Prodrugs of Persulfides,Sulfur Dioxide,and Carbon Disulfide: Important Tools for Studying Sulfur Signaling at Various Oxidation States

Introduction

Sulfur signaling is implicated in a large number of physiological and pathological processes (27, 34, 40 –43, 64, 68, 69, 73, 92, 93). Early on, attention in the field of sulfur signaling was largely focused on hydrogen sulfide (H₂S), which is often referred to as the third gasotransmitter with some disagreement (85, 89). Gasotransmitter is generally defined as endogenously produced gaseous signaling molecules, although with some debate (62). Previously, donors of various sulfur species at different oxidation states were sometimes collectively referred to as precursors of H₂S (41). However, in recent years, researchers have increasingly realized that sulfur species at different oxidation states might have their distinct biological functions due to their difference in chemical properties (6, 60, 64). For example, H₂S and RSSH are excellent reducing agents and play important roles in maintaining the intracellular redox balance (32). Further, sulfur species are known to chelate transition metals, and play roles in regulating enzyme activities and functions (6). In addition, persulfides are known to react with the cysteine residue in a protein, leading to persulfidation, which can in turn regulate enzyme functions and/or offer protection against oxidative damage (32, 60).

The involvement of multiple species and the diversity of the chemistry in sulfur signaling are keys to its pleiotropic effects in anti-inflammation, vasodilation, cytoprotection, metabolic rate regulation, among others. As if the involvement of multiple species is not complex enough, an added difficulty in studying sulfur signaling is the rapid interconversion of the various sulfur species under physiological conditions. Such conversions probably are also dependent on the redox state of the intracellular environment because many such reactions are redox reactions (83). All these factors convolute the study of the detailed mechanistic issues in sulfur signaling, complicate the interpretation of results, and lead to sometimes seemingly contradictory results and conclusions. All these complicating factors also mean that donors of various species with well-defined chemistry, stability, release rates, and triggering mechanisms are needed for further studies and for “apple to apple” comparisons of data from various donors (7, 30, 58, 71, 78, 98). One indeed needs to examine the specific sulfur species involved, the cellular conditions, and the release kinetics in interpreting the mechanistic meaning of the relevant biological responses. Another factor that needs to be considered is the fast exchange among various sulfur species at different oxidation states. For example, H₂S (oxidation state: −2) can be oxidized to persulfide (oxidation state: −1), then to sulfite (oxidation state: +4), and finally to sulfate (oxidation state: +6) (26). Reduction of some sulfur species such as disulfide can also happen very quickly (1, 91). In this special issue, there are other manuscripts that examine various aspects of sulfur signaling and redox chemistry. This review specifically focuses on donors of persulfide, sulfur dioxide (SO₂), and carbon disulfide (CS₂).

Persulfide Species: Endogenous Production, Physiological Activities, and Their Donors

Persulfide species R-SS_nH (n ≥ 1) play very important physiological roles (20). These persulfide species can be formed through enzyme catalysis or chemical reactions. For example, from cysteine, Cys-S-S-H can be formed by cystathionine-β-synthase or cystathionine-γ-lyase catalysis. Cys-S-S-H can go through disproportionation reaction to form Cys-S-S-S-Cys and H₂S, which could generate more complex persulfide species and hydrogen polysulfide through thiol exchange. Glutathione (GSH) can also react with Cys-S-S-H to produce glutathione persulfide (GSSH), which can go through the same process as Cys-S-S-H. Glutathione reductase can also participate in this process by reducing G-S-S-Sn-G (n ≥ 1) to G-S-S_n-H. 3-Mercaptopyruvate sulfurtransferase is also found to be able to generate persulfide species (35). In mouse organs such as liver and heart, the concentration of Cys-S-S-H and GSSH has reported to be in the range 50–100 and 1–5 μM, respectively (20).

On appearance, sometimes hydrogen polysulfides and persulfide (RS_nH, n ≥ 2) seem to have similar physiological effects to H₂S. For example, donors that deliver pure persulfide species and pure H₂S both were reported to exhibit cytoprotective effects from oxidative damage and cardioprotective effects in a murine model of myocardial ischemia-reperfusion (MI/R) (41, 99). However, often hydrogen polysulfides or persulfides exhibit greater potency than H₂S. For example, hydrogen polysulfides were found to be able to induce Ca²⁺ influx by activating transient receptor potential A1 channels in rat astrocytes and are 320 times more potent than H₂S (36). H₂S₂ and GSSH have been reported to directly modify glyceraldehyde 3-phosphate dehydrogenase (GAPDH) activity through cysteine persulfidation while H₂S itself incapable of doing so (22, 94). Such results are understandable because the oxidation state of H₂S does not allow it to directly persulfidate a thiol group; H₂S would have to be oxidized first before doing so. Therefore, in a persulfidation-related signaling process, H₂S might simply act as a precursor molecule. All these findings further state the importance of examining the roles of oxidation state of sulfur species in sulfur signaling. As a result, delivering donors at the right oxidation state for mechanistic studies is an important issue. Below, we discuss the various available donors of persulfide in detail.

