Indanone-derived b-ketoamide 1a and phenylacetylene 2a were selected as the model substrates to conduct our research. Firstly, several cooperative catalytic systems, which showed good ability in catalytic enantioselective Conia-ene reaction, including Pd(II)/Yb(III) dual catalyst system, Zn(II)/Yb(III) catalyst system and amine–silver, were investigated13,16,19. But all of them gave only trace amount of product without enantioselectivities even rising the reaction temperature to 70 °C (Table 1, entries 1–3). Then chiral N,N'-dioxide ligand-metal complexes were chosen as the activators of ketoamides, in connection with AuCl∙PPh3/AgOTf for the activation of 1-alkyne. Firstly, Sc(OTf)3 was used to coordinate with chiral N,N'-dioxide L-PiEt2 to promote the reaction under air atomosphere, the byproduct 3bb was obtained as the main product along with the desired product 3aa in 11% yield with 60:40 e.r. (entry 4). Further research showed that the reaction could possess efficiency in an absolute anaerobic condition, delivering the product 3aa in 92% yield with 60:40 e.r. (entry 5). Then Ga(OTf)3 which showed efficient catalytic activity in Shi’s report41 was used to coordinate with chiral N,N'-dioxide L-PiEt2 to promote the reaction, however, only trace of product 3aa was obtained (entry 6). To our delight, In(OTf)3 could improve the reaction activity greatly and deliver the desired product with 62:38 e.r. (entry 7). To improve the enantioselectivity, other conditions were carefully studied. Changing the N,N'-dioxide ligand to L-PiEt2Me, which has ethyl groups at ortho-positions and methyl group at para-position of aniline, the yield could be improved to 99% with 63:37 e.r. (entry 8). Moreover, the addition of trace amount of H2O (entry 9) and increasing the amount of ligand L-PiEt2Me (entry 10) benefited the improvement of the enantioselectivity. Further exploration showed that the solvent had a great influence on the reaction, when para-xylene was used as solvent, the desired product was isolated in 98% yield with 90:10 e.r. (entry 11). The enantioselectivity enhanced into 94.5:5.5 e.r. after the concentration of 1a reduced to 0.067 mol/L by enhancing the amount of solvent (entry 12). The steric hindrance of the ligands on [Au] catalyst was another key factor. Changing the AuCl·PPh3 into more sterically hindered XPhosAu(TA)OTf, only trace product could be obtained (entry 13). In comparison, other indium catalysts of the typical chiral ligands such as Pybox L3, Box L2, or CPA organocatalyst were used, the product 3aa was obtained in extremely low yield with poor enantioselectivity (entries 14–16).
With the optimized reaction conditions in hand (Table 1, entry 12), the substrate scope was then evaluated (Fig. 2). A variety of ketoamides 1 derived from 1-indanones with different substituents were tested. Substrates with electron-donating groups exhibited excellent yields and enantioselectivities (3ba–3ea) at 50 °C. Substrate 1f bearing an electron-withdrawing group transformed to the desired product 3fa in 98% yield with 85:15 e.r. at higher temperature (60 °C). With respect to 1-alkynes 2, when the substituents at the aromatic ring of the phenylacetylenes varied, both steric hindrance and electronic properties had little effect on the reaction (3ab–3ai). However, substrate 1,4-diethynylbenzene 2j just delivered the product 3aj in moderate yield with excellent enantioselectivity. It might be caused by the competitive coordination of the alkyne-bearing product with AuOTf∙PPh3. The thienyl-substituted alkynes (2k and 2l) were also suitable. Various aliphatic 1-alkynes (2m–2q) could also transformed to the desired products in moderate to brilliant yields with good enantioselectivities (3am–3aq). Importantly, the methodology was applicable to the alkyl-alkyne derived from saccharide 2r. Next, ring structure of ketoamides was studied. The substrate 1h derived from 1-tetralone got good results (3ha-3hl), while 1i derived from 1-benzosuberone gave much lower yield and e.r.. It might be caused by steric hindrance between methylene of substrate 1i with AuOTf∙PPh3-activated 2a. Meanwhile, aliphatic substrate 1j was also tolerated, affording the product 3ja in moderate yield with good enantioselectivity. The absolute configuration of 3ae was determined to be R by X-ray crystallographic analysis and the absolute configurations of 3aa–3ac and 3ag–3ah were determined to be R by comparison of the CD spectra with that of 3ae.
