Thionoesters: A Native Chemical Ligation-Inspired Approach to Cysteine-Triggered H₂S Donors

Abstract

Native chemical ligation (NCL) is a simple, widely-used, and powerful synthetic tool to ligate N-terminal cysteine residues and C-terminal α-thioesters via a thermodynamically-stable amide bond. Building on this well-established reactivity, as well as advancing our interests in the chemical biology of reactive sulfur species including hydrogen sulfide (H₂S), we hypothesized that thionoesters, which are constitutional isomers of thioesters, would undergo a similar NCL reaction in the presence of cysteine to release H₂S under physiological conditions. Herein, we report mechanistic and kinetic investigations into cysteine-mediated H₂S release from thionoesters. We found that this reaction proceeds with high H₂S-releasing efficiency (~80%) and with a rate constant (9.1 ± 0.3 M⁻¹ s⁻¹) comparable to copper-catalyzed azide-alkyne cycloadditions (CuAAC). Additionally, we found that the final product of the reaction of cysteine with thionoesters results in the formation of a stable dihydrothiazole, which is an iron-binding motif commonly found in siderophores produced by bacteria during periods of nutrient deprivation.

Graphical Abstract

Introduction

H₂S is now recognized as an important biological signaling molecule¹ that is produced endogenously, cell membrane permeable, and reactive toward cellular and/or molecular targets.² The endogenous production of H₂S stems primarily from catabolism of cysteine and homocysteine by cystathionine β-synthase (CBS), cystathionine γ-lyase (CSE), and 3-mercaptopyruvate sulfurtransferase (3-MST).³ Recently, increasing interest has focused on harnessing H₂S as a potential therapeutic agent⁴ based on its role in vasodilation,⁵ neurotransmission,⁶ and angiogenesis.⁷ Although the majority of prior reports have used sodium hydrosulfide (NaSH) or sodium sulfide (Na₂S) as sources of H₂S, the addition of these salts to buffer leads to an almost instantaneous increase in H₂S concentration, which is in stark contrast to the slow, gradual endogenous production of H₂S.⁸ In efforts to provide more physiologically-relevant rates of H₂S release, researchers have developed different types of H₂S-releasing molecules (Figure 1a).^9-11 For example, Lawesson’s Reagent and related derivatives^12-13 have been used as hydrolysis-activated H₂S donors that function at physiological pH, and dithiolethiones, such as ADT-OH, have been conjugated to non-steroidal anti-inflammatory drugs (NSAIDs) to access H₂S prodrug conjugates.¹⁴ More recently, “triggered-release” scaffolds have also been reported, including those activated by light¹⁵ and enzymatic activation.¹⁶ In addition, recent work has demonstrated that carbonyl sulfide (COS)-releasing scaffolds can also function as H₂S donors via the rapid conversion of the released COS to H₂S by carbonic anhydrase.¹⁷

Figure 1.

a) Representative examples of common synthetic, small-molecule H₂S donors b) Selected small-molecule, thiol-triggered H₂S donors.

Drawing parallels to the enzymatic conversion of cysteine or homocysteine to H₂S, a number of H₂S donor motifs have been developed that are activated by thiols, such as cysteine and reduced glutathione (GSH) (Figure 1b). Polysulfides, such as the commonly-used diallyl trisulfide (DATS)¹⁸ or more recently reported synthetic trisulfides¹⁹ and tetrasulfides,²⁰ release H₂S in the presence of thiols via an intermediate persulfide. Building in complexity, Xian and co-workers have reported thiol-triggered H₂S donors based on protected N-mercaptan²¹ or persulfide²² platforms. Similarly, Matson and co-workers reported S-aroylthiooxime compounds,²³ which generate a thiol-reactive intermediate thiooxime. Thiol-mediated H₂S release from arylthioamides²⁴ and aryl isothiocyanates²⁵ has also been reported, although the mechanisms of H₂S release remains uninvestigated and low releasing efficiencies (~2% and 3%, respectively) are observed. To the best of our knowledge, the only reported cysteine-selective H₂S donor utilizes the established reactivity of acrylate Michael acceptors toward cysteine,²⁶ to subsequently trigger the generation of COS, which is quickly converted to H₂S by carbonic anhydrase.²⁷

