry analysis of intact proteins has been essential to assess the kinetics of redox transformations in proteins and to develop workflows for redox proteomics studies [14].

Gel-based methods to assess protein redox modifications

There are a number of ways in which disulfide bonds between two cysteinyl residues on the same or different proteins can be observed and potentially measured quantitatively or semi-quantitatively in order to assess the oxidized-to-reduced ratio (i.e. the redox status or redox poise) of the population of molecules of a given protein. In purified proteins, reducing and non-reducing SDS polyacrylamide gels stained for protein often give a straightforward way to assess the relative amounts of reduced and oxidized forms of the proteins and can be readily interpreted as long as only two or a small number of cysteine residues are present. Within a mixture, a protein that can be specifically detected with antibodies can also be assessed for redox status by gel methods that include detection by Western blot. When antibodies are used for detection with non-reducing gels, however, proper controls must be included to evaluate the extent to which the sensitivity of the detection method differs between oxidized and reduced (and/or chemically modified) forms of the protein.

The simplest way to tell the difference between oxidized and reduced forms of proteins that form a single intramolecular disulfide bond is to block free thiols with alkylating agents like NEM or IAM, preferably added in advance of the SDS-containing sample buffer, and analyze the samples on non-reducing gels (recommendations for avoiding artifacts are noted in Box 1). Using such an approach, small proteins like Trx and Grx (~9000–12,000 Da, respectively) with local disulfide bonds at their CXXC active site may exhibit a detectable shift in migration to a lower apparent MW on non-reducing gels relative to the reduced protein [48]. This also works to track oxidation with larger proteins like lipid phosphatase PTEN (~47 Da) in which more distant Cys residues become linked in the disulfide (Cys71 and Cys 124) [49]. In large, multidomain proteins with multiple redox centers, engineered constructs with disulfide bonds between domains (mimicking intermediates in the electron transfer pathways between domains) have been generated that exhibit large shifts to higher apparent molecular weight on non-reducing gels, as shown for bacterial Trx reductase [50] and the distantly related reductase protein AhpF [29]. To amplify the difference in migration of oxidized and reduced protein species on SDS gels, even in cases with multiple oxidizable redox centers (as in Trx2 of Escherichia coli), a common approach has been to add a larger thiol blocking reagent than NEM, shifting the reduced and blocked protein to a higher apparent MW than the oxidized protein, using such reagents as 4-acetamido-4′-maleimidylstilbene-2,2′-disulfonic acid (AMS, adding 490 Da per thiol to the protein mass) [51,52] and methoxypolyethylene glycol-maleimide (MAL-PEG, adding 5000 Da per thiol) [53].

Other gel-based approaches to detect disulfide bonds and potentially other oxidative modifications at cysteines use a sequential process, blocking reduced thiols first, then alkylating with a second type of blocker with unique properties after reducing reversibly oxidized thiols with general reductants such as DTT or TCEP (Figure 3; see below for more about “tag-switch” methods designed to be selective for particular modifications). Several gel-based methods have been developed to track the redox status of human Trx1 and Trx2, for example [48,54,55]. For this purpose, Merrill et al. adapted an approach that bases the separation of individual redox forms in the gel on the difference in charge imparted by initially labeling cells/lysates with IAA in a urea-containing buffer, then resuspending the acetone-precipitated proteins in urea buffer containing DTT, then excess IAM. Separation of the labeled products by urea-PAGE followed by a Western blot for Trx, with comparison to appropriate mobility standards, then gives a readout of the extent of oxidation (up to two disulfides) within even the five Cys-containing human Trx1 protein [55]. Holmgren and colleagues reversed the order of alkylating agents, with IAM used before IAA, due to the higher rate of alkylation by IAM [56].

Figure 3. Approaches for detecting reversibly oxidized cysteine residues using thiol blocking and modification-selective reduction.

Following the blocking of free thiols with thiol-selective reagents (step 1), general reductants (step 2a) can be added to reveal new thiols that can then be labeled with a second alkylating reagent. Alternatively, the second step (step 2b) may involve addition of a modification-selective reducing agent like Grx (for glutathionylated cysteines), ascorbate (for nitrosothiols) or arsenite (for sulfenic acids), followed by addition of the second alkylating reagent. The thiol-selective reagent in step 2a or 2b can be tagged with biotin, fluorescent dyes, or other reporters depending on the downstream detection method. For persulfides (2c), two methods have emerged that use either (a) MSBT as the first thiol reagent and CN-biotin for step 2c, or (b) NBD-Cl as the first thiol (and sulfenic acid) reagent and a dimedone-like reagent for the second labeling step. See text for references and details.

