Main

The development of effective and logical strategies that sustainably access valuable and challenging molecules from simple feedstocks is one of the central tenets of modern synthetic chemistry, as well as relentless pursuit of chemists. Methyl thioesters not only are found in natural products and drug molecules but also represent important synthetic precursors in biosyntheses, such as the native chemical ligation reactions towards increasingly large polypeptides and proteins (Fig. 1a)1,2,3,4,5,6. Classical approaches to forging methyl thioester moieties generally rely on multistep syntheses or substitution reactions between electrophilic acyl compounds with either nucleophilic sodium methylthiolate or methyl mercaptan7,8,9. Nonetheless, these methods are plagued by the undesirable chemical properties of these two nucleophiles (Fig. 1b). Specifically, the handling of gaseous methyl mercaptan is not operationally straightforward, and sodium methylthiolate hydrolyses in moist air to the former compound. In addition to require of non-ideal reagents, other synthetic challenges such as extra preparatory steps of the starting materials, narrow functional group compatibility and harsh reaction conditions have severely curbed the efficient construction of valuable methyl thioesters through classical methods10,11,12,13,14. To address these liabilities, the Wu group successfully developed a palladium-catalysed carbonylation–thiomethylation of aryl halides with CO and thioesters15. In summary, reported methods can form limited thioesters, but methyl thioesters remain challenging to synthesize. Thus, broadly substrate-compatible, easy-to-operate and biocompatible strategies for methyl thioesters are yet underdeveloped but in high demand.

Fig. 1: Development of a strategy for the synthesis of thiomethyl esters.
figure 1

a, Representative biologically active thiomethyl esters. b, Synthetic challenges of methyl thioesters. c, Synergistic mode of photocatalysis and nickel catalysis. d, Working hypothesis. e, This work: nickel-catalysed, sensitized ET radical methyl thioesterification. Ni, nickel-catalysed process; PPh3, triphenylphosphine; [Ir] and Ir, Ir[dF(CF3)ppy2(dtbbpy)][PF6]; hv, visible light irradiation; cat., catalysis.

Full size image

Over the past decade, radical reactions triggered by photocatalysis have provided a powerful and efficient platform for the construction of challenging molecules16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38. In this field, the combination of transition metal catalysis for bond formation with photoinduced electron- and/or energy transfer processes has attracted increasing attention (Fig. 1c)39,40,41. Nickel-catalysed, photoinduced electron transfer (ET) strategies have been deeply investigated and successfully applied to synthetic chemistry42,43,44,45,46,47,48,49,50. Moreover, the amalgamation of energy transfer catalysis (EnT) with nickel catalysis has offered new opportunities for organic cross-coupling reactions (Fig. 1c). Energy transfer-mediated nickel catalysis for the construction of C–O and C–C bonds was pioneered by the groups of MacMillan51 and Molander52, respectively. Intriguingly, successful combination of photoinduced energy transfer process (EnT) with ET process by Xiao53, König54, Gilmour55, Weaver56 and their co-workers paved the way to new radical reaction scenarios (Fig. 1c). So far, only a few examples that operate through ET/nickel catalysis followed by single electron transfer (SET) have been reported by Rueping57 and Chu58, successfully enabling the generation of tri-substituted alkenes via a three-component cross-coupling reaction (Fig. 1c). In this type of reactions, energy transfer from excited photocatalysts often plays a role in the last step, enabling alkene isomerizations57,58. Nevertheless, more attractive roles for energy transfer in multi-catalytic mode remain to be exploited. For instance, energy transfer with a photosensitizer can promote chemical bond cleavage17,59,60, which offers more possibilities for constructing functional molecules.