Model persulfide substrates have been prepared for the study of the fundamental chemistry of persulfides such as their stability and degradation product(s) (2, 3, 16, 18). However, those persulfide substrates are not generated under physiological conditions and are not the focus of this review. There are various thiol-activated H₂S donors, which would generate a persulfide intermediate and release H₂S in the presence of other thiol species such as cysteine. For example, diallyl disulfide and diallyl trisulfide have been shown to release H₂S via S-allyl persulfide intermediate in the presence of cellular thiols (Fig. 1A) (51). A series of S-acyl disulfide derivatives developed by Xian's laboratory (Fig. 1B) and dithioperoxyanhydrides developed by Galardon's laboratory (Fig. 1C) have been also used as H₂S donors via a persulfide intermediate (29, 72, 96, 97). Although those thiol-activated donors can generate persulfide, these persulfide species would quickly exchange with the thiol present in the system to generate H₂S. Thus, they are often viewed as H₂S donors instead of persulfide donors, although these donors can probably play the role of both.

OPEN IN VIEWER

FIG. 1.

H₂S donors utilizing perthiol as intermediate. (A) DADS and DATS H₂S donors, (B) S-acyl disulfide H₂S donors, and (C) Dithioperoxyanhydrides H₂S donors. DADS, diallyl disulfide; DATS, diallyl trisulfide; H₂S, hydrogen sulfide.

Our laboratory has developed a series of esterase-sensitive persulfide prodrugs (99). In the first example reported in 2017, the design took advantage of a masked “hydroxymethyl disulfide” intermediate for stability and esterase sensitivity. Specifically, a hydroxyl group was protected through ester formation (Fig. 2). Esterase-mediated hydrolysis of the ester would lead to a “hemithioacetal-like” intermediate 2, which would collapse spontaneously to yield the corresponding persulfide and an aldehyde. Using this strategy, six persulfide prodrugs with various R groups were synthesized by a one-step reaction. To tune the release rate, esters of varying steric hindrance were used to mask the “hemithioacetal-like” moiety. The half-lives of prodrugs were determined to be in the range of 12–145 s in the presence of 0.5 unit/mL porcine liver esterase (PLE) when studied by monitoring the formation of the aldehyde. 2,4-Dinitrofluorobenzene (DNFB) was used to directly trap the persulfide released from the prodrugs to form a relatively stable product. The trapping yields were in the range of 82%–92% at room temperature in phosphate-buffered saline (PBS) solution.

OPEN IN VIEWER

FIG. 2.

Esterase-sensitive persulfide prodrugs through a “hemithioacetal-like” intermediate. DNFB, 2,4-dinitrofluorobenzene.

The prodrugs were used in clarifying one chemistry question and in demonstrating the protective effect of persulfide in MI/R injury. Specifically, there had been a debate in the literature as to whether S-methyl methanethiosulfonate (MMTS) would only selectively react with thiol, but not persulfide groups. This is an important question because MMTS had been used to study intracellular thiol contents. The selectivity of MMTS between thiol and persulfide affects the interpretation of such results. Using those prodrugs, it was unequivocally shown that both benzyl thiol and benzyl persulfide were able to react with MMTS, leading to trapped products as confirmed by liquid chromatography–mass spectrometry (LC-MS). As a result, one can unequivocally say that MMTS does not have sufficient selectivity for thiol over persulfide to be used as a thiol-selective probe. It was found that 50 or 100 μg/kg of the prodrug was able to effectively protect mice from MI/R injury. Either lower (12.5 μg/kg) or higher dosage (500 μg/kg) did not exhibit any protective effects, demonstrating dose dependence. The amount of sulfane sulfur was also quantified in the plasma after the administration of the prodrugs. The sulfane sulfur level in plasma was elevated to ∼0.6 μM after treatment with 100 μg/kg of the prodrug. Sulfane sulfur refers to “bound sulfur” that can be converted to H₂S by reducing reagents such as dithiothreitol (39). No obvious cytotoxicity was observed in rat myoblast H9c2 cells after 24 h of incubation with up to 100 μM of these esterase-sensitive persulfide prodrugs.

Besides delivering general persulfide species, our laboratory took advantages of a “trimethyl lock”-facilitated lactonization reaction to deliver some endogenous persulfide species such as hydrogen persulfide (H₂S₂) and GSSH (Fig. 3A, B) (94, 95). The “trimethyl lock” concept was first reported by Cohen and Borchardt in the 1970s to offer entropic advantages in promoting lactonization through the introduction of three pendant methyl groups (labeled with red in compound 4-1) (45), with rate enhancement on the order of 5.1 × 10¹⁰-fold (5). In practice, the “trimethyl lock” system offers the chance for rapid and spontaneous lactonization upon unmasking of the hydroxyl group. As a result, similar concepts have been used for the preparation of esterase-sensitive prodrugs of various types (45). Along this line, our laboratory took advantage of the “trimethyl lock” system for the preparation of esterase- and phosphatase-sensitive prodrugs of sulfur species. Esterase-mediated hydrolysis of the phenol ester prodrug 4-1 and 6 or phosphatase-mediated cleavage of phosphate prodrug 4-2 would initiate the spontaneous lactonization, leading to the release of H₂S₂ or GSSH, respectively (Fig. 3A, B).

