Introduction

Sulfenamides are a class of divalent sulfur-derived scaffolds featuring an S-N bond1. For many years, they have been found to be a superior cross-linking agent to elemental sulfur in rubber production, providing greater operational safety and higher cross-linkage yields2,3,4. In the field of medicinal chemistry, sulfenamides have been identified as the active metabolites of proton pump inhibitors in the treatment of acid-related gastrointestinal diseases5,6. Other uses of sulfenamides in industry include load-capacity improvers in lubricants7, wood preservatives8, and insecticides in agricultures9. From the standpoint of organic chemistry, they continue to elicit interest due to their utility as protecting groups in peptide synthesis and their stereochemical properties arising from hindered rotation about the S–N bond1. Very recently, we and other groups have exploited sulfenamides as a versatile reagent to prepare other organosulfur pharmacophores with higher oxidation states, such as S(IV)-derived sulfilimines and S(VI)-derived sulfoximines10,11,12,13,14,15,16. Notably, C–S bond forming pathways override the alternative C–N bond formation processes either in the presence of transition-metal catalysts or under transition-metal-free conditions.

Considering the unique structural features of sulfenamides, and their widespread applications in academia and industry, much effort has been devoted to their preparation. Conventionally, synthesis of sulfenamides predominantly relies on the construction of S–N bonds via a nucleophilic attack from amines to disulfides, sulfenyl halides or their surrogates (Fig. 1A, a, left)17. Moreover, dehydrogenation of amines and thiols under oxidation conditions could lead to the formation of S–N bonds, though condensation of thiols to disulfides often competes (Fig. 1A, a, right)18. Alternatively, direct arylation of NH-sulfenamides represents a straightforward and step-economical pathway to afford N-aryl sulfenamides (Fig. 1A, b). Despite this appeal, C–N arylation of NH-sulfenamides has not been explored previously, to the best of our knowledge, primarily due to the fact that transition-metal-catalyzed or transition-metal-free functionalization of NH-sulfenamides prefers to take place on the sulfur site rather than the nitrogen site10,11,12,13,14,15,16, even though N-functionalization is thermodynamically favored11. Furthermore, the S–N bond of sulfenamides is notoriously labile, as it is prone to hemolysis under thermal or photo-induced conditions19 or heterolysis in the presence of nucleophiles or electrophiles (Fig. 1B, b)20. In addition, sulfur(II)-derived sulfenamides readily convert to the sulfur(IV)-derived sulfinamides or sulfur(VI)-derived sulfonamides upon exposure to oxidants (Fig. 1B, b)21,22. To efficiently construct the C–N bond while retaining the fragile S–N bond of sulfenamides, Chan–Lam coupling represents an appealing arylation protocol owing to inexpensive and abundant copper catalysts, mild reaction conditions, broad functional group tolerance, and neutral pH conditions23,24. Our group has pioneered a copper-catalyzed Chan–Lam type coupling of sulfenamides assisted by an acyl-based protecting group on nitrogen in the absence of an external ligand, and a variety of sulfilimines have been prepared11. The chemoselectivity favoring less thermodynamically stable product arising from C–S bond formation was attributed to the bidentate sulfenamide coordination in the transmetalation event. Very recently, we have introduced an enantioselective copper-catalyzed Chan–Lam type S-arylation of N-aryl sulfenamides with arylboronic acids to furnish chiral sulfilimines, in which the chemoselectivity favoring C–S bond as well as the enantioselectivity was steered by a bidentate pyridyl oxazolidine ligand15. Subsequently, Yang and coworkers disclosed that transition-metal-free arylation of N-aryl sulfenamides with diaryliodonium salts occurs on the sulfur atom even without an acyl directing group on nitrogen25. To achieve C–N bond formation of sulfenamides via a Chan–Lam coupling by overriding the kinetically favored C–S bond formation, we envisioned a ligand-controlled Chan–Lam coupling protocol for two major reasons: (1) Considering that most Chan–Lam couplings occur readily without a ligand, generation of a ligand-coordinated copper species would prevent background S-arylation by non-ligated copper complexes; (2) A multi-dentate ligand could block S,N-bis-chelation of sulfenamides to copper center, thereby forcing only N-binding, which gives rise to a suitable intermediate for C–N bond reductive elimination. Herein, we report a ligand-controlled Chan–Lam coupling of sulfenamides with arylboronic acids to provide facile access to a variety of N-arylated sulfenamides with high level of chemoselectivity favoring C–N bond formation over C–S bond formation (Fig. 1B, a).