To further the development of thiol-triggered H₂S donors, we were inspired by the well-established chemistry of native chemical ligation due to the high biological compatibility and presence of a sulfur atom. Native chemical ligation is the chemoselective reaction between a thioester and an N-terminal cysteine residue to generate a new amide bond.²⁸ This reaction has been applied extensively in the field of protein synthesis, including in the semisynthesis of a potassium channel protein.²⁹ The mechanism of this important ligation reaction begins by the nucleophilic addition of a cysteine sulfhydryl group to form an intermediate thioester, which then undergoes a rapid S to N acyl transfer to generate the more thermodynamically-stable amide product (Figure 2).³⁰ Despite the broad use of thioesters as activated coupling partners for native chemical ligation, to the best of our knowledge there have not been investigations into similar reactions with thionoesters, which are a constitutional isomer of thioesters. Building from our interest in the chemistry of reactive sulfur species,^31-33 we hypothesized that thionoesters would undergo a similar reaction pathway in the presence of cysteine, but would also generate hydrogen sulfide (H₂S) during the S to N acyl transfer step of the reaction. Such reactivity would not only provide access to new H₂S-releasing motifs, but also provide insights into new mechanisms of chemical ligation that could be accessed by simple interchange of oxygen and sulfur atoms in a reactive electrophile. Additionally, such platforms are also attractive because they mimic the enzymatic conversion of cysteine to H₂S. Herein, we present a mechanistic and kinetic investigation of thionoesters with cysteine and related species and also demonstrate that thionoesters function as cysteine-selective H₂S donors that proceed through a native chemical ligation-type mechanism.

Figure 2.

Generalized reaction scheme for native chemical ligation and release of H₂S upon addition of cysteine to a bis(phenyl) thionoester.

Results and Discussion

To prepare a model thionoester system, we treated phenyl chlorothionoformate with phenylmagnesium bromide at −78 °C in anhydrous THF to yield O-phenyl benzothioate (DPTE).³⁴ Despite previous reports,³⁵ we found that treatment of phenyl benzoate with Lawesson’s reagent required extended reaction times and afforded undesirable yields, which is consistent with the predicted decrease in reactivity of esters towards Lawesson’s reagent.³⁶ The structure and purity of DPTE was confirmed by NMR spectroscopy and HPLC (see Supporting Information). To determine whether thionoesters are a viable platform for H₂S release, we added 25 μM DPTE to buffered aqueous solutions (10 mM PBS, pH 7.4) containing varying concentrations of cysteine (25-500 μM) and monitored H₂S generation using the spectrophotometric methylene blue assay³⁷ (Figure 3a). Consistent with our design hypothesis, we observed an increase in H₂S release from DPTE at higher cysteine concentrations, suggesting that thionoesters are a viable platform for cysteine-triggered H₂S donation.

Figure 3.

a) Release of H₂S from DPTE in the presence of increasing cysteine concentrations (25, 125, 250, and 500 μM) in 10 mM PBS, pH 7.4 at 25 °C. b) Lack of H₂S release from structurally-related compounds (25 μM) in the presence of cysteine (500 μM, 20 equiv.)

To assess the H₂S-releasing efficiency from thionoesters, we used a methylene blue calibration curve to quantify the H₂S release (Figure S4). We measured that 20 μM of H₂S was released from a 25 μM solution of DPTE in the presence of 500 μM cysteine (20 equiv.), which corresponds to a releasing efficiency of 80%. In addition to the thionoester system, we also investigated H₂S release from structurally-related diphenyl ester (2) and diphenyl thioester (3) compounds under our conditions (Figure 3b). As expected, neither of these compounds released H₂S when treated with excess cysteine. Similarly, a representative secondary thioamide (4) failed to release H₂S in the presence of cysteine, suggesting the release of H₂S occurs exclusively from the thionoester moiety in the presence of cysteine.

To further investigate the selectivity of H₂S release from thionoesters, we treated DPTE with other biologically-relevant nucleophiles (Figure 4).³⁸ In the absence of any added nucleophiles, no hydrolysis-mediated H₂S release was observed from DPTE at physiological pH, although we note prior reports show that thionoesters are hydrolyzed under basic conditions to afford the corresponding thioacid and alcohol.³⁹ Treatment of DPTE with serine or lysine, chosen as representative alcohol- and amine-based nucleophiles respectively, did not result in H₂S release, although prior reports suggest that amines can react with thionoesters to yield thioamides via displacement of the corresponding alcohols.⁴⁰ To investigate this potential side reactivity, cysteine-triggered (500 μM) H₂S release from DPTE (25 μM) was measured in the presence of lysine (500 μM) and we observed no change in H₂S releasing-efficiency. We also investigated the reactivity of DPTE with thiol-based nucleophiles. Treatment of DPTE with homocysteine also resulted in H₂S release, although at a slower rate than from treatment with cysteine. This observation is consistent with a larger, less favorable transition state required for an intramolecular S to N acyl transfer in the homocysteine system in comparison with the cysteine system. Alternatively, the reduced rate may be reflective of the significant pK_a difference between cysteine (pK_a ~ 8.5) and homocysteine (pK_a ~ 10),⁴¹ meaning that under physiological conditions, the effective concentration of cysteine thiolate is much greater than homocysteine thiolate (~10% vs. ~0.03%). Surprisingly, treatment of DPTE with penicillamine did not result in H₂S release. We anticipated geminal methyl groups would help to preorganize the intermediate dithioester generated after nucleophilic attack and would result in faster H₂S release.⁴² However, the geminal methyl groups also likely significantly reduce the nucleophilicity of the thiol moiety due to steric congestion, which would subsequently disfavor the initial nucleophilic attack on the thionoester.