For acyclic b-ketoamide 3a, which without other substituent on α-position transformed to thermodynamically stable achiral α,β-conjugated carbonyl product 4aa through olefin isomerization (Fig. 3). When acyclic b-ketoamides 3b–3e which bearing methyl, phenyl, benzyl or chlorine group on the α-position were used as the nucleophiles, the corresponding products could not be observed due to steric hindrance of substituents. Therefore, α-fluoro substituted 3f with smaller steric hindrance and stronger acidity of α-proton was evaluted (Fig 4). Moderate yields with good e.r. could be obtained after adjusting the ligand to L-PiEt2, increasing the reaction temperature and prolonging the reaction time. Electron-donating or electron-withdrawing substitutes on the para-position of phenyl ring were tolerated well. Generally, the 1-alkynes 2 with an electron-donating substituent led to better yields than the ones with electron-withdrawing substituents. Compared with the phenylacetylene, the more electron-rich aromatic alkynes like 2l and 2s showed better reactivities (4fl and 4fs). When aliphatic 1-alkynes 2m and 2n were applied to the reaction, the products were delivered in moderate yields with good e.r. values.
To evaluate the synthetic potential of the catalytic system, a gram-scale synthesis of the product 3aa was carried out (Fig. 5). Under the optimized conditions, 3.5 mmol of 1a and 7.0 mmol of 2a reacted smoothly, delivering 1.14 g (98% yield) of 3aa without any erosion of the enantioselectivity. The reduction of carbonyl group of 3aa using LiAlH4 provided secondary alcohol product 5aa in 90% yield with 92:8 d.r. and 94.5:5.5 e.r.. The absolute configuration of the major isomer was confirmed to be (1S, 2R) by X-ray crystal analysis, and the stereo-arrangement at the quaternary carbon center is in consist with that of 3ae. Besides, the epoxidation of 3aa in the presence of m-CPBA afforded the epoxide derivative 6aa in 98% yield with 90:10 d.r. and 94.5:5.5 e.r. (Fig. 5).
Next, the reaction mechanism was investigated (Fig. 6). Some control experiments were carried out (Fig. 6a). In the absence of AuCl∙PPh3/AgOTf or In(OTf)3/L-PiEt2Me, only trace amount of the product 3aa was detected, which indicates that the two catalysts work cooperatively. N,Nʹ-dioxide/In(OTf)3 crystal structure obtained in our previous study48 showed that a OH-bridged dinuclear indium complex forms in the presence of H2O, in which N,Nʹ-dioxide coordinates to In(III) in a tetradentate manner. Nevertheless, the investigation of relationship between the e.e. value of L-PiEt2Me and that of 3aa showed a clear linear effect (Fig. 6b), implying that the active catalytic species is likely to be the mixture of In(OTf)3 and L-PiEt2Me in a 1:1 ratio. The OH anion generated from the water in situ preparation of the chiral indium catalyst might act as a base to accelerate the enolization of 1,3-dicarbonyl compounds. In addition, the M+ peak (Found: 561.1058), which corresponded to a 1:1 complex C of [Au·PPh3]+ and phenylacetylene 2a, was detected by ESI-TOF analysis in the positive-ion mode. The mixture of L-PiEt2Me, In(OTf)3, and 1a (1:1:1) in p-xylene displaying an ion at m/z 1114.4025 ([L-PiEt2Me+In3++OTf–+1a-H+] m/z calcd 1114.4036) suggested that enolized 1a coordinate to the catalyst in a 1:1 molecular ratio (Fig. 6c), which is consistent with our non-linear effect.
Based on above analysis and previous work, a catalytic cycle with a possible transition state is proposed. As illustrated in Fig. 7, in In(OTf)3/L-PiEt2Me cycle, initially, the tetradentate L-PiEt2Me coordinates to InIII to form a six-coordinate octahedral geometry complex A' and dimer A. When ketoamide 1a was added, the basic anion of the catalytic species accelerates the deprotonation process, and the enol ion of 1a coordinates tightly to chiral indium(III) center through two oxygens to form the carbanion nucleophile intermediate B. On the other hand, as for [Au] cycle, the [Au]OTf which is the more reactive species would bind to the π-bond of 1-alkyne 2 in an unsymmetrical fashion to form species C. The intermediate C then reacts with the complex B to form the Au/In stabilized reactive intermediate TS which is the origin of the stereoselectivity. Due to the steric hindrance of [Au] unit with the amide moiety of the ligand L-PiEt2Me, cat1 activated π-bond of 2a approaches preferably from the Re-face to undergo an energetically favorable C–C bond forming reaction, forming the complex D with R absolute configuration at the newly formed stereogenic center. Subsequent protonation of D gives the desired product 3 and releases the two catalysts.
In summary, we have successfully realized an efficient catalytic asymmetric Nakamura reaction of cyclic and acyclic 1,3-dicarbonyl compounds with unactivated 1-alkynes by developing a bimetallic synergistic catalysis. The combination of π-acid gold(I)/chiral N,Nʹ-doxide-indium(III) complex enabled the activation of alkyne and the efficiency and stereoselectivity of nucleophile. Various β-ketoamide derivatives with a cyclic all-carbon quaternary center and acyclic quaternary center with a fluorine substituent were obtained in moderate to excellent yields with brilliant enantioselective ratios under mild conditions. A possible catalytic cycle with a transition-state model was proposed to elucidate the process of the reaction and origin of chiral induction. Further studies on hetero bimetallic synergistic or relay catalysis are underway in our laboratory.