Considering the importance and abundance of carboxylic acids in nature and the easy availability of dimethyl disulfide (DMDS), construction of thiomethyl esters from these two starting materials is an attractive strategy. In this Article, to address the synthetically challenging methyl thioesterification of carboxylic acids, we envisioned that merging energy transfer with ET could offer a cooperative manifold with nickel catalysis. According to the hypothesis, we designed a possible catalytic process (Fig. 1d). Thiomethyl radical, formed from DMDS via photocatalytic energy transfer60, would react with a nickel catalyst to form nickel complex (II). In this system, an active phosphorus radical cation might be generated via photocatalytic ET, as described by Doyle and co-workers61. Subsequently, nucleophilic attack of carboxylic acids produces the intermediate III. Then, acyl nickel complex (IV) would be obtained via the interaction of intermediate II with III. Further reductive elimination of IV would forge the desired product. Herein we describe the successful merger of a nickel-catalysed, energy transfer and ET for the radical thiomethyl esterification of carboxylic acids (Fig. 1e).

Results and discussion

Evaluation of the reaction conditions

To evaluate this multi-catalytic hypothesis, 4-phenylbutyric acid (1) and DMDS (2) were chosen as reaction partners. Pleasingly, after extensive optimization, 95% yield of thiomethyl ester (3) could be obtained with NiBr2(diglyme) and Ir[dF(CF3)ppy2(dtbbpy)][PF6] [Ir-F] as co-catalysts, triphenylphosphine (PPh3) and pyridine as additives (Fig. 2a, entry 1). It is worth mentioning that no decarboxylation by-product was observed in this reaction system, which shows high chemoselectivity62,63. The effect of other essential reaction parameters was further investigated (Fig. 2a). Noteworthily, the reaction yield dropped substantially in the absence of nickel catalyst (Fig. 2a, entry 2). Similarly, no desired product was observed without either photocatalyst or irradiation under visible light (Fig. 2a, entry 3). Further screening revealed that PPh3 plays an essential role in this transformation (Fig. 2a, entry 4). In addition, 80% yield of product was obtained in the absence of pyridine (Fig. 2a, entry 5). Increasing the amount of PPh3 did not affect the reaction yield, while the yield dropped substantially when less PPh3 was used (Fig. 2a, entries 6 and 7). Moreover, changing the amount of DMDS had no obvious effect on the reaction efficiency (Fig. 2a, entries 8 and 9). Higher or lower amount of the nickel catalyst led to low reaction yields (Fig. 2a, entries 10 and 11). Additionally, other nickel catalysts could also work for this protocol, albeit with lower efficiency (Fig. 2a, entries 12 and 13).

Fig. 2: Investigation of reaction parameters.
figure 2

a, Effect of the reaction parameters, conditions: 1 (0.2 mmol, 1.0 equiv.), 2 (0.3 mmol, 1.5 equiv.), [Ir-F] (1 mol%), Ni catalyst (8 mol%), PPh3 (2.5 equiv.), pyridine (2.0 equiv.), and CH3CN (1 ml), argon, 450 nm, 24 h. 1H NMR spectroscopic yield with dibromomethane as internal standard. Isolated yield is given in parentheses. LED, light-emitting diode; ND, not detected. b, Sensitivity assessment for high levels of reproducibility. c, Additive-based robustness screen for the test of functional group tolerance.

Full size image

Sensitivity assessment

A condition-based sensitivity screening approach was carried out (Fig. 2b)64, which demonstrates that the reaction of synthesizing compound 3 is sensitive to high oxygen concentration, low temperature and water content. Pleasingly, good efficiency of this transformation was observed at high temperature and light intensities (Supplementary Section 4).

Additive-based robustness screen

The robustness and the functional group tolerance of this radical two-component thioesterification protocol were investigated by applying an intermolecular additive-based screening method65. Most of the examined additives could be well tolerated, and high reaction yields were observed. These findings highlight the remarkable mildness and tolerance of our methodology (Fig. 2c and Supplementary Section 5).