Figure 4.

Selectivity of H₂S release from DPTE in the presence of different analytes. Data were acquired at 1, 5, 10, 15, 30, 45, and 60 min. Methylene blue absorbance values are relative to the maximum absorbance value obtained from H₂S release in the presence of cysteine (1). Analytes: H₂O (2), serine (3), lysine (4), L-homocysteine (5), DL-penicillamine (6), L-cysteine methyl ester hydrochloride (7), N-acetyl-L-cysteine (8), N-acetyl-L-cysteine methyl ester (9), S-methyl-L-cysteine (10), GSH (11), cysteine + GSH (12), cysteine + lysine (13), PLE (1.0 U/mL) (14).

We also investigated whether different cysteine derivatives could generate H₂S release from DPTE to further understand the requirements for H₂S release from thionoesters. Treatment of DPTE with cysteine methyl ester did not affect H₂S production, suggesting that the carboxylic acid is not required for H₂S generation. By contrast, treatment of DPTE with N-acetylcysteine, N-acetylcysteine methyl ester, or S-methylcysteine completely abolished H₂S release, highlighting the requirement of a 2-aminoethanethiol moiety for productive H₂S release. Consistent with these results, treatment of DPTE with GSH, the most abundant biological thiol, did not generate H₂S, which is consistent with the requirement of a pendant amine to generate H₂S release. Despite the lack of H₂S release, we anticipated that GSH would still attack DPTE to form an intermediate dithioester, which should still be sufficiently electrophilic to react with cysteine to generate H₂S. To test this hypothesis, we treated DPTE (25 μM) with GSH (1 mM) and cysteine (500 μM), and observed a reduced rate of H₂S release. These results suggest that the competitive, non-productive, addition of GSH to the thionoester is reversible, and that the thionoester moiety can still react with Cys in the presence of GSH to release H₂S. Adding to the selectivity investigations, treatment of DPTE with porcine liver esterase (PLE) failed to generate H₂S, however, we cannot rule out consumption of the thionoester moiety by PLE or other native enzymes. Taken together, these results demonstrate the high selectivity of the thionoester moiety towards cysteine and homocysteine for H₂S release.

Building from the selectivity studies, as well as from the established mechanism of native chemical ligation, we proposed a mechanism for cysteine-mediated H₂S release from thionoesters (Scheme 1). Initial nucleophilic addition by cysteine on 1 generates tetrahedral intermediate 5, which collapses to form dithioester intermediate 6 and extrude one equivalent of phenol. Similar to native chemical ligation, subsequent nucleophilic attack by the pendant amine on the thiocarbonyl leads to the formation of substituted thiazolidine 7. Loss of H₂S, either by direct extrusion of HS⁻ or solvent-assisted extrusion of H₂S, results in formation of dihydrothiazole 8, which could be further hydrolyzed to form N-benzoyl-cysteine (9).

Scheme 1.

Proposed mechanism of H₂S release from DPTE in the presence of cysteine.

As a first step toward investigating our proposed mechanism, we determined the reaction order in cysteine by treating DPTE (25 μM) with varying concentrations of cysteine under pseudo-first order conditions at 25 °C and measuring H₂S release using the methylene blue assay (Figure 5). As expected, we observed that increased cysteine concentrations led to increased rates of H₂S production. The resultant releasing curves were fit to obtain pseudo first-order rate constants (k_obs), and plotting log[Cys] versus log[k_obs] confirmed a first-order dependence in cysteine, which is consistent with our proposed mechanism. Additionally, the obtained k_obs values were plotted against Cys concentrations to obtain a second-order rate constant of 9.1 ± 0.3 M⁻¹ s⁻¹ for the reaction. In comparison to other known reactivities, the rate of cysteine-triggered H₂S release from DPTE is comparable to the rate (10 – 100 M⁻¹ s⁻¹) of copper (I)-catalyzed azide-alkyne cycloadditions (CuAAC), a classic example of a “click reaction.”⁴³

Figure 5.

a) H₂S release by DPTE in the presence of increasing cysteine concentrations (250, 500, 1000, and 1250 μM) b) Plot of log(k_obs) vs log([Cys]) for DPTE. c) Plot of [Cys] vs. k_obs.

