Optimization study. Initially, the coupling of phenethylpyridinium salt 1a and phenylacetylene 2a was selected as the model reaction for optimization (Table 1). To realize such transformation, we envisaged that a pincer ligand might be feasible due to its strong and tridentate bonding mode to the metal center, thereby possibly stabilizing the alkylnickel intermediate50 and suppressing the undesired Glaser coupling51,52. Thus, the ligands were firstly screened by using 10 mol% NiCl2(glyme) as catalyst, K3PO4 as base in tetrahydrofuran (THF) at 80 oC. When pyridine bisoxazoline (pybox), the most efficient ligand in Liu’s work18, was applied to this reaction, the desired product 3a was obtained in 4% yield and the main product was 1,4-diphenylbutadiyne derived from the homocoupling of 2a (entry 1). While the use of a more electron-rich and bulky 4,4′,4″-tritert-butyl terpyridine (ttbtpy) in this process, the yield of 3a was improved to 53% (entry 2). Much to our delight, the yields of 3a could be further improved to 87% and 83% respectively, when amide-type pincer ligand (e.g. N-(pyridin-2-ylmethyl)picolinamide (L1) or N-(quinolin-8-yl)picolinamide (L2)) was used (entries 3–4), though they were seldom used as ligands in transition metal-catalyzed cross-coupling reactions53–56. This discovery encouraged us to synthesize two new sterically more hindered methylated derivatives L3 and L4 as ligands. Gratifyingly, the yield was significantly improved to 96% by employing L4 (entry 6). The reasons for the high efficiency of L4 are still unclear at present, but probably related to its steric hindrance and rigidity. Screening of nickel catalysts revealed that Ni(acac)2 was ineffective (entry 7), whereas the inexpensive, air- and moisture-stable NiCl2·6H2O gave the best result (entry 8). Subsequently, the effect of base was examined. K2CO3 resulted in a slightly diminished yield (entry 9). However, the reaction completely shut down by using Et3N, a frequently-used base in palladium-catalyzed Sonogashira coupling of aryl halides (entry 10)11. Lowering the amount of catalyst or reaction temperature led to a reduced yield in different extent (entries 11–12). Control experiments indicated that NiCl2·6H2O, L4 and K3PO4 were all essential for achieving the transformation (entries 13–15) (For a detailed optimization study, see Supplementary Information).
Substrate scope. With the optimized coupling conditions in hand, the scope of alkynes was first evaluated using 1a as the coupling partner. For some cases that the products were unseparated from the excess terminal alkynes, p-methoxylphenethylpyridinium salt 1b was used instead of 1a. As shown in Fig. 2, the arynes bearing both electron-donating and electron-withdrawing groups could participate in this transformation delivering the products (3b-3 k) in excellent yields. Various synthetically important functional groups including methoxyl, arylhalide, ester, acetyl, trifluoromethyl, formyl and free amino were all perfectly accommodated. Particularly noteworthy was that aryl chlorides and bromides, popular electrophilic partners in Sonogashira reactions10, remained inert under our optimized reaction conditions, highlighting the exquisite chemoselectivity of this transformation. Additionally, the presence of an ortho formyl did not hamper the reaction. Strikingly, terminal alkyne (2 l) containing a boronate ester group was also successfully engaged in this transformation with its C-B bond intact, thus allowing for further diversification. Heteroaromatic rings such as pyridine and thiophene that might deactivate a metal catalyst by coordination, and 1-ethynylcyclohexene could also smoothly undergo the transformation giving the corresponding products (3 m-3o) in excellent yields. More importantly, aliphatic alkynes (2p-2t) could also be coupled in high efficiency. The functional groups such as Cl, NHBoc, and OH were well tolerated, affording the products (3r-3t) in high to excellent yields with excellent selectivity. Finally, triisopropylsilyl- and trimethylsilyl-capped alkynes were also suitable substrates to obtain the products (3u-3v) in high yields.