Mechanistic investigations

To provide mechanistic details to support the proposed catalytic cycle, Stern–Volmer luminescence quenching experiments were carried out. As shown in Fig. 3a, the excited photocatalyst could be quenched by either the PPh3 or DMDS (2), which supports our hypothesis. Additionally, UV/Vis absorption experiments were also performed (Supplementary Fig. 5). Then, this multi-catalytic process was completely inhibited in the presence of TEMPO (tetramethylpiperidine-1-oxyl), revealing that a radical process might be involved (Fig. 3b). Thiomethyl radical was trapped by the (2-cyclopropylallyl)benzene (4), and no acyl radical 8 was observed. This probe experiment further suggests the intermediacy of radical in this system (Fig. 3c). Subsequently, control experiments show that the sulfur radical could be formed from DMDS (2) in absence of nickel catalyst or PPh3 (Fig. 3c). Furthermore, various triplet sensitizers were evaluated (Fig. 3d). The yield of the methylthiolated product correlates to the triplet energy rather than the redox potential of the photocatalysts, which means that the production of thiomethyl radicals from disulfides is probably promoted by EnT60,66.

Fig. 3: Mechanistic investigations for the synthesis of thiomethyl esters by this protocol.
figure 3

a, Stern–Volmer fluorescence quenching analysis. b, Radical inhibitor and probe experiments. c, Control experiments. d, Comparison of various triplet sensitizers. e, Transient spectroscopy of [Ir-F]. f, Transient spectroscopy of [Ir-F] with DMDS (2). g, Transient spectroscopy of [Ir-F] with PPh3. h, Transient spectroscopy of [Ir-F] with 2 and PPh3. GC–MS, gas chromatography–mass spectrometer; PC, photocatalyst.

Full size image

Transient absorption spectroscopic studies

Nanosecond transient electronic absorption spectroscopy (ns-TEAS) was employed to obtain further insight into the kinetics of the initial activation of PPh3 and DMDS (2) by the [Ir-F] (Fig. 3 and Supplementary Section 6.7). The ns-TEAS results of the sole [Ir-F] as well as mixtures of [Ir-F] with DMDS 2, [Ir-F] with PPh3, and [Ir-F] with DMDS 2 and PPh3 in Ar-saturated MeCN upon excitation at λexc = 387 nm are displayed in Fig. 3. Global and target analysis of the data was performed using the python package KiMoPack, which yielded the species-associated difference spectra (SADS, bottom panel)67. The pure [Ir-F] (Fig. 3e) exhibits a broad excited state absorption band peaking at λprobe = 465 nm, which is ascribed to the triplet metal-to-ligand charge-transfer/ligand-centred (3*MLCT/LC) state of the photocatalyst60. Addition of 2 to the solution (Fig. 3f) leads to a notable shortening of the 3*MLCT/LC lifetime and to the formation of a second species with distinct absorption peaks at λprobe = 400, 440 nm that has previously been assigned to the triplet state of 2 (3*2), which is generated from triplet–triplet energy transfer60,68. The mixture of [Ir-F] and PPh3 (Fig. 3g) shows the accelerated depletion of the 3*MLCT/LC state of the photosensitizer in favour of a species with a broad absorption band centred at λprobe = 520 nm indicates the excited state ET leading to the formation of the radical cation PPh3+⦁ as well as to the reduction of the ([Ir-F]II) photosensitizer67,68,69,70,71. The mixture of all three compounds (Fig. 3h) mostly resembles the photodynamics resulting from excited state ET between the [Ir-F] and 2. However, some superimposed triplet–triplet energy transfer contribution caused by the interaction of the photocatalyst with PPh3 can be identified from the decomposed spectra. In summary, the ns-TEAS results provide evidence that the initial activation of the PPh3 to PPh3+⦁ is due to one-electron oxidation from the 3*MLCT/LC state of the photocatalyst and that the previously observed TTEnT between the photocatalyst and the disulfide (2) also occurs in the presence of PPh3.