When proteins with reactive, modified forms of cysteine (particularly sulfenic acid) interact with other proteins or domains in transient or stable complexes, disulfide bonds can form if thiols are present (Figure 1), with potential effects including the stabilization of complexes and/or alternative conformations. With disulfide bonds between two proteins or between subunits of an oligomer, the shift in apparent molecular weight of the covalent complex on non-reducing SDS-PAGE approximately reflects the combined molecular weights of the two polypeptide partners. This can be observed as multiple bands at higher apparent molecular weight when there are multiple partners forming such bonds under oxidizing conditions, detectable by Western blot for the protein(s) of interest; it is important to also assess such samples under disulfide bond reducing conditions to ensure that the “cross-link” is reversible (Box 2). Such a shift reporting on multisubunit (i.e. oligomeric) proteins that form covalent dimers in their oxidized form can be readily used to track redox status, as is often implemented for naturally substrate-oxidized peroxiredoxins of the “typical 2-Cys” or Prx1-like class [23,57,58]. This phenomenon of intermolecular disulfide bond formation can also be queried across a proteome by using a 2D gel approach whereby a sample is first resolved under non-reducing conditions, then the gel slice from that lane, soaked in a reductant-containing buffer, is placed atop a second gel and resolved under reducing conditions. In this case, proteins involved in disulfide bonds between polypeptides appear as “off-diagonal” spots on the gel, exhibiting a higher apparent molecular weight in the first dimension than in the second [24,44,59,60]. Because of the influence of disulfide bond formation within a polypeptide (i.e an intramolecular disulfide) on migration in polyacrylamide gels noted above, such oxidized proteins may also be observable on the other side of the diagonal (i.e. exhibiting a lower apparent molecular weight in the first dimension than in the second).

Sample treatments that take advantage of fluorescent thiol labeling agents combined with the separation capabilities of PAGE are another way in which proteomic analyses tracking redox status can be conducted, in both 1D and 2D gel formats. Such multistep labeling approaches are described further in the next section.

Multistep workflows and tag-switch methods

A number of approaches to evaluate reversibly oxidized cysteines in proteins rely on multi-step processes (Figure 3) whereby free thiols are initially trapped by alkylation, then either general or selective methods that produce new thiols at the sites of interest are followed by secondary labeling with alkylating agents (so-called tag-switch methods) or enrichment of proteins with nascent thiols by chromatography on activated thiol sepharose columns (referred to as PROP, purification of reversibly oxidized proteins [61]). These approaches can be used to generate samples for analysis by SDS-PAGE or by mass spectrometry, or for imaging-based analyses where one or both tags are fluorescent (e.g. through imaging of labeled proteins on gels, or cell- or tissue-based imaging approaches by microscopy or flow cytometry). Differential cysteine labeling in paired samples to compare protein amounts was implemented in the 1990s for analysis either by 2D gels (DIGE, e.g. using the CyDyes) [62] or mass spectrometry [63]. In the latter case, iodoacetylated derivatives of biotin with distinct masses due to incorporation of different stable isotopes, known as isotope-coded affinity tags (ICAT), were developed and applied to samples treated with reducing agents to label all cysteines. Recognizing that the cysteine labeling will be affected by oxidative modifications in non-reduced samples, workflows following the general protocol of Figure 3 (steps 1 and 2a) were combined with the ICAT technology and MS to report on redox status of each cysteine-containing protein, called OxICAT [64]. Gel imaging or MS-based approaches can all be used in such workflows that incorporate a detectable and/or enrichment-friendly tag.

The double alkylation strategies used in these tag switch methods are similar to the IAA/IAM tagging strategy described above for use with SDS-PAGE. In all of these procedures, the first step, the blocking of thiols by an alkylating agent, must be selective, efficient and complete. Removal (or inactivation) of the initial alkylating agent must also be effective without significant loss of sample in order to ensure that the next step of reduction does not incorporate more of the initial reagent. Strategies for recovery of the proteins free of the initial alkylating agent typically involve small gel filtration columns (gravity or spin columns) or addition of a protein precipitant. As mentioned, if the second step of labeling includes general reductants like DTT or TCEP, many of the reversibly oxidized modifications will be reversed, allowing for a second alkylating agent to be incorporated into such sites (Figure 3). Such a strategy that incorporates a biotinylated alkylating agent in the second step of alkylation is generally referred to as a biotin or tag-switch assay.