Fig. 1: Strategies and Challenges in Synthesis of Sulfenamide.
figure 1

A Synthetic Approaches to Sulfenamides. a Classic Synthetic Strategies to Sulfenamides. b This Work: Synthesis of N-Arylated Sulfenamides by Chan–Lam Coupling. B Conceptual Design and Major Challenges of Our Approach. a Conceptual Design of Our Approach. b Potential Challenges of Our Approach.

Full size image

Results

Reaction Optimization

We initiated the investigation by employing S-(4-fluorophenyl)-N-(p-tolyl)thiohydroxylamine (1a) and p-tolylboronic acid (2a) as substrates (Table 1). After a series of optimizations (see Supplementary Tables 1-5), the optimal conditions for Chan–Lam coupling of sulfenamides were determined to be: sulfenamide 1a as the limiting reagent, boronic acid 2a (2.0 equiv) as coupling partner, Cu(TFA)2•H2O (10 mol %) as catalyst, pybox L326 (20 mol %) as ligand, Cy2NMe (1.5 equiv) as base, in MeCN (0.3 M) at room temperature for 24 h under an O2 atmosphere. These conditions provided an 87% assay yield of the desired product 3aa, with an 84% isolated yield (Table 1, entry 1). Notably, only 5% of the alternate S-arylation product was observed. The electron-donating p-OMe in ligand L3 was beneficial to the transformation, as parent L2 or L1 bearing an electron-withdrawing p-Cl leads to lower yields of 3aa (entries 2, 3). In control experiments, the copper complex, Cy2NMe, and ligand were found to be essential for the catalytic reaction, as evidenced by no or trace of 3aa in the absence of any of these reagents (Table 1, entries 4-6). Only 8% assay yield of 3aa was observed in the absence of the external-oxidant O2 (Table 1, entry 7). The theoretical yield of 3aa under anaerobic condition should be 5%, but the error of 3% could either be attributed to the error of 19F NMR technique used to determine the assay yield, or caused by the inevitable trace amount of air in the microwave vial. When air was used instead of O2 atmosphere, the assay yield of 3aa dramatically dropped to 37% (Table 1, entry 8).

Table 1 Optimization for Chan–Lam Coupling of Sulfenamide (1a) and p-Tolylboronic Acid (2a).a
Full size table

Substrate Scope

With the optimized conditions in hand, we explored the scope of arylboronic acids in the coupling with N-phenyl-S-(p-tolyl)thiohydroxylamine (1b) (Fig. 2). Arylboronic acids with electron-donating groups, such as p-Me (2a), p-OMe (2b), and p-SMe (2c) were well tolerated, leading to the formation of desired products (3ba-3bc) in 57-87% yields. The coupling reaction proceeded smoothly with arylboronic acids possessing electron-withdrawing groups, including p-F (2d), p-Cl (2e), p-CF3 (2 f) and p-CN (2 g), affording 3bd-3bg in 57-81% yields. Sterically hindered 2-tolylboronic acid (2 h) was compatible with our protocol, providing 3bh in 80% yield. Similarly, meta-substituted arylboronic acids (1i and 1j) reacted well with 1b under the standard condition, providing the desired products (3bi and 3bj) in 86% and 72% yield, respectively. Notably, arylboronic acids equipped with different functional groups, including ketone (1k), ester (1 l), and alkenyl (1 m), could be utilized to generate 3bk-3bm in 54-72% yields. Attesting to the mild reaction conditions and broad functional group tolerance, a myriad of heteroarylboronic acids were found to be compatible under the optimal conditions; the N-pyrimidyl (3bn), benzofuranyl (3bo), thiophenyl (3bp), indolyl (3bq) and quinolinyl (3br, 3bs) sulfenamides were obtained in yields ranging from 40% to 82%, highlighting the expediency of our protocol. The structure of 3bn was unambiguously assigned by X-ray crystallography (CCDC: 2142998, see Supplementary Fig. 1 for details), which confirms that arylation occurs on nitrogen rather than on sulfur.