OPEN IN VIEWER

FIG. 3.

Esterase- and phosphatase-sensitive H₂S₂ donors (A) and esterase-sensitive GSSH donors (B). GSSH, glutathione persulfide.

To study the release kinetics, high-performance liquid chromatography (HPLC) was used to measure the generation of byproduct lactone 5 during the incubation. The half-live was determined to be 24 min with 1 unit/mL PLE for 4-1. The phosphatase-sensitive H₂S₂ prodrug 4-2 was designed to improve water solubility of the H₂S₂ prodrug and to increase the diversity of biological stimulus (Fig. 3A). In the presence of 2 units/mL alkaline phosphatase, the half-life of the H₂S₂ prodrug 4-2 was measured to be 28 min. For detecting H₂S₂ release, a H₂S₂ fluorescent probe (DSP-3) and a trapping agent monobromobimane (mBB) were used. The total trapping yield was ∼68%. In terms of the biological effect, H₂S₂ released from the prodrug was shown to independently inactivate GAPDH by increasing the S-persulfidation level of the enzyme, while H₂S was not able to do so without additional oxidizing agent such as H₂O₂. No obvious cytotoxicity was observed on rat myoblast H9c2 cells after 48 h incubation with 100 μM of the esterase- or phosphatase-sensitive H₂S₂ prodrugs.

Our laboratory also utilized the “trimethyl lock” system to build a GSSH donor (Fig. 3B). The GSSH released from the prodrug was trapped by DNFB for the purpose of detection and quantification. However, the trapping yield was only ∼10%, which might indicate the unstable nature of GSSH. Further mass spectrometric studies showed that GSSH itself quickly underwent disproportionation to give GSSSG and H₂S. In this study, the GSSH prodrug showed protective effect on H₂O₂-induced toxicity on rat myoblast H9c2 cells, while H₂S, GSH, and the byproduct lactone failed to rescue the cells from the same kind of damage. No obvious toxicity was observed on H9c2 cells at 200 μM after 24 h of incubation with the GSSH donor.

Matson and colleagues reported an ROS-sensitive N-acetyl cysteine (NAC) persulfide donor (70). The design used a boronic acid moiety, which is sensitive to peroxide-mediated oxidation. As shown in Figure 4A, removal of the aryl boronic ester by H₂O₂ would lead to the formation of an unstable phenol 8, which is known to undergo 1,6-elimination leading to the release of NAC persulfide. The NAC persulfide released was directly validated by LC-MS. The cytoprotective effect of the NAC persulfide donor on H9c2 cells was examined. In the presence of 100 μM H₂O₂, cell viability dropped to 30%. When 200 μM of the NAC persulfide donor was applied, the cell viability increased from 30% to 100%. However, it should be noted that some protective effects were also observed from the aryl boronic ester, which might be due to the consumption of the peroxide species by the boronate molecule. On rat myoblast H9c2 cells, compound 7 did not show toxicity at 200 μM after 1 h incubation.

OPEN IN VIEWER

FIG. 4.

(A) ROS-sensitive N-acetyl cysteine persulfide prodrugs and (B) ROS-sensitive benzyl persulfide prodrugs.

Chakrapani and coworkers also reported an ROS-triggered benzyl persulfide donor. This design relied on a retro Michael addition reaction (4). They also utilized a vinyl boronate ester as the ROS-sensitive moiety (Fig. 4B). In the presence of H₂O₂, the oxidation would yield an enol intermediate 11, which is expected to collapse by a 1,4-O-S-relay mechanism to yield benzyl persulfide and an α,β-unsaturated carbonyl compound (cinnamaldehyde). In the presence of 10 equivalents of H₂O₂ at pH 7.4 and 37°C, the decomposition rate of the donor was measured to be 5.3 × 10⁻² min⁻¹. Both DNFB and mBB were used to detect the released persulfide species. The rate constants for the formation of the trapped products with DNFB and mBB were determined to be 0.15 and 0.12 min⁻¹, respectively. It was also found that the byproduct cinnamaldehyde was able to react with the released persulfide through Michael addition, which led to a low yield of persulfide. The cytoprotective effect of the persulfide donor was examined on colon carcinoma DLD-1 cells. This persulfide donor was able to effectively rescue DLD-1 cells from damage caused by menadione and a juglone analog named JCHD, which are known to induce oxidative stress.