An early example of a biotin switch assay was introduced by Jaffrey and Snyder to reveal RSNO sites using a reportedly nitrosothiol-selective reagent, ascorbate, as the reductant of choice in the second step (Figure 3) [65]. This assay has been widely used in the field to reveal such sites, although issues of reproducibility and selectivity have been noted. In particular, when revisiting aspects of the conditions used in such procedures, Gladwin and coworkers found that “contaminating” copper is required for efficient and reproducible reduction of nitrosothiols by ascorbate [66]. Ascorbate also has some capacity to reduce moieties other than nitrosothiols, such as sulfenic acids, in certain contexts [67]. In another approach, sodium arsenite has been utilized as the selective reductant of sulfenic acids (Figure 3, lower right) [68]; a concern with this approach is the need for denaturation in order to efficiently alkylate the initial free thiols and enable access of the arsenite to sulfenic acid-containing sites, and the instability typically observed for such meta-stable sulfenic acid species, particularly outside of the protein environments that may stabilize them. Another selective reductant, glutaredoxin, has been employed in a parallel way to reveal cysteinyl sites modified through a disulfide bond with glutathione [69].

Methods to evaluate persulfide (and potentially polysulfide) sites in proteins have been a challenge to develop, in part due to the comparable reactivity of the terminal sulfur in per/polysulfides and the sulfur of thiols with alkylating agents. Earlier methods developed by Filipovic, Xian and others [70,71] have been subsequently replaced by a method that employs NBD-Cl as the initial alkylator of thiol, persulfide and sulfenic acid species, then selective alkylation of only the persulfide product with 1,3-cyclohexadione-containing derivatives like dimedone (Figure 3) [72]. A somewhat more direct method introduced by Nagy and colleagues proposed the use of biotin-iodoacetamide as the initial blocking agent, affinity capture with a streptavidin matrix, then elution using DTT toward which only the persulfide or polysulfide products should be susceptible [73]. Again, cross-reactivity of the iodoacetamide with sulfenic acid and partial or full reversibility of the ensuing product could create false positives among the persulfidated and polysulfidated proteins identified by such methods.

When a workflow for a multistep procedure for identifying modifications avoids pitfalls like incomplete alkylation in the first step and lack of chemoselectivity of the reductant for the targeted modification, and the reliability of the method has been confirmed by independent means, these methods may be quite powerful for obtaining proteome-wide information. Even in the best case, though, with a minimum of 15 possible reversible modifications at cysteine sites, there should be an understanding that there may always be some false positive (as well as false negative) data among the findings. Performing thorough validation of selectivity and yield at each step of the workflow and using a number of replicates (both biological and technical – multiple repeated runs of the same original sample) is important in all cases to understand more about the variability of the results, as well.

Direct trapping of cysteine oxidative modifications

As noted above, several thiol-reactive compounds demonstrate cross-reactivity with protein sulfenic acids. However, the chemical attributes of the sulfur and oxygen within the sulfenic acid moiety (-SOH), to some degree shared by sulfenyl halides and sulfenyl amides (i.e., sulfenamides) but distinct from thiols, can be harnessed to enable selective labeling of these groups in vitro and in vivo. As a number of nucleophilic agents including dimedone are reactive with sulfenic acids but not thiols [74], our group and others have taken advantage of this selective reactivity to develop and implement reagents and workflows since the late 1990s to identify proteins undergoing oxidation and to query the timing, extent and location of such modifications [75-82]. For more detailed descriptions of the chemistry and covalent targeting of sulfenic acids in proteins, the reader is referred elsewhere [13,83].

Based on the core 1,3-cyclohexadione reactive group of dimedone [74], our group and others have created sets of reagents that incorporate reporter or affinity groups into proteins (Figure 4) [76,77,80,81,84]. Versions of these reagents that incorporate either alkyne or azide groups with the reactive core, available for subsequent incorporation of detectable or affinity-mediated fragments via click chemistry, have also been generated [78,79,85,86]. A recent report has also highlighted the selectivity of several of the reagents, DCP-Rho1 and DCP-NEt₂C, toward labeling of mitochondria-localized sulfenylated proteins [82]. In addition to these 1,3-diketone bearing reagents, strained alkynes and alkenes have been used to selectively trap and label sulfenic acids [74,87], including bicyclo[6.1.0]nonyne (BCN) [88]. While BCN was also shown to react with persulfidated proteins, the reaction products can be distinguished by mass spectrometry (−16 Da mass difference relative to the adduct with protein sulfenic acids). Both norbornene and trans-cyclooctenol-derived probes have been used to trap protein sulfenic acids in live cells [89,90]. Many of these reagents are commercially available and have been pivotal in the identification and characterization of sulfenylated proteins involved in many cellular processes [91-97].

