A general method for site-selective Csp3-S bond formation via cooperative catalysis.

Herein, we report a copper-catalysed site-selective thiolation of Csp3-H bonds of aliphatic amines. The method features a broad substrate scope and good functional group compatibility. Primary, secondary, and tertiary C-H bonds can be converted into C-S bonds with a high efficiency. The late-stage modification of biologically active compounds by this method was also demonstrated. Furthermore, the one-pot preparation of pyrrolidine or piperidine compounds via a domino process was achieved. This journal is © The Royal Society of Chemistry.

Introduction

The sp3 carbon–sulfur bond plays a pivotal role in modern organic synthesis, natural products and pharmaceuticals. For example, Butoconazole is an imidazole antifungal used in gynaecology, and Nelfinavir is an antiretroviral drug which was used in the treatment of the human immunodeficiency virus (HIV) (Scheme 1). Clinical trials demonstrated that the introduction of Csp3–S moieties could improve their biological activities, such as anti-tumour, anti-inflammatory and immunomodulatory properties.[1] Besides that, the Csp3–S bond can also be found in the natural amino acid, such as cysteine, and facilitate the metabolism process of the related protein in the organism.[2] Given the importance of Csp3–S containing compounds, methods that can directly assemble the Csp3–S bond through direct C–H functionalization with a high efficiency and selectivity are highly demanded. Moreover, these methods could facilitate the late-stage functionalization of complex molecules that enable the rapid diversification and construction of a library of molecule analogues.

Scheme 1

Pharmaceutical molecules containing sp3 C–S thiol moieties.

Csp3–S bonds are usually constructed by nucleophilic substitution of the alkyl halides with metal thiolate (Scheme 2a). Also, similar to other kinds of nucleophilic substitution reactions, these methods are sensitive to the sterics.[3] With the development of transition-metal catalysis, more efficient protocols to introduce the Csp3–S bond were realized via the transition-metal catalysed cross-coupling of organo halides/alkene and thiols/thioethers.[4] For example, Li reported the palladium catalysed cross-coupling of aryl halides and thiols; however, only primary and secondary thiols were shown to be suitable substrates.[5] Furthermore, the preparation of alkyl halide substrates was tedious and was not step- and atom-economic (Scheme 2b). Alternatively, hydrothiolation of alkenes has been developed to construct the Csp3–S bond. However, pre-functionalization of the substrates is necessary, and the regioselectivity greatly depends on the structure of the substrate (Scheme 2c).[6] Recently, the direct thiolation via the alpha-sp3 C–H bonds of ether and amine was achieved by Li,[7] Yuan and Xiang,[8] independently (Scheme 2d). The thiolation of Csp3–H was also developed, facilitated by a directing group strategy (Scheme 2e).[9] However, to date, a general method to realize the C–H bond thiolation remains a significant challenge.

Scheme 2

The strategies to construct the Csp3–S bond: (a) Csp3–S bond formation via nucleophilic substitution. (b) Transition-metal catalysed cross-coupling. (c) Catalytic addition to unsaturated double bond. (d) Direct Csp3–H bond thiolation. (e) Directing group assisted Csp3–H bond thiolation. (f) Site selective Csp3–S bond formation.