DFT studies

To gain more mechanistic insights, density functional theory (DFT) calculations were performed to investigate the radical methyl thioesterification of carboxylic acids. Firstly, computational studies suggest that the activation of PPh3 through the single ET from thiomethyl radical is endergonic by 28.9 kcal mol−1 (Supplementary Fig. 11), indicating the SET pathway is thermodynamically unfeasible, which further supports the rationality of activation of PPh3 by excited state photosensitizer72,73,74,75. With the proposal of the activation of PPh3 and thiomethyl radicals, we paid attention to the detailed reaction pathway of methyl thioesterification of carboxylic acids, especially the C(acyl)–O activation mechanism.

As shown in Fig. 4a, the combination of carboxylate anion with PPh3 radical cation, which leads to the formation of complex 7, is exergonic by 6.9 kcal mol−1. The calculated Mulliken atomic spin population reveals that complex 7 is a phosphorus-centred radical species as the phosphorus atom has the largest spin density (0.55). From 7, the C(acyl)–O activation can be achieved through the β-scission of phosphorus radical 7 (via TS-1), which generates the acyl radical 8 and triphenylphosphine oxide. This metal-free pathway requires an activation free energy of 17.8 kcal mol−1.

Fig. 4: DFT studies for this photochemical nickel-catalysed synthesis of thiomethyl esters.
figure 4

a, Metal-free C(acyl)–O activation through the β-scission of phosphorus-centred radical 7. b, Free energy profile of Ni-promoted radical methyl thioesterification of carboxylic acids. c, Ni(I)-mediated C(acyl)–O activation in phosphorus radical 7. The purple numbers denote the Mulliken atomic spin densities at certain atoms. All energies were in kcal mol−1 and were calculated at the M06/6-311 + G(d,p)-SDD/SMD(acetonitrile)//B3LYP-D3(BJ)/6-31G(d)-LANL2DZ level of theory. The bond lengths shown in 3D structures are in angstrom. R = Ph(CH2)3.

Full size image

In the presence of the nickel catalyst (Fig. 4b), nickel(I) bromide 10 can be used to stabilize the thiomethyl radical by forming a triplet thiomethyl nickel(II) complex 311. The corresponding singlet structure is less stable than the triplet structure by 7.8 kcal mol−1. Subsequently, thiomethyl nickel(II) complex 311 can further trap the phosphorus radical 7 and forms a stable doublet nickel complex 12. As shown in Fig. 4b, most of the spin is located at the nickel centre (0.97). Moreover, natural population analysis (NPA) of 12 reveals that the NPA charge of the [Br-Ni-SMe] fragment is −0.97. These results support that the metal centre in complex 12 is nickel(I). We surmise that the SET between thiomethyl nickel(II) complex and phosphorus radical 7 can occur rapidly, thereby resulting in the single-electron reduction of nickel(II) to nickel(I). The energy barrier of the SET76 between thiomethyl nickel(II) complex and phosphorus radical 7 is only 0.9 kcal mol−1 (Supplementary Section 7), which verifies the facile single-electron reduction of nickel(II) to nickel(I) can occur rapidly. Considering the free energy of nickel(I) complex 12 with respect to 10 is –26.0 kcal mol−1, we can conclude that the successive trapping of thiomethyl radical and phosphorus radical using nickel(I) bromide 10 is thermodynamically feasible and irreversible.