Fig. 2: Substrate Scopea.
figure 2

aReaction conditions: 1a (0.15 mmol), 2a (2.0 equiv), Cu(TFA)2•H2O (10 mol %), L3 (20 mol %), and Cy2NMe (1.5 equiv) in MeCN (0.5 mL) under O2 at room temperature for 24 h. b35 °C, 24 h. c12 h. dCu(TFA)2•H2O (15 mol %), L3 (30 mol %), 12 h. e35 °C, 2 h. f35 °C, 12 h.

Full size image

Next, the sulfenamide component was evaluated with 2a (Fig. 2). The aryl group on the sulfur of the sulfenamide was especially insensitive to electronic or steric effects. An S-aryl-N-phenyl sulfenamide possessing electron-donating p-OMe group (1c) afforded 3ca in 56% yield. Electron-withdrawing substituents appended to the S-aryl-N-phenyl sulfenamide, such as p-F (1a), p-Cl (1d) and p-NO2 (1e) were well tolerated under the standard conditions to deliver the expected products in good yields. Sterically demanding ortho-substituents, such as o-Me (1 f), o-OMe (1 g), o-Cl (1 h) and o-COOMe (1i), were not detrimental, furnishing 3fa-3ia in a range of 83-90% yields. In particular, the coupling reaction proceeded smoothly with sulfenamide 1j containing two flanking o-Me groups to afford 3ja in 80% yield. In addition, S-aryl sulfenamide (1k) with an m-Br substituent was viable, providing 3ka in 80% yield. Our Chan–Lam coupling protocol was also successful with an S-heteroaryl sulfenamides to provide 3la in 54% yield.

The substituent group on the nitrogen of the sulfenamide was also varied. Aryl groups bearing electron-donating substituents such as p-Me (1 m) or p-OMe (1n) underwent the coupling reaction with 2a to afford 3ma and 3na in 75% and 62% yield, respectively. Electron-withdrawing substituents, including p-F or Cl (1o, 1p), were equally successful leading to 3oa (75%) and 3pa (75%). Aryl groups on the nitrogen bearing sterically demanding ortho-substituents, such as o-Me (1q) and o-F (1r), were compatible with our protocol, delivering 3qa and 3ra in 61% and 85% yield under slightly modified conditions. Again, a range of functional groups on the N-aryl, such as ester (1 s), nitrile (1t), ketone (1 u), sulfonyl (1 v) and amide (1w), were well tolerated to furnish 3sa-3wa in 52-65% yields. Even N-indolyl-S-(p-tolyl) sulfenamide (1x), an example of an N-heteroaryl, could be utilized affording 3xa in 57% yield under slightly modified conditions. Remarkably, challenging N-alkyl sulfenamides, which are usually less stable than their aryl counterparts, were also well-suited under the catalytic conditions. N-Methyl sulfenamide 1 y coupled with 2d to give 3yd in 53% yield under slightly modified conditions. Substrate bearing bulky tert-butyl group on nitrogen (1z) could be well tolerated to provide 3zd in good yield in 12 h. N-Benzyl sulfenamide (1aa) underwent the coupling reaction smoothly, as evidenced by the formation of 3aad in 64%. 2-Nitrobenzyl, a photocleavable protecting group was orthogonal to the coupling protocol affording 3abd in 72% yield, offering a potential site for the downstream functionalization. Of note, heteroaryl groups, including 3-pyridyl (1ac), 2-thiophenyl (1ad), and 2-furanyl (1ae), appended on the benzyl position were compatible, extending the substrate generality of our protocol.