Khodade and Toscano reported S-substituted thiosiothioureas as persulfide donors under physiological conditions (Fig. 5A) (33). In this design, the terminal sulfhydryl of the persulfide moiety was protected by an S-alkylthioisothiourea moiety. Under near physiological conditions (pH = 7.4), the neutral intermediate 14 was shown to undergo an elimination reaction to produce persulfide species with yields in the range of 75%–96% and an aryl cyanamide byproduct. It should be noted that the gem-dimethyl group was used to provide enhanced stability of the prodrug. The decomposition rates of the persulfide precursors were also examined in PBS containing 20% deuterated methanol (CD₃OD) at pH 7.4. The half-lives ranged from 5.0 to 17.3 min depending on the substitution on the nitrogen of the thiourea moiety.

OPEN IN VIEWER

FIG. 5.

(A) S-substituted thiosiothioureas as pH sensitive persulfide donors and (B) acid- or fluoride triggered persulfide donors.

Xian and coworkers developed several persulfide donors based on an O → S relay protection and deprotection mechanism (Fig. 5B) (31). Different protecting groups including trimethylsilyl and triethylsilyl were used for acid- or fluoride-triggered release. Upon desilylation, the unstable intermediate 17 or 19 would decompose to release hydrogen persulfide, benzyl persulfide, or a modified cysteine persulfide (N-acetyl-cysteine methyl ester) and aldehyde 20. The release kinetics were studied in DMF/PBS buffer (4:1). For the hydrogen persulfide donors, at pH 5.0, the time for total consumption of the donors ranged from 2.5 to 13 h. When the pH was increased to 7.4, the total decomposition time also increased by about two- to threefolds. For the persulfide donors, the total decomposition time ranged from 3.5 to 4.5 h at pH 5–6. At pH 7.4, the decomposition time was >12 h. In the presence of potassium fluoride, the total consumption time ranged from 5 min to 8 h. The released persulfides were directly captured by iodoacetamide. The yield was determined to be ∼90% for benzyl persulfide and 38%–64% for the modified cysteine persulfide.

SO₂: Endogenous Production, Physiological Roles, and Prodrugs

Although traditionally viewed as a toxic gas to human body and environment, SO₂ has been reported to have various pharmacological effects, especially in the cardiovascular system (14). Both SO₂ gas and its derivatives (Fig. 7) induce concentration-dependent relaxation of isolated rat aortic rings with similar potency, suggesting SO₂'s role as a potential vasoactive molecule (14, 59). SO₂ exposure was reported to cause a consistent decrease in blood pressure in salt-induced hypertension rat model. Further in spontaneously hypertensive rats, a significant decrease in SO₂ concentration and aspartate aminotransferase (AAT) activity in the plasma was observed, suggesting SO₂'s role in blood pressure regulation (13, 19). It has been shown that SO₂ provides protective effects in monocrotaline-induced pulmonary hypertension, MI/R injury, and atherosclerosis (25, 49, 88). Endogenous SO₂ has also been shown to offer protection against oleic acid-induced acute lung injury by suppressing oxidative stress (8). The fact that SO₂ has its own enzymatic production pathway and plays various roles under physiological/pathological conditions has led to the proposal that SO₂ is a new gasotransmitter (19).

OPEN IN VIEWER

FIG. 7.

Equilibrium between various SO₂ derivatives.

Although the story of SO₂ as an endogenous signaling molecule is not without controversy, one thing is for sure: SO₂ has been shown to have pharmacological effect and is produced endogenously through the metabolism of sulfur-containing amino acids or through oxidation of H₂S (Fig. 6) (28, 75). Some of these reactions are catalyzed by enzymes (26). For example, cysteine dioxygenase is known to oxidize l-cysteine to l-cysteine sulfinate, which is subsequently converted to β-sulfinylpyruvate by AAT. SO₂ is generated through the decomposition of β-sulfinylpyruvate (75, 79). Other enzymes such as NADPH oxidase are also known to produce SO₂ by the oxidation of H₂S (61). SO₂ is also an intermediate of H₂S metabolism. H₂S can be first oxidized to thiosulfate by sulfide oxidase and then in the presence of thiosulfate sulfurtransferase, thiosulfate can be converted to SO₂. GSSH generated during the H₂S metabolic process can also be oxidized to SO₂ by sulfur dioxygenase (21, 28). It is worth noting that sulfite can be converted to thiosulfate, which can be a potential H₂S donor (76). SO₂ concentration in rat plasma was determined to be in the range of 10–20 μM by different research groups (14, 46). SO₂ concentrations in various tissues are known to vary with the highest concentration reported to be in aortic tissue (5.55 ± 0.35 μmol/g protein), while in pulmonary artery, mesenteric artery, tail artery, and renal artery, the concentrations were reported to be 3.27 ± 0.21, 2.67 ± 0.17, 2.50 ± 0.20, and 2.23 ± 0.19 μmol/g protein, respectively (14).

OPEN IN VIEWER

FIG. 6.

Endogenous generation of SO_2. SO₂, sulfur dioxide.