From 12 (Fig. 4b), the C(acyl)–O activation can be achieved through the C–O oxidative addition to the nickel(I) centre (via TS-2). The activation free energy is 9.9 kcal mol−1, which is 7.9 kcal mol−1 lower than that of TS-1 in the metal-free C(acyl)–O activation pathway. Subsequent C–S reductive elimination from the acyl nickel(III) complex 13 (via TS-3) is demonstrated to be a barrier-less step. The final ligand exchange with two molecular PPh3 can release the thiomethyl ester product 3 and regenerate the nickel(I) bromide 10. Besides the metal-free and two-electron oxidative addition mechanisms for C(acyl)–O activation, Ni(I)-mediated radical C(acyl)–O activation mechanism has also been considered in DFT calculation (Fig. 4c). An open-shell singlet transition state TS-4 containing a three-membered cyclic core structure was located. Vibrational frequency calculation and intrinsic reaction coordinate calculation of TS-4 suggest that this process is analogous to nickel(I)-mediated β-scission of phosphorus radical. However, the higher activation free energy (26.2 kcal mol−1) indicates this radical pathway is disfavoured. Therefore, the radical methyl thioesterification of carboxylic acids prefers to occur through successive radical trapping with nickel(I) complex, C(acyl)–O oxidative addition, and C–S reductive elimination. The nickel catalyst plays a critical role in tuning the stability and reactivity of different radical species and promoting the C(acyl)–O activation.

Scope for thioesterification

Subsequently, we examined the substrate scope under the optimal condition (Tables 1 and 2). Firstly, primary carboxylic acids were evaluated in their reactions with DMDS (2) (Table 1). Good yields of the corresponding products were observed under standard condition (3 and 15–18, 50–92% yields). Additionally, secondary carboxylic acids were tested (19–25, 56–92% yields). The substrates containing electron-rich aromatic ring (19, 56% yield), keto group (23, 60% yield) and cyclic alkyl groups (21, 22, 24, 68–92% yields) were compatible with this radical system. Importantly, tertiary carboxylic acids could also react smoothly with DMDS (2) to offer desired products in moderate to good yields (26–28, 54–78% yields), exemplifying the generality of this methyl thioesterification protocol. Then, aryl carboxylic acids with either electron-donating groups (30–32, 36, 37, 60–75% yields) or electron-withdrawing groups (34 and 35, 50% and 67% yields, respectively) on the aromatic ring were successfully converted to thiomethyl esters. It is worth mentioning that the efficiency of the reaction was not impeded by ortho substituents on the aromatic ring (38 and 39, 60% and 50% yields, respectively). The transformation of heterocyclic carboxylic acids successfully occurred (40–44, 57–91% yields). With respect to disulfides as coupling partners, thiophene (45, 98% yield), aryl (46 and 47, 90% and 96% yield, respectively) and alkyl disulfides (49–51, 69–73% yields, respectively) were also amenable to this reaction, providing the corresponding products in good to excellent yields and highlighting the utility of this thioesterification protocol. Pleasingly, sec-butyl disulfide and isopropyl disulfide could react well with 4-phenylbutyric acid (1) to forge compounds 52 and 53 in 86% and 95% yields, respectively. Unfortunately, in the case of sterically demanding disulfides, such as tert-butyl disulfide, no corresponding product was obtained.

Table 1 Scope for thioesterification
Full size table
Table 2 Scope for drugs and amino acids
Full size table

Scope for drugs and amino acids

Encouraged by the above results, drug molecules containing the carboxylic group and amino acids were examined. As shown in Table 2, a series of non-steroidal anti-inflammatory drugs, such as ibuprofen (54, 70% yield), ketoprofen (55, 86% yield), flurbiprofen (56, 73% yield), zaltoprofen (57, 80% yield) and loxoprofen (58, 70% yield) could react with DMDS (2) to give the desired products in good yields. Interestingly, citronellic acid was also compatible with this system, offering the desired product in 50% yield (59). The drug, containing primary carboxylic acid, was well tolerated (60, 76% yield). Probenecid, which is mainly used to treat gout and hyperuricaemia, was converted to its thiomethyl ester in 69% yield (61). Additionally, various α-, β- and γ-amino acids were evaluated under the standard condition. Thiomethylated products could be obtained using Boc-protected (62, 65% yield) and Cbz-protected (63–65, 52%, 74%, 83% yields) α-amino acids as reaction partners. β-Amino acids featuring azetidine (67), piperidine (69) and pyrrolidine (70) rings were introduced in this protocol, and delivered the corresponding products in 60–85% yields. The successful reactions of other β-amino acids further demonstrate the functional group tolerance of this radical approach (68, 71 and 72, 46–80% yields). γ-Amino acids could also be successfully reacted, and formed the desired product in 55% yield (73), which shows the compatibility of this method with a wide range of amino acids.