Next, the generality of alkyl amines was evaluated as shown in Fig. 3. Various primary alkyl (1b-1 l) and benzylpyridinium salts (1 m-1o) were all suitable substrates for this transformation, and the desired products (4b-4o) could be obtained in high to excellent yields. However, the secondary alkylpyridinium salts (e.g. 1t) exhibited a dramatic drop in reaction efficiency (4t, 74%) under the optimized conditions. Then reoptimization of secondary alkylpyridinium salts was conducted by exploring various reaction parameters. Gratifyingly, 98% yield of 4t could be obtained by changing the solvent to DMF. Under the slightly modified conditions, diverse secondary alkylpyridinium salts underwent this coupling smoothly to give the desired products (4p-4w) in high to excellent yields. Similarly, good functional group tolerance was observed, as exemplified by the well compatible with methoxyl, trifluoromethoxyl, bromide, indole NH, alkenyl, tert-amine, acetal, hydroxyl and chloride. More importantly, heterocyclic units such as thiophene (1f), pyridine (1 g), indole (1 h), tetrahydropyran (1 s) and piperidine (1t) which are prevalent in medicinally relevant molecules were competent substrates. In addition, benzylpyridinium salts especially electron-rich benzylic salts which are not suitable in Gryko’s work46 could be coupled with high efficiency (4 m-4n), emphasizing the robustness of our strategy in synthetic applications. It is worth noting that both cyclic (1p-1u) and acyclic secondary amines (1v-1w) could be readily applied to this protocol with high to excellent yields.
It is worth highlighting that this protocol was amenable to a one-pot transformation in which pyrylium salt, alkyl amine and the cross-coupling reagents were added simultaneously in a single step, and 78% yield of the product 3a could be obtained without further reoptimizing the reaction conditions (Fig. 4a). Additionally, a gram-scale reaction was successfully performed using 1a and 2c under the optimized conditions producing 3c in 87% yield, exemplifying the practicability and scalability of this process (Fig. 4b).
Late-stage derivatizations. To further demonstrate the broad applicability of this method, late-stage functionalization of natural products and medicinally relevant molecules were conducted (Fig. 5). A series of pyridinium salts and alkynes derived from drugs and bioactive compounds underwent this transformation with good to excellent yields (5–20). This general protocol could be successfully applied for rapid construction of alkyne-labelled derivatives of biomolecules (5–9). The readily attached alkynyl group is expected to serve as a labeling tool to facilitate further chemical biology studies and as a handle for rapid entry to complex derivatives. Likewise, this versatile method can be also applied in the further functionalization of alkynyl-containing bioactive molecules or intermediates (10–14). Notably, the virtues of the current method were further illustrated by the successful coupling of two drug molecules for assembling their drug-like hybrids 15–20, highlighting the potential applications of this chemistry in the discovery of pharmaceutical candidates.
Mechanistic studies. To understand the reaction mechanism, a series of experiments were performed. When the radical trapping reagent TEMPO was added to the reaction mixture, only trace of 3a was obtained with the concurrent formation of TEMPO-adduct 21 in 16% yield (Fig. 6a). In addition, a radical-clock experiment was also conducted by employing cyclopropylmethyl pyridinium salt 1x. Instead of normal cross-coupling product 4x, a ring-opened product 22 was achieved in high yield (Fig. 6b). These results suggest that an alkyl radical may be involved in this transformation. To further elucidate the role of the nickel catalyst in this reaction, a Ni(II) complex Int-1 was synthesized by simple exposure of NiCl2·6H2O and L4 in tetrahydrofuran at room temperature and characterized by X-ray crystallography. Gratifyingly, a high yield of 3a was obtained when Int-1 was applied to this catalytic transformation (Fig. 6c). However, when Ni(cod)2 was used as the catalyst, a remarkable decrease in efficiency was observed and only a moderate yield of 3a was achieved (Fig. 6d). These results indicate that Ni(II) complex Int-1 is likely involved as a competent catalytic species in this chemistry rather than a Ni(0) species.
Although a detailed mechanism awaits further studies, a plausible mechanism is depicted in Fig. 6e based on the reactions of Katritzky salts27–47 and Ni/pincer-ligand catalyzed cross-coupling of other alkyl electrophiles54,57−59. Initially, coordination of L4 to the Ni center followed by ligand exchange to form a Ni(II) complex Int-1. It may undergo transmetalation with alkyne promoted by base to give intermediate A. Single-electron transfer (SET) from A to pyridinium 1 generates a Ni(III) complex B and an alkyl radical, which recombines with another molecule of A to give a Ni(III) species C. Reductive elimination from C furnishes the desired product and an unstable Ni(I) intermediate D, which quickly undergoes comproportionation with B to regenerate A and Int-1 for the next catalytic cycle.