SO₂ can readily dissolve in water and form a hydrated SO₂ complex (SO₂·H₂O). This complex would afford bisulfite (HOSO₂ ⁻ or HSO₃ ⁻) and sulfite (SO₃ ²⁻) ions depending on the pH. Bisulfite has two tautomers in aqueous solution (Fig. 7) (15, 63). The first and second pKa of SO₂·H₂O are 1.81 and 6.97, respectively. Therefore, under physiological conditions (pH = 7.4), the molar ratio of bisulfite to sulfite is ∼1:3 (80 –82). Disulfite (S₂O₅ ²⁻) can be formed by the dimerization of bisulfite. These species exist in equilibrium in the biological system. It is important to recognize that studying SO₂ under physiological conditions is actually the examination of various SO₂ derivatives as a whole, not a single species (Fig. 7).

Despite the understanding of SO₂ chemistry and its known pharmacological roles, the mechanism of action for SO₂ is far from clear yet. Whether SO₂ has biological targets is still a question. How SO₂ might bind and/or modify possible targets has not been well studied. Further, the role of SO₂ in affecting redox balance is not clear. All these unanswered questions are more of a reason to develop tools that can deliver SO₂ in a controllable fashion to enable detailed mechanistic studies. The section below describes the recent development of SO₂ prodrugs and donors.

In 2012, Chakrapani's laboratory reported a series of thiol-sensitive SO₂ donors as antimycobacterial agents (Fig. 8) (57). This was the first reported effort in making SO₂ prodrugs. The design was based on the 2,4-dinitrophenylsulfonamide scaffold, which was previously used as a thiol probe. The SO₂-releasing mechanism involved ipso-nucleophilic substitution on an electron-deficient aromatic ring (Fig. 8). Various analogs with different R₁ and R₂ groups were synthesized to evaluate the effects of the substituent on SO₂ production yields and rates. In the presence of 10 equivalents of cysteine, 100 μM of donors were able to release SO₂ in 24%–100% yield within 30 min in PBS (pH = 7.4). The release half-life ranged from 2 to 64 min, depending on the substituents on the sulfonamide nitrogen. One of the SO₂ donors was found to inhibit Mycobacterium tuberculosis (Mtb) with a minimum inhibitory concentration (MIC) of 0.05 μM, while its IC₅₀ on human embryonic kidney HEK293 cells was 7 μM. The inhibitory effects on M. tuberculosis were said to possibly involve the depletion of thiol species during the activation stage and the ability of SO₂ to induce oxidative stress and to damage biomacromolecules such as lipids, proteins, and DNA. Later, the Chakrapani laboratory reported the synthesis of additional 2,4-dinitrophenylsulfonamide analogs as SO₂ donors and antimicrobial agents against M. tuberculosis (56). The antimicrobial activity of those 2,4-dinitrophenylsulfonamide SO₂ prodrugs was also tested on various Gram-positive bacteria such as methicillin-resistant Staphylococcus aureus (MRSA) (66). Among these compounds, the most potent one has a MIC of 2 μg/mL. However, these compounds showed no activity against Gram-negative Escherichia coli at up to 32 μg/mL, probably because of permeability issues for the prodrug. Evidence showed that the SO₂ release yield is not the only factor that affects the antimicrobial potency. Incidentally, these 2,4-dinitrophenylsulfonamide SO₂ prodrugs have also been widely used as a source of SO₂ to validate SO₂ detection methods (47, 52).

OPEN IN VIEWER

FIG. 8.

Thiol-activated SO₂ donors.

In 2018, Chen's group used this thiol-responsive sulfur dioxide prodrug to form nanocarriers for doxorubicin (74). The released SO₂ showed a synergistic effect with doxorubicin toward drug-resistant cancer cells in cell culture studies.

These thiol-triggered 2,4-dinitrophenylsulfonamide SO₂ prodrugs have a well-defined release mechanism, half-life, and yields, and are great research tools. However, the use of thiol as a trigger may complicate mechanistic studies of sulfur signaling, and the dinitrophenyl byproduct may also limit its application in therapeutic application due to its potential metabolic toxicity.

In 2013, the Chakrapani laboratory reported a 1-phenyl-benzosultine scaffold for the design of thermally responsive SO₂ donors (Fig. 9) (55). Such a structure can undergo a thermal retro-Diels-Alder reaction, resulting in the release of SO₂, which can further react with 1,3-dienes to form benzosulfone. Specifically, in acetonitrile (ACN) at 37°C, 1-phenyl-benzosultine would decompose to produce the more stable 1-phenyl-benzosulfone in 92% yield. Such results were interpreted as to suggest that the SO₂ released from 1-phenyl-benzosultine further reacted with the 1,3-dienes to generate 1-phenyl-benzosulfone. However, in phosphate buffer at 37°C and pH 7.4, the yield of SO₂ from 1-phenyl-benzosultine was ∼89% as calculated based on the formation of SO₂ as sulfite. Such results suggest that solvent plays an important role in determining whether the released SO₂ would undergo further reactions with a 1,3-diene. Although the mechanistic aspect of this solvent effect was not further explored, it is possible that the enhanced solubility of SO₂ and the conversion of SO₂ to sulfite species in aqueous medium could help shift the equilibrium in favor of SO₂ generation. Through the modification of electron density on the phenyl ring, various SO₂ donors were designed with half-life (calculated based on sulfite formation) ranging from 10 to 68 min in phosphate buffer containing 30%–40% ACN, at 37°C and pH 7.4. Compound 24 (100 μM, R = H) was tested in a supercoiled plasmid cleavage assay and was found to have a nearly identical DNA damage profile with 100 μM sulfites (Na₂SO₃ and NaHSO₃ in a 1:3 ratio) in the presence of an equimolar amount of Cu (II). These SO₂ donors have tunable release rates and do not require thiol as a trigger, which avoids the potential disruption of the thiol redox balance in cells.