Inspired by the Hofmann–Löffler–Freytag (HLF) reaction,[10] C–H bond functionalization via a radical pathway provides a new route to construct different types of targeted molecules.[11] Elegant related works have been documented by Suárez,[12] Muñiz[13] and Nagib[14] independently by using the stoichiometric amounts of iodine, catalytic iodine, and triiodide for the HLF type reaction. More recently, a breakthrough of metallaphotoredox-catalysed remote functionalization of amides was demonstrated by the groups of Knowels,[15] Rovis,[16] and Meggers.[17] In addition, Cu-modified HLF-type reactions with a simple NH substrate are particularly valuable in primary C–H bond amination.[18] Although the generated amidyl radical induced 1,5-HAT (HLF reaction) process is well established, further transformation of C-centred radicals is limited to the halogenation, cyclization or interceptions by alkenes or heteroaryls.[19] The metal mediated C-centred radical resulted in the functionalization of unactivated Csp3–H bond selectively and efficiently, a great challenge. We proposed that by using a N–F precursor, the low tendency toward N–F homolysis and instability of the fluorine radical, can enable the 1,5-HAT process to facilitate Csp3–H bond functionalization over C–F bond formation. Previously, and independently, Zhu[20] and Nagib[21] have reported a remote arylation of an aliphatic amine via activation of the N–F precursors. Very recently, Muñiz demonstrated a cyclization reaction of an aliphatic amine via N–F bond activation by a copper catalyst, albeit under a high temperature.[22] Li has successfully achieved a copper catalysed trifluoromethylation by using N–F precursors.[23] Given the importance of thiol moieties in functional molecules, here we describe a protocol for the site-selective thiolation of aliphatic amines through the combination of light-driven and copper-facilitated 1,5-hydrogen atom transfer (Scheme 2f). This strategy enables the transformation of primary, secondary and tertiary C–H bonds into C–S bonds with a high efficiency. Meanwhile, a late-stage modification of biologically active compounds was demonstrated to show the synthetic utility of such a methodology.

Results and discussion

To initiate our proposal, N-fluoro-tosylamide 1a and phenyl disulfide 2a were chosen as the model substrates. After extensive optimization, we found that a cocktail containing 10 mol% of Cu(acac)2, 10 mol% of 1,10-phenanthroline (L1), 1.4 equivalent Na2HPO4 and 1.7 equivalent of indium powder as additives could promote the desired δ thiolated product 3a in 75% isolated yield in the presence of blue LED. Control experiments were conducted to determine the role of each ingredient (Table 1, entries 2–5). Without a copper catalyst or ligand, the yield decreased significantly (Table 1, entries 2 and 3). The desired product was only observed in 36% yield, indicating the pivotal role of light in this transformation (Table 1, entry 4). Moreover, a lower yield was obtained in the absence of indium powder and base (Table 1, entry 5). Other copper complexes were also tested in this reaction, resulting in low yields (Table 1, entries 6–8). The aliphatic amines are generally regarded as good reducing reagents in photoredox reactions. However, the addition of triethylamine inhibited this reaction (Table 1, entry 9). To determine the role of copper complexes in this transformation,[24] we replaced the copper with an Ir complex or Ru complex, and no desired results were obtained, indicating the role of copper is more than as a photoredox catalyst in this reaction (Table 1, entries 11 and 12).

Effect of reaction parametersa

Entry	Variations from “standard” conditions	Yieldb (%)
1	None	78 (75)
2	Without Cu(acac)2	27
3	Without L1	13
4	Under dark	36
5	Without In powder and Na2HPO4	65
6	Cu(OTf)2 instead of Cu(acac)2	47
7	cat.[Cu]1c instead of Cu(acac)2	31
8	cat.[Cu]2d instead of Cu(acac)2	20
9	Et3N instead of Na2HPO4	0
10	2,6-Lutidine instead of Na2HPO4	19
11	Ir(ppy)3 instead of Cu(acac)2	10
12	Ru(1,10-phen)3Cl2 instead of Cu(acac)2	7

Reaction was conducted on a 0.1 mmol scale.

Yields were determined by 1H NMR analysis versus 1,1,2,2-tetrachloroethane as the internal standard. Isolated yield in parentheses.

cat.[Cu]1 is Cu(CH3COCHCOCF3)2.

cat.[Cu]2 is Cu(CF3COCHCOCF3)2.

With the optimal conditions in hand, we examined the scope of different disulfides with N-fluoro-tosylamide 1a (Table 2). The substituted disulfides with electron-donating groups, such as methyl (3b), methoxyl (3c) and acetoxyl (3d), were all well tolerated. Moreover, the halide-substituted aryl disulfides also underwent this thiolation reaction (3e–3i), even with multi-halides or fluoro substituents. Remarkably, the electron-withdrawing nitro-group on the disulfide led to a moderate drop in reactivity (3j, 3k). Interestingly, the thiophenyl group generally has special properties in optoelectronic materials, and can afford the desired product in an acceptable yield (3m). Furthermore, to our satisfaction, the benzyl disulfide can give the δ benzyl thioether in a slightly compromised yield (3n, 45%). Interestingly, the δ-selenation could also be achieved when using diselenide instead of disulfide (3o).