Synthetic Applications

To showcase the potential synthetic applications, further transformations of 3ba were explored (Fig. 3a). After the Chan–Lam coupling between 1b and 2a, direct treatment of the reaction with oxone (2.0 equiv) and diethylamine (40 mol %) in an one-pot fashion led to the formation of sulfinamide 4 in 48% yield (Fig. 3a)27. Likewise, adjusting the amount of oxone to 5.0 equivalents could deliver the corresponding sulfonamide 5 in 45% yield (Fig. 3a)28.

Fig. 3: Synthetic Applications.
figure 3

a Downstream Transformations of Products. b Synthesis of Sulfenamide-Analog of Sulfacetamide. c Synthesis of Sulfenamide-Analog of Oxybuprocaine.

Full size image

To improve poor aqueous solubility and slow dissolution rates of therapeutics containing acidic N-H bonds, Guarino and coworkers have attempted to use sulfenamide-type analogs as a prodrug strategy29. Toward this end, our approach provides a facile tool to directly functionalize sulfenamide-type derivatives of drugs attesting to the mild reaction conditions as well as the broad functional group tolerance, while retaining the labile S-N bond. Sulfacetamide, which is a marketed anti-infective agent in the treatment of conjunctivitis, trachoma and other eye infections30, could be readily transformed to the sulfenamide-derived compound (8) in two steps, 39% overall yield (Fig. 3b). Of note, our Chan–Lam coupling favored the C–N bond formation on the sulfenamide but left secondary amide group intact, highlighting the excellent chemoselectivity. Oxybuprocaine, a short-acting anesthetic for ophthalmology and otorhinolaryngology31, which was synthesized from commercially available 9 in 4 steps, could be effectively decorated with the N-aryl sulfenamide moiety using our Chan–Lam coupling (Fig. 3c).

Mechanistic Studies

A combined experimental and computational study was performed to gain insight into the mechanism of Chan–Lam coupling (Fig. 4–5). Interestingly, even though the chemoselectivity of Chan–Lam coupling of sulfenamides was affected by the substituents on the pybox ligands (Table 1, entries 1-3), the conversion rates using L1–L3 were similar as illustrated by the kinetic studies (Fig. 4a), suggesting the chemoselectivity favoring C-N bond over C-S bond was not determined by the rate-limiting step.

Fig. 4: Mechanistic Studies.
figure 4

a Kinetics. b UV–Vis Spectra. c EPR Spectra.

Full size image
Fig. 5: Computational Mechanistic Studies.
figure 5

a DFT Computational Study: Formation of N-Arylation Product 21’. All Free Energies Were Computed Using UM06/6-311 + + G(d,p)-SDD(Cu)-CPCM(DME)//UB3LYP-D3/6-31 G(d)-SDD(Cu). b Proposed Catalytic Cycle for Chan–Lam Coupling of Sulfenamides.

Full size image

UV-Vis spectra of sulfenamide 1a, boronic acid 2a, and ligand L3 did not exhibit appreciable absorptions above 400 nm as expected, whereas a mixture of Cu(TFA)2•H2O and L3 exhibited a strong absorption at 740 nm, in agreement with the d-d transition of Cu(II) complexes (Fig. 4b). Similarly, a characteristic band was observed at 700 nm upon monitoring the reaction of 1a with 2a indicating that a copper(II) species serves as the resting state of the catalyst during the catalytic cycle. To identify this copper(II) (3d9, S = 1/2) species in the reaction, electron paramagnetic resonance (EPR) spectra were recorded (Fig. 4c). In this case, EPR-silent Cu(I)Tc was used with dioxygen gas to mimic the turnover process to form the Cu(II) species in the catalytic cycle. In doing so, two EPR active Cu(II) species were identified. One species (species Cu-1) corresponds to Cu(II) bound with L3 and 2a in a dx2−y2 ground state, with an axial g tensor of [2.296, 2.054, 2.053] and a hyperfine tensor of A(63Cu) = [510, 35, 35] MHz. In addition, significant superhyperfine coupling with the 14N (I = 1) from the ligand L3 gives rise to a hyperfine splitting signals at the g⊥ ~ 2.054 region. Spectral simulation suggests only two 14N nuclei coupled to the electron spin center Cu(II) (see Supplementary Fig. 2-4 for simulation parameters). However, upon further introduction of 1a, a second species (species Cu-2) was observed, which is also in a dx2−y2 ground state, but with three 14N nuclei coupled to the electron spin center Cu(II). When compared to Cu-1, this result indicates that the third 14N of species Cu-2 most likely arises from 1a binding to Cu(II) via nitrogen. This assignment was confirmed by the EPR isotope response when using 15N labeled 1a (Fig. 4c, Supplementary Fig. 5-6).