OPEN IN VIEWER

FIG. 9.

Thermally activated SO₂ prodrugs.

In 2015, the Chakrapani group reported a series of photochemically activated SO₂ donors based on a benzosulfone scaffold (Fig. 10A) (54). Various benzosulfone compounds 26 with different substitutions on the aromatic ring or at the 1-positions were synthesized. Under ultraviolet (UV) radiation (200–400 nm) for 10 min at 37°C in phosphate buffer (containing 30%–40% ACN or ethanol, pH 7.4), the release yield for 100 μM of these compounds ranged from 12% to 90% (SO₂ was determined as SO₃ ²⁻ using ion chromatography); while without UV irradiation, no SO₂ release was detected under the same conditions. In 2015, Uchida's laboratory reported a UV-sensitive diarylethene SO₂ donor scaffold (Fig. 10B) (37). The closed-ring form 28 is thermally stable, and no SO₂ was generated after heating at 70°C for 7 days in hexane. However, open-ring form 29 can release SO₂ after heating at 70°C for 1 day in hexane. Under visible light (λ > 480 nm), the closed-ring form 28 can be converted to the open-ring form 29 without SO₂ formation; under UV light such as 365 nm, closed-ring form 28 can be converted to the open-ring form 29, and 29 would then release SO₂ in the presence of UV light. Mouse fibroblast NIH/3T3 cells attached to polymethylmethacrylate films loaded with 28 were used to assess the toxicity of donor 28 after UV exposure. Specifically, NIH/3T3 cells were incubated with 28 at a concentration of 7.2 mg cm⁻², and cells were irradiated for 8 min (365 nm, 95 mW cm⁻²). Cellular damage and detachment were observed 2 h after exposure to UV light, while control experiments without 28 treatment showed no damage after the same level of UV radiation. In 2019, Yang's group utilized the same SO₂ release strategy to build SO₂-releasing nanoparticles useful in potential cancer therapies (48).

OPEN IN VIEWER

FIG. 10.

Photochemically activated SO₂ prodrugs based on benzosulfone structure (A) and diarylethene structure (B).

In 2016, Xian's laboratory reported a SO₂ donor utilizing a benzothiazole sulfinate (BTS) moiety (Fig. 11) (10). In the presence of water, BTS 31 was found to release SO₂. BTS was shown to have very good water solubility (up to 100 mM) and to release SO₂ in a pH-dependent manner. The half-life of BTS was determined to be 7.5 min at pH 4 and 13 days at pH 7.4 by monitoring the formation of benzothiazole with HPLC. Thus, BTS can serve as a slow-release SO₂ donor. The vasorelaxation activity of BST was tested on rat thoracic aorta rings. BST with a concentration range from 250 μM to 4 mM showed concentration-dependent relaxation with a ∼70% relaxation of rat thoracic aorta rings at 4 mM BTS.

OPEN IN VIEWER

FIG. 11.

Water-activated SO₂ donors.