Substrate scope of disulfidesa

Unless otherwise noted, the reaction conditions were as follows: 1a (0.1 mmol), 2 (1.2 equiv.), Cu(acac)2 (10 mol%), 1,10-phen (10 mol%), In powder (1.7 equiv.), Na2HPO4 (1.4 equiv.), DCE 0.6 mL, blue LED (50 W), 40 °C, 9 h.

80 °C.

This C–H thiolation protocol was also successfully applied to functionalize a wider range of electronically and sterically diverse primary, secondary and tertiary aliphatic C–H bonds (Table 3). The thiolation yield for primary C–H bond is slightly increased with a N-fluorotosylamide substrate containing a substituent at the α-position (4a, 4b). N-Fluorocarboxamides are also good substrates for this transformation (4c, 4d). Notably, functionalization of the secondary C–H bond afforded a series of thiolether derivatives selectively. Even with longer carbon chains in the substrates (4e–4g), the reaction occurred only at the δ position. Surprisingly, a gem-dialkyl group on the linear N-fluoroamide substrates at the β-position does not improve the product yield (4f). The biologically relevant functionalities, such as Nphth (4h), carboxylic ester (4i) and OTBDMS (4j, 4k), as well as OBz (4l) are well tolerated. Specifically, exclusive δ-selectivity was observed in the case of 4h and 4j bearing a heteroatom near the reaction centres. Cycloalkanes with various ring sizes ranging from four to seven carbons were thiolated in good yields (4m–4q, 4s–4t). The adamantyl group can be functionalized selectively on secondary C–H bonds rather than tertiary ones (4r). Obviously, in heteroatom-containing cycles, functionalized products were obtained at the α-position with a better diasteroselectivity for the six-member ring than the five-member ring (4u, 4v). Disappointingly, the steric hindered substituents at the α- or β-position of N-fluoroamides led to a poor diasteroselectivity (4w–4y). Both the 1,5-HAT and 1,6-HAT processes occurred in a substrate which involved a tertiary C–H bond at the ε-position (4z). Finally, this transformation can be successfully applied to construct the quaternary carbon centre with a thiol group, highlighting the synthetic utility of this methodology (4aa, 4ab).

Selective thiolation of aliphatic amidesa

Unless otherwise noted, the reaction conditions were as follows: 1 (0.1 mmol), 2 (1.2 equiv.), Cu(acac)2 (10 mol%), 1,10-phen (10 mol%), In powder (1.7 equiv.), Na2HPO4 (1.4 equiv.), DCE 0.6 mL, blue LED (50 W), 40 °C, 9 h.

dr was determined by crude 1H NMR.

Having demonstrated the applicability of this remote Csp3–H thiolation protocol, late-stage functionalization of biologically active molecules was carried out (Table 4). The amino alcohol derivatives can be coupled with varieties of disulfides, which greatly demonstrated the powerful utility of this method (5a–5g). These functionalized amino alcohols have a great potential in the synthesis of biological compounds. A non-symmetric thiolation of the methyl group occurred on the menthol derived N-fluoroamide substrate (5h). It's important to note that pharmaceutical molecular (5i) and complex natural products (5j, 5k) can be readily modified with excellent regioselectivity.

Selective thiolation of biologically active compoundsa

Unless otherwise noted, the reaction conditions were as follows: 1 (0.1 mmol), 2a (1.2 equiv.), Cu(acac)2 (10 mol%), 1,10-phen (10 mol%), In powder (1.7 equiv.), Na2HPO4 (1.4 equiv.), DCE 0.6 mL, blue LED (50 W), 40 °C, 9 h.

dr was determined by crude 1H NMR.