Figure 4. Chemical probes for selective labeling of cysteine sulfenic acids.

(A and B) Sulfenic acids react with 1,3-cyclohexadione derivatives to form thioethers (A), or with strained alkyne reagents like bicyclononyne (BCN) to form the adduct that incorporates the oxygen (B). (C) Biotin-linked reagents are used for affinity enrichment of tagged proteins through 1,3-cyclohexadione, 1,3-cyclopentadione, or BCN reactive groups. (D) Fluorescent groups can be incorporated into sulfenic acid-reactive reagents for imaging applications, including a recently developed dimedone-based reagent, F-DiNAP, which enables ratiometric imaging of sulfenic acids in live cells [17], (E) Sulfenic acid-reactive reagents with alkyne or azide groups allow for tag addition post labeling using the click reaction. Among those depicted, the recently developed benzothiazine-based probe, BTD, exhibits enhanced reactivity toward sulfenic acids relative to the 1,3-cyclohexadione based probes [86].

On the other hand, the detection of sulfinic acids using direct labeling by reagents showing appropriate chemoselectivity has been quite challenging. While stability of products generated with sulfinic acid using aryl-nitroso reagents was less than ideal [98], recent development of a diazene probe has enabled a proteome-wide exploration of sulfinic acid sites that have revealed additional substrates for the sulfinic acid-reducing protein sulfiredoxin [99]. In another approach, the reactivity of sulfinic acids with nitrosothiols was exploited to detect both sulfinylated and nitrosated proteins [17].

Conclusions

Redox processes play an important role in cell signaling and metabolism through the direct influence of transcriptional control mechanisms or specific signaling cascades that translate extracellular signals to intracellular responses. The sulfhydryl group of protein cysteinyl residues constitutes the chemical focal point for diverse redox chemical transformations that mediate these biological processes. This minireview summarizes work over the past ~30 years on methods to describe the nature, location and timing of chemical modifications that mediate redox control of biology in cysteine-containing proteins. A combination of improved biochemical workflows to handle and preserve redox sensitive species, development of specific molecular probes that can be either spectroscopically or immunochemically detected, and improved analytical (mass spectrometric) and genetic methods provide the basis of our current understanding of redox-mediated chemical biology. Despite these significant advances, much more work (e.g. shorter, easier workflows and more specific chemical traps) remains to better understand thiol-mediated redox control of biological processes.

Box 2. Applying best practices in sample preparation for thiol-containing proteins and their oxidation products.

• Block all free thiols before and during sample preparation where possible (including before and during denaturation) using thiol-selective reagents to suppress (i) thiol-disulfide exchange reactions, (ii) adventious oxidation of free thiols and (iii) migration of moieties (e.g. NO⁺ of RSNO) present on modified cysteines.

• More acidic conditions disfavor thiol-disulfide exchange reactions that rearrange (i.e. scramble locations of) the existing disulfide linkages.

• Pretreatment of cells with NEM prior to lysis may minimize redox perturbations in some proteins like peroxiredoxins, but trapping of all thiols cannot be relied on as rapidly and fully effective across all thiol-bearing molecules and results will undoubtedly vary with the target of interest and with cell type.

• Caution is urged as traditionally used thiol blockers like NEM, IAM and MMTS have been shown to cross-react with sulfenic acids. Thiol-selective alkylation reagents lacking sulfenic acid cross-reactivity include MSBT and MSTP. MMTS is also not an ideal thiol blocker on the basis of its generation of a disulfide-containing product that can promote introduction of or scrambling to yield non-native disulfides.

• If samples with and without chemical reductants (e.g. DTT or ME) are analyzed on the same polyacrylamide gel, it is best to leave an empty lane between them as the small molecule reductant readily diffuses between lanes.

• Where possible, use multiple complementary approaches to evaluate and validate the redox species and pathways uncovered by the analyses.

Abbreviations

IAA

iodoacetic acid

IAM

iodoacetamide

MMTS

methyl methanethiosulfonate

NEM

N-ethylmaleimide