In 2017, our laboratory designed a series of SO₂ donors via a click-release reaction between thiophene dioxide and a strained alkyne (Fig. 12) (86). This strategy was inspired by our previous work on click-release CO prodrugs (24, 84). As a proof of concept, tetrachlorothiophene dioxide 33-1 (with R₁ to R₄ being Cl) was used to react with endo-bicyclo[6.1.0]nonyne (BCN) 34 to generate SO₂. The formation of the cycloaddition product suggests SO₂ generation. The released SO₂ was detected by its reaction with DTNB (50). The second-order rate constant between tetrachlorothiophene dioxide and endo-BCN was determined to be 1.50 M ⁻¹ s⁻¹ in MeOH at room temperature. The rate constant of the reaction between thiophene dioxide and a strained alkyne can be tuned from 0.01 to 1.5 M ⁻¹s⁻¹ by altering the lowest unoccupied molecular orbital of thiophene dioxide through different substitutions. To achieve real-time monitoring of SO₂ generation, compound 33-2 was prepared; its reaction with BCN led to the generation of a fluorescent product (quantum yield: 0.138) 35-2 with excitation and emission wavelength being 296 and 470 nm, respectively. These SO₂ donors have a well-defined release mechanism, a specific trigger, and tunable release rate. However, SO₂ release requires two components, which limit their application in biological systems because of the requirement of local enrichment of both components. To improve this click-release SO₂ system, our laboratory also developed a unimolecular SO₂ release system by combining the thiophene dioxide and alkyne into one molecule (Fig. 12) (23). In this unimolecular SO₂ release system, a strained alkyne such as that in BCN was no longer required for the Diels-Alder reaction. Instead, a nonactivated linear alkyne was used as the dienophile. By modifying the electronic properties of the alkyne moiety, the tethering linker and the substituents on the tether, the half-life was tuned from 0.17 h to >7 days at 37°C (solvent system: DMSO: H₂O = 1:3). One of the SO₂ donors (t _1/2 = 0.28 h) was used to test for its ability to cleave pBR322 supercoiled plasmid. It was found that 100 μM of this donor induced pronounced DNA cleavage in comparison with the negative control 37 at 37°C. One thing worth noting is that intramolecular cycloaddition is spontaneous in aqueous solution, and to some degree in organic solvents as well. The latter is a problem in the synthesis of the pure SO₂ prodrug.

OPEN IN VIEWER

FIG. 12.

SO₂ donors via click-release chemistry.

In 2017, our laboratory reported a class of esterase-sensitive SO₂ donors based on a modified Julia olefination reaction (87). As shown in Figure 13A, after esterase-catalyzed hydrolysis of the ester bond, a hydroxyl group was revealed to form intermediate 39. Subsequent intramolecular addition–elimination would lead to intermediate 41, which spontaneously collapsed to generate SO₂ and 2-hydroxybenzothiazole as a side product. Compound 38 (BW-SO-202, R = CH ₃ ) was synthesized to validate this concept. Fifty micromolars of BW-SO-202 was incubated with 5 unit/mL PLE in 10% dimethyl sulfoxide (DMSO)/PBS at 37°C. The formation of 2-hydroxybenzothiazole was readily observed (t _1/2 ∼1 h), indicating the release of SO₂. No degradation of BW-SO-202 was found within 3 h without PLE. By using different acyl groups to control the esterase-catalyzed hydrolysis rate, the SO₂ release rate can be tuned. When 50 μM of BW-SO-202 was incubated with 5 unit/mL PLE in 10% DMSO/PBS at 37°C, the slowest SO₂ donor (with R being t-butyl group) showed a 10% release after 2 h, while when R was ethyl group, the half-life was only 10 min.

OPEN IN VIEWER

FIG. 13.

(A) Esterase-sensitive SO₂ donors based on modified Julia olefination and (B) Esterase-sensitive self-immolative SO₂ donors.

In 2018, the Chakrapani group reported another class of esterase-sensitive SO₂ donors (67). As shown in Figure 13B, after esterase-catalyzed hydrolysis, the ionization of the unmasked hydroxyl group would lead to an elimination reaction and C-S bond cleavage to release SO₂ from the sulfonate moiety. As a proof of concept, compound 44 was synthesized with R₁ being coumarin, which allows for fluorescence turn-on after release. When 50 μM of compound 44 was incubated with 1 unit/mL PLE in PBS (pH = 7.4, 10 mM) at 37°C, the half-life was 5 min based on the disappearance of 44. On the contrary, without esterase, compound 44 was stable at pH of 7.4 within 30 min. Esterase-catalyzed SO₂ release was also validated by colorimetric and fluorescent probes. Compounds with different R₁ groups were also synthesized to evaluate the effect of leaving groups on SO₂ generation efficiency. SO₂ delivery from those donors in a biological system was tested on A549 cells by using fluorescence imaging. Cells were treated with a SO₂-selective fluorescent dye for 30 min, and then the fluorescent dye was removed. After that, cells were treated with the esterase-sensitive SO₂ donor or NaHSO₃ for 30 min in PBS (pH 7.4, 10 mM) at 37°C. Twenty-five micromolars of a SO₂ donor (R₁ = 4-OMePh) was able to turn on a sulfite fluorescence probe with a similar intensity to that of 200 μM of NaHSO₃. The fluorescence imaging results suggest that this esterase-sensitive SO₂ donor is cell permeable, which offers the advantage of efficient intracellular SO₂ delivery. However, much of the end results will be governed by the kinetics of permeation, release, and species interexchange in aqueous solution. Thus, the biological implications of using cell-permeable SO₂ donors versus sulfide need more work to delineate. Most of these donors showed moderate cytotoxicity (cell viability >60%) on cervical cancer cells HeLa and lung cancer cells A549 lines at 10 μM.