To understand the factors affecting chemoselectivity and canonical steps, the mechanism for this transformation with L2 was probed using density functional theory (DFT) [UM06/6-311 + + G(d,p)-SDD(Cu)-CPCM(DME)//UB3LYP-D3/6-31 G(d)-SDD(Cu)32,33,34,35,36,37,38,39,40, see Supplementary Fig. 7-49 for full computational details]. In Fig. 5a, four- and five-coordinate cationic copper complexes were explored computationally based on X-ray crystallographic evidence for similar structures41,42,43. Initially, the arylboronic acid is proposed to form the pre-reacting complex 16’ (Fig. 5a, uphill in energy by 3.5 kcal/mol) from the starting cationic CuII complex 16. This complex then undergoes transmetalation (via [16-17], 9.9 kcal/mol) to form intermediate 17 (−10.3 kcal/mol) observed by EPR spectroscopy (species Cu-1 in Fig. 4c). The boronic acid disrupts the tridentate binding of the pybox ligand via H-bonding with one nitrogen atom of the oxazoline ring, accounting for the observed binding of only two nitrogen atoms in the EPR spectra. Next, sulfenamide exchanges with boronic acid to form 18 (−12.9 kcal/mol) which also has only two nitrogen atoms of the pybox ligand bound to the copper center, consistent with EPR studies (species Cu-2 in Fig. 4c) (for a comparison of the energetics of 18 and its conformer with different methods, see Supplementary Fig. 7).

Intermediate 18 then undergoes deprotonation with NMe3 (as a simplified model for Cy2NEt) at the nitrogen atom of the sulfenamide via [18-19] (overall energetic span of 13.1 kcal/mol) to form intermediate 19. CuII species 19 undergoes disproportionation with CuII intermediate 16 to generate CuIII intermediate 20 and CuI hydroxide24 (transition state not located). Following disproportionation, the selectivity-determining arylation occurs. On the other hand, intermediate 20, which has the sulfur atom of the sulfenamide bound to the copper center, can undergo reductive elimination via [20-21] (overall span of 5.0 kcal/mol from 19) to generate S-aryl product 21. Alternatively, the sulfenamide can coordinate via the nitrogen atom as in CuIII intermediate 20’. Intermediate 20’ will then undergo facile reductive elimination via [20’-21’] (span of 0.4 kcal/mol) to yield the experimentally observed and thermodynamically favored N-aryl product 21’ (downhill in energy by 61.5 kcal/mol) and regenerate CuI. The CuI species is oxidized by oxygen to CuII to restart the catalytic cycle24, as shown in Fig. 5b.

Next, we explored the nature of the ligand-controlled chemoselectivity by comparing the experimental results with L1, L2, and L3 to our computational results. The potential energy surface for the formation of S-aryl and N-aryl products from intermediate 18 with these ligands is given in Supplementary Fig. 8. While the spans for N-arylation for L1, L2, and L3 are fairly similar (see Supplementary Fig. 9-10), the S-arylation for L3 has the highest span followed by L2 and L1. Notably, ligand L3, which experimentally gives the best chemoselectivity for the N-arylation product, has the largest difference between energy spans for S-arylation and N-arylation, supporting our mechanistic proposal.