Gram-scale synthesis

To evaluate the scalability of this photochemical protocol, selected reactions were carried out on a larger scale (Table 2 and Supplementary Section 8). Probenecid (74) and an azetidine-based amino acid (75) were used as substrates, respectively. Pleasingly, desired products 61 and 67 could be successfully obtained in 70% yield (0.88 g) and 82% yield (0.87 g), respectively.

Application of product

We next turned our attention to the potential applications of thiomethyl esters (Supplementary Section 9). Compound 61 was submitted to the conditions presented in reported work and provided other valuable compounds77,78,79. For examples, 96% yield of compound S-76 could be generated via decarbonylation of compound 61. In addition, alkynylation (S-77, 62% yield) and boronation (S-78, 92% yield) were also successfully achieved using compound 61 as coupling partner, demonstrating that thiomethyl esters are functional synthetic precursors in a series of chemical transformations.

Conclusion

In summary, we have reported a nickel-catalysed, sensitized ET strategy that features high chemoselectivity, a broad substrate scope, mild reaction conditions and good functional group tolerance. Importantly, various drugs and amino acids were successfully functionalized using this system. The facile scalability and the synthetic utility of this radical protocol were demonstrated by the gram-scale synthesis and application of products, respectively. Mechanistically, the successful amalgamation of photoinduced ET, energy transfer and nickel catalysis plays an essential role in this two-component radical deoxysulfurization. Therefore, we anticipate that this combination of multiple catalytic systems would have diverse applications in synthetic chemistry and beyond.

Methods

General procedure for synthesis of methyl thioesters

A dried 5 ml Schlenk tube was charged with PPh3 (0.5 mmol, 2.5 equiv.), carboxylic acid (0.2 mmol, 1.0 equiv.), NiBr2(diglyme) (8 mol%) and Ir[dF(CF3)ppy2(dtbbpy)][PF6] (1.0 mol%). Then, reaction mixture was degassed by vacuum-argon purging. After the mixture was thoroughly degassed and filled with argon, the DMDS (2, 0.3 mmol, 1.5 equiv.), pyridine (0.4 mmol, 2.0 equiv.) and CH3CN (1 ml) were added to the vessel under argon atmosphere. Then, the Schlenk tube was sealed tightly and stirred under irradiation with light (λmax = 450 nm) for 24 h. The solvent was removed in vacuo after the reaction. The crude residue was purified by flash column chromatography on silica (n-pentane/EtOAc mixtures).

General procedure for synthesis of other thioesters

A dried 5 ml Schlenk tube was charged with PPh3 (2.5 or 3 equiv.), carboxylic acid (1, 0.2 mmol, 1.0 equiv.), NiBr2(diglyme) (8 mol%) and Ir[dF(CF3)ppy2(dtbbpy)][PF6] (1.0 mol%). Then, reaction mixture was degassed by vacuum-argon purging. After the mixture was thoroughly degassed and filled with argon, disulfide (0.3 mmol, 1.5 equiv.), pyridine (0.4 mmol, 2.0 equiv.) and CH3CN (1 ml) were added to the vessel under argon atmosphere. Then, the Schlenk tube was sealed tightly and stirred under irradiation with light (λmax = 450 nm) for 24 h. The solvent was removed in vacuo after the reaction. The crude residue was purified by flash column chromatography on silica (n-pentane/EtOAc mixtures). Note: PPh3 (3.0 equiv.) was necessary with the benzyl and alkyl disulfides as substrates. Compounds 52 and 53 were synthesized by using [Ir-F] (2 mol%), disulfide (0.4 mmol) and PPh3 (0.8 mmol) without pyridine.