Based on the currently available data, it seems that one of the major areas of application of SO₂ donors is in the treatment of cardiovascular disease. However, the application of SO₂ in the treatment of bacterial infection has also been explored (9). Thiol-activated SO₂ prodrugs developed by Chakrapani's laboratory have shown promising results against M. tuberculosis and MRSA, probably by disrupting redox balance and damaging biomacromolecules such as lipids, proteins, and DNA. However, those thiol-activated prodrugs face the same dilemma as many narrow-spectrum antibiotics: they are far more effective against Gram-positive bacteria than Gram-negative bacteria, possibly due to the difficulty for the donors to penetrate the outer membrane of Gram-negative bacteria (38). Furthermore, translating antimicrobial effect in bacterial culture to in vivo applications faces the same hurdle as many other known antimicrobials; that is, they have to deal with issues of toxicity, selectivity, potency, and drug delivery and other pharmacokinetic challenges.

CS₂: A Precursor to H₂S or an Active Compound by Itself?

CS₂ has been implicated in various biological processes on its own, although CS₂ is only produced naturally by soil-based microorganisms, and there is currently no evidence of its enzymatic production in mammalian cells (90). CS₂ is relatively inert to hydration to form H₂S at biologically relevant pH, and the approximate half-life was estimated to be 1.1 years at pH 9 (12). CS₂ is also inert to carbonic anhydrase (CA), which catalyzes the hydrolysis of CO₂. When hydrolyzing CS₂, CA II is about nine orders of magnitude less effective compared with the hydrolysis of CO₂ (11, 12, 17, 77). However, free thiol such as cysteine and GSH can react with CS₂ to form trithiocarbonates, which might be a form of post-translational cysteine residue modification to modulate protein function much the same way as protein S-persulfidation. Free amines can also react with CS₂ to form dithiocarbamates (DTCs). For example, the free amino group on the dopamine molecule can form DTC with CS₂. This DTC formed from dopamine and CS₂ is known to chelate the copper center of dopamine β-hydroxylase and inhibit its activity (53). CS₂ can also bind and inhibit proteins with the Ni/Fe–S cluster in a reversible manner; this can also be another way of CS₂ signaling (44). More comprehensive and detailed discussions can be found in a recent review (12). There have been efforts to build CS₂-releasing molecules to further probe CS₂'s biological effect. Along this line, the Ford group reported a series of DTCs derived from 47 as CS₂ donors (Fig. 14) (11, 77). At pH = 7.4, 37°C, the k_obs for CS₂ release is 775, 3.5, and 0.043 × 10⁻³ s⁻¹ for imidazolidyldithiocarbamate (ImDTC), diisopropyl DTC, and morpholiniumdithiocarbamate (MorDTC), respectively. Such results suggest the tunability of release kinetics for this class of donors. The release of CS₂ from DTCs is an acid-catalyzed reaction. For example, at pH 6.5, the k_obs for MorDTC is 0.36 × 10⁻³ s⁻¹, an acceleration of >10-fold compared with that of pH 7.4. However, the catalysis mechanism is different among the various types of analogs. For example, for the aliphatic analogs, CS₂ release from the protonated dithiolate group involves an obligatory deprotonation step and possibly proton transfer to the nitrogen. As a result, CS₂ release is slower from the protonated form. On the contrary, ImDTC protonation would first occur on the imidazole nitrogen, leading to enhanced release of CS₂ because the protonated form is a better leaving group than imidazole itself (Fig. 14).

OPEN IN VIEWER

FIG. 14.

DTC molecules as CS₂ donors. CS₂, carbon disulfide; DTC, dithiocarbamate.

Conclusions

One of the difficulties in studying sulfur signaling is the rapid interconversion of various sulfur species through redox reactions, which leads to scrambling among the various species. Furthermore, the reciprocal effects of cellular oxidative stress and availability of various sulfur species convolute the interpretation of results from mechanistic studies. The availability of a range of donors of sulfur species at various oxidation states hopefully would help deciphering results from mechanistic studies. It is our hope that this review article will stimulate additional work to further improve the tools available, and more discussions of the challenging issues facing sulfur-signaling studies, especially in the context of redox chemistry, with the ultimate goal of achieving a thorough understanding of the complex chemistry and biology issues in sulfur-mediated biological processes.

With the many demonstrated pharmacological effects of various sulfur species, there is always the question of whether any of these pharmacological effects would ever translate into successful development of therapeutic agents. This is truly a billion-dollar question, and it is a long-term issue. Intuitively, the demonstrated pharmacological effects would surely suggest a high degree of likelihood for further pharmaceutical development of these donors of sulfur species. However, the reality might be very different. It is our opinion that a key to success in this area is to identify an area of unmet medical need that can tolerate the complication of the rapid interconversions among the various sulfur species. Further, one has to put the question of clinical indication in the context of desired pharmacokinetic profiles, the route of delivery, and the possibility of feedback controls in sulfur signaling and thus development of drug resistance. The last point would require in-depth understanding of sulfur-signaling biology and much more data from animal models. We might still be a long way from realizing the therapeutic potential of sulfur signaling. Only in the context of a specific clinical indication, one can truly assess the possibility for a particular sulfur donor to be developed for further clinical application.