To understand the origin of ligand-controlled chemoselectivity, interaction/distortion analysis was performed on the S-arylation and N-arylation transition states as described by Houk and Bickelhaupt44 (see Supplementary Fig. 11). To estimate the electronic energy of the transition states, a comparison analysis of the favorable interaction energy between CuL and aryl substrate with the disfavorable distortion energy from the intermediate into the transition state geometries was performed. For S-arylation, the interaction energy appears to control the energy of the transition states. Specifically, S-arylation with L1 has the lowest overall energy and the most favorable interaction energy, followed by L2 and L3. This favorable interaction energy for S-arylation with L1 leads to the observed lower ratio of N:S product experimentally. Plots of the noncovalent interactions (NCI) show the slightly larger interaction between the aryl of the sulfenamide and the ligand in L1 compared with L2 or L3 (Supplementary Fig. 12). On the other hand, the interaction energies between the LCu fragment and the sulfenamide in the N-arylation are fairly similar for all three ligands. Instead, it is the distortion energy which appears to control the relative energies of the transition states, with L3 having the largest distortion energy, corresponding with its highest energy. This distortion can be observed in the bond lengths of the forming C-S bond (Supplementary Fig. 13), where L3 has the shortest C-S bond. It follows that this transition state is the latest of the three, which leads to greater distortion from the initial ground state.

From the combined experimental results and computational studies, a plausible catalytic cycle was proposed (Fig. 5b). Oxidation of CuI species A leads to the formation of CuII complex B under exposure to dioxygen gas, followed by transmetalation with arylboronic acid 2 to yield arylated CuII complex C. Then, sulfenamide 1 binds to copper center via the nitrogen atom, as supported by the EPR study. Next, deprotonation occurs, followed by disproportionation to furnish a CuIII species (F) along with a CuI byproduct. Ultimately, F undergoes reductive elimination to afford the desired sulfenamide 3, and A was formed to close the catalytic cycle.

In this work, we have developed a copper-catalyzed Chan–Lam coupling of sulfenamides with arylboronic acids to afford diverse N-aryl sulfenamides in a direct and step-economical fashion. Sulfenamide-type derivatives of two marketed therapeutics have been directly functionalized using our protocol, highlighting its synthetic potential in medicinal chemistry. The chemoselectivity favoring C–N bond formation over C-S bond formation is steered by a tridentate pybox ligand, which dictates mono-coordination of the sulfenamide to the copper center via nitrogen, and facilitates the subsequent reductive elimination of C–N bond. The key copper species in the catalytic cycle are spectroscopically investigated, and mono-coordination of sulfenamide to copper via the nitrogen atom is characterized by EPR and verified by the EPR isotope response using 15N-labeled sulfenamide. DFT studies show that the reaction proceeds via transmetalation of the arylboronic acid followed by coordination of the sulfenamide substrate, deprotonation of the nitrogen atom, disproportionation, and reductive elimination. With this ligand, N-arylation is kinetically and thermodynamically favorable. The exact nature of the ligand determines the degree of S-arylation, with the electron-rich L3 having the highest N:S product ratio. Our work provides strong evidence that Chan–Lam coupling, besides its mild conditions and broad functional group tolerance, could be manipulated by ligands as a precise synthetic tool, which is currently under investigation in our laboratories.

Methods

General Procedure for Catalysis

To an oven-dried microwave vial equipped with a stir bar was added Cu(TFA)2•H2O (4.5 mg, 10 mol %), L3 (7.4 mg, 20 mol %) S-(4-fluorophenyl)-N-(p-tolyl)thiohydroxylamine (1a) (35.0 mg, 0.15 mmol) and p-tolylboronic acid (2a) (40.8 mg, 0.3 mmol). Cy2NMe (48.2 μL, 0.23 mmol) and MeCN (0.5 mL) were added via syringe under an air atmosphere. The vial was sealed with a septum, and refilled from a dioxygen balloon for 3 min. The solution was then stirred at room temperature for 24 h under an O2 atmosphere. Upon completion of the reaction, the solvent was removed under vacuum to give a residue, which was further purified by flash chromatography to give the pure product.