To further evaluate our proposed mechanism, we sought to identify the rate-determining step in cysteine-triggered release of H₂S from thionoesters. In native chemical ligation, the initial nucleophilic attack by thiols to form intermediate thioesters is reversible and has been utilized to enhance the reactivity of alkyl thioesters for native chemical ligation. However, in the presence of cysteine, the transthioesterification resulting from nucleophilic attack of the sulfhydryl group on the thioester is thought to be rate-limiting due to the rapid, and irreversible subsequent S to N acyl transfer to form the more thermodynamically-stable amide bond.³⁰ In the thionoester system, the initial attack by a thiol on DPTE results in extrusion of phenol, which is a much weaker nucleophile than a thiol and should not attack the generated dithioester intermediate. If other thiols are present in solution, then it is likely that they could attack the dithioester intermediate in a transdithioesterification reaction. This thiol exchange is supported by the observed reduced rate of H₂S generation from DPTE in the presence of competing thiols, suggesting that the initial nucleophilic attack on dithioesters is reversible.

Using similar pseudo first-order conditions as those used for the cysteine order dependence investigations (25 μM DPTE, 500 μM cysteine), we performed an Eyring analysis to determine the activation parameters for the reaction in an effort to further understand the amount of disorder in the rate-limiting transition state for the reaction (Figure 6). Our expectation was that if initial thiol addition is the rate limiting step, then we would observe a negative entropy of activation (ΔS^‡) of approximately −20 e.u., which is typical for a bimolecular reaction. In contrast, if the intramolecular S to N thioacyl transfer to form the substituted thiazolidine is the rate limiting step then we would expect a larger, more negative ΔS^‡ due to the highly-ordered structure required for the intramolecular cyclization Under our experimental conditions, we observed ΔS^‡ = −38 ± 3 eu, which is most consistent with intramolecular cyclization being the rate-determining step of the reaction.

Figure 6.

(a) Effect of temperature on rate of H₂S release from DPTE (25 μM) in the presence of cysteine (500 μM, 20 equiv.) (b) Eyring analysis of H₂S release from DPTE.

As a final step of characterizing the proposed mechanism, we performed a preparative scale reaction and isolated the reaction products. In addition to recovered starting material, we isolated a cysteine-derived dihydrothiazole (CysDHT) rather than N-benzoyl-L-cysteine as the major product of the reaction (Figure 7). These results suggest that the dihydrothiazole is stable under aqueous conditions and is not further hydrolyzed to N-benzoyl-L-cysteine. To further confirm the formation of CysDHT from DPTE, we synthesized an authentic sample of CysDHT and used HPLC to monitor the reaction progress. We treated a 100 μM solution of DTPE with 20 equiv. of cysteine and observed nearly complete conversion to phenol and CysDHT within 1 h. Using known concentrations of phenol and CysDHT to construct an HPLC calibration curve, we measured that the concentrations of phenol and CysDHT after 1 h were approximately 76 μM and 64 μM, respectively, which supports the high H₂S-releasing efficiency of thionoesters.

Figure 7.

a) Reaction conditions b) 100 μM DPTE in PBS (10 mM, pH 7.4) with 10% THF c) 100 μM PhOH in PBS (10 mM, pH 7.4) with 10% THF d) 100 μM CysDHT in PBS (10 mM, pH 7.4) with 10% THF e) Reaction aliquot after 1 h.

Although we were initially surprised by the inherent stability of the dihydrothiazole product, we note that biological formation of the dihydrothiazole moiety is a known post-translation modification of cysteine residues in bacteria.⁴⁴ For example, the cyclodehydration of internal cysteine residues results in formation of Fe(III)-coordinating dihydrothiazole, which are commonly found in sideophores,⁴⁵ such as yersiniabactin⁴⁶ and pyochelin.⁴⁷ Additionally, adjacent dihydrothiazole moieties can be oxidized to a bis(thiazole), and the planarity of this motif allows for intercalation of DNA as seen in bleomycin.⁴⁸ Taken together, these observations highlight the biological significance of the dihydrothiazole motif and provides new areas of investigation using this established reactivity.

Conclusion

By investigating the reactivity of DPTE with cysteine, we not only demonstrated the inherent reactivity of thionoesters toward cysteine in a native chemical ligation-type mechanism, but also demonstrated that this functional group provides a novel platform for highly efficient H₂S donation. We demonstrated that this reaction occurs at rates similar to those for the commonly-used Cu(II)-mediated azide/alkyne click reaction, with a second order rate constant of 9.1 ± 0.3 M⁻¹ s⁻¹. Our mechanistic investigations suggest that in comparison to native chemical ligation, the rate-determining step has been shunted from the addition of cysteine to the intramolecular S to N thioacyl transfer. Taken together, these investigations demonstrate that thionoesters are novel, cysteine-triggered H₂S releasing scaffold. Additionally, the high selectivity of DPTE towards cysteine warrants future exploration into the thionoester functional group for cysteine-selective reactive probes.