Furthermore, to illustrate the utility of this new transformation with respect to constructing drug scaffolds, we tested a type of substrates which can proceed by a domino process involving the 1,5-HAT followed by cyclization and sequence thiolation, affording thiol substituted pyrrolidine or piperidine derivatives (Scheme 3).

Scheme 3

Preparation of pyrrolidine or piperidine derivatives via remote thiolation.

To gain insight into the reaction mechanism, competition experiments between 1°/2° C–H bonds, 2°/3° C–H bonds and 1°/3° C–H bonds were carried out independently (Scheme 4a). The total reactivity of the hydrogen atoms decreased in the order tertiary > secondary > primary. Furthermore, to test the selectivity of thiolation reagents, the mixed disulfides were subjected to this transformation (Scheme 4b). For an alkyl phenyl disulfide, only the phenyl thioether product was observed. For a mixed aryl disulfide with an electron-donating group OMe and an electron-withdrawing group CF3, the thiolation occurred only with the electron-donating OMe substituted thiol group.[25] These selectivities may result from the electron density on different sulfur moieties. Based on this selectivity, we developed a direct cross-coupling between two biologically interesting molecules to give the corresponding conjugated compounds (Scheme 5a). Subsequently, the thioether group can be directly oxidized to a sulphone group, which broadly exist in pharmaceuticals (Scheme 5b).

Scheme 4

Control experiments: (a) competition experiments; (b) reactivity of mixed disulfides.

Scheme 5

Late-stage functionalization of biological active compounds: (a) cross-coupling with biological active compounds; (b) oxidized thioether to sulphone.

Moreover, radical control experiments were carried out to uncover evidence for the presence of radical intermediates. Radical-clock experiments were conducted (Scheme 6a), the observation of products 15a, 17a and 19a indicating the formation of a carbon-centred radical in the transformation. Especially for 19a, the observed linear selectivity deduced that a η1- or η3 metal allylic intermediate maybe involved in this process. Subsequently, the radical process was further demonstrated by the results in the presence of TEMPO and BHT. Furthermore, the experiments of PhSH and In(SPh)3 were conducted to test the possible active species of this thiolation reaction. The results indicated that by using PhSH, 20% of the desired product was obtained, while no desired product was observed by using In(SPh)3 as the thiolation reagent (Scheme 6b). Finally, blank experiments were carried out to confirm the effective role of each additive in this transformation (details in the ESI†).

Scheme 6

Mechanistic investigation: (a) radical clock reactions; (b) comparative experiments.

Based on these results, we propose that this transformation is preferred by a light-driven, copper initiated 1,5-HAT process (Scheme 7). Ligand exchange between an in situ generated Cu(i) catalyst and disulfide occurred to generate Cu(i)–I species, which could be excited by visible light followed by reducing the N–F amide A to afford the N-centred radical B and Cu(ii)–II intermediate. The Cu(ii)–II released one molecular of F-SAr[26] through a metathesis process to afford the Cu(ii)–III. The selective formation of F-SAr is likely to result from the coordination ability of the sulfur group. The electron-rich sulfur group coordinated to form the Cu(i)–I species. This selectivity is in conjunction with the result of the mixed disulfide. Translocation of the N-centred radical B to the C-centred radical Cvia 1,5-HAT occurred, and then the δ C-radical was intercepted by the Cu(ii)–III to form the Cu(iii)–IV intermediate. Finally, the reductive elimination of Cu(iii)–IV gave the desired product D and regenerated active Cu(i) catalyst. Furthermore, efforts to expand the scope of this transformation are underway.

Scheme 7

Proposed mechanism.

Conclusions

In summary, we have developed a light-driven, copper-catalysed site-selective thiolation of Csp3–H bonds for aliphatic amines. Primary, secondary and tertiary C–H bonds can all be successfully converted into C–S bonds. The broad amine scope, good functional compatibility, and late-stage modification of biologically active intermediates underscore the unique synthetic potential of this method.[27]

Conflicts of interest

The authors declare no conflicts of interest.