N-Heterocyclic carbene-catalyzed enantioselective hetero-[10 + 2] annulation

Higher-order cycloadditions are a powerful strategy for the construction of polycycles in one step. However, an efficient and concise version for the induction of asymmetry is lacking. N-heterocyclic carbenes are widely used organocatalysts for asymmetric synthesis and could be an ideal choice for enantioselective higher-order cycloadditions. Here, we report an enantioselective [10 + 2] annulation between catalytically formed aza-benzofulvene intermediates and trifluoromethyl ketone derivatives. This protocol exhibits a wide scope, high yields, and good ee values, reflecting a robust and efficient higher-order cycloaddition. Density functional theory calculations provide an accurate prediction of the reaction enantioselectivity, and in-depth insight to the origins of stereocontrol.

In brief, NHC-catalyzed cycloadditions ranged from [2 + 2] to [8 + 2] have been extensively investigated over the past few years, but there is a remarkable lack of higher-order cycloadditions (e.g., [10 + 2] 62,63 and [14 + 2]). Although the intricately competitive pathways make the reaction-control difficult, these higher-order cycloadditions can provide a direct way to efficiently build polycyclic scaffolds.
Herein, we report a hetero-[10 + 2] higher-order cycloaddition of indole-2-carbaldehydes with trifluoromethyl ketone derivatives, proceeding via an NHC-bounded aza-benzofulvene intermediate (Fig. 1c). This discovery represents the initial use of NHC-bounded aza-arylfulvene intermediates in catalytic and enantioselective [10 + 2] or [14 + 2] reaction. In addition, in medicinal chemistry, the incorporation of "F"-containing fragments normally provides an effective route to enhance the metabolic stability, as well as other chemical or physical properties, of target molecules [64][65][66] . Based on the importance of polycyclic structures and incorporated "F"-containing fragments, the potential of these synthesized molecules in drug discovery is worth our expectation.

Results
Reaction optimization. We commenced our studies by investigating the reaction of indole-2-carbaldehyde 1a and 2,2,2-trifluoroacetophenone 2a as the model substrates, K 2 CO 3 as the base, DQ as the oxidant, tetrahydrofuran as the solvent, and the results are briefly summarized in Table 1. When L-phenylalaninederived triazolium NHC precatalyst A was exploited, the expected cycloadduct 3a was not observed. Replacing the mesitylene group with pentafluorophenyl group triazolium NHC precatalyst B gave desired product 3a in 40% yield and 0% ee, whereas the use of precatalyst C and D resulted in almost no reaction. To our delight, when indanol-derived triazolium catalyst E was tested, the [10 + 2] cycloadduct 3a was successfully formed in 61% yield with 35% ee and implies that this highly enantioselective [10 + 2] annulation can be achieved in the presence of ideal conditions. The catalytic performance could be further improved by changing the X group of precatalyst E from H to NO 2 (entry 6). After evaluating bases and solvents, we found that a combination of PhCO 2 Na as the base and hexane as the solvent gave the product 3a in 80% yield and 88% ee (entry 10). Improvements in yield and enantioselectivity were found when thiourea was used as the additive to form 3a (entry 12, 85% yield, 91% ee).
Substrate scope. With the optimal catalytic system in hand, we moved our attention to exploring the generality of this asymmetric higher order [10 + 2] annulation. As illustrated in Fig. 2,  by reacting with indole-2-carbaldehyde 1a, an array of aryl trifluoromethyl ketones 2 was examined first. In the reactions to generate the [10 + 2] cycloadducts 3, yields and enantioselectivities were found to be independent of the electronic properties of the substituents on the aryl group in 2 (3b−i). When the heteroaryl trifluoromethyl ketones were reacted with indole-2carbaldehyde 1a under optimal conditions, an [10 + 2] annulation was efficiently realized in all cases (3j−n). Reactions attempted using the alkyl trifluoromethyl ketones gave their corresponding [10 + 2] cycloadducts in good yields with high ee values (3o and 3p). Whereas the alkenyl trifluoromethyl ketone 2q was reacted with 1a, product 3q was also obtained in a good yield (73%) but with a slightly diminished enantioselectivity (72% ee). Switching the fluorinated substituent from CF 3 to CF 2 H, ClCF 2 , or C 2 F 5 in ketones, synthetic useful yields, and high to excellent enantioselectivities were still obtained under current conditions (3r−t). Next, we turned our focus to investigate the scope of substrate 1 (Fig. 3). Different substituents and substitution patterns on the indole skeleton were examined comprehensively. Electron-     Electron-releasing groups such as methyl (4c, 4e, 4f, and 4g) and methoxyl unit (4d) could also be installed on the indole scaffold of the aldehyde substrates. It is worth to note that this [10 + 2] protocol could be extended to a higher-order [14 + 2] cycloaddition, affording their corresponding cycloadducts (4h and 4i) in good enentioselectivities albeit with acceptable but dropped yields under the current standard conditions. The absolute configuration of 3e (CCDC 1961662) was determined by singlecrystal X-ray analysis and other products were assigned by analogy.
Postulated mechanism. A postulated catalytic mechanism of [10 + 2] annulation is summarized in Fig. 4. Deprotonation of NHC precatalyst F gives the corresponding NHC, which adds to aldehyde 1 to give the corresponding tetrahedral intermediate, 67,68  were conducted to gain a better insight into the mechanistic details. The initial rate constants of the reaction were determined in situ 1 H-nuclear magnetic resonance (NMR) and 19 F-NMR spectroscopy. The results show that the reaction appeared to have a nearly first-order dependence on NHC catalyst F (Fig. 5a), and zero-order dependence on substrates 1a (Figs. 5b), 2a (Fig. 5c), and DQ (Fig. 5d).
To further reveal the enantioselectivity of this [10 + 2] annulation, density functional theory (DFT) calculation was performed to study the key step of nucleophilic attack of intermediate II onto trifluoroacetophenone. As shown in Fig. 6a, two transition states named TS(II-III)R and TS(II-III)S was located, where the re-or si-face of trifluoroacetophenone was attacked, respectively. The calculated relative free energy of transition state TS(III-IV)R is 5.0 kcal/mol lower than that of TS (II-III)S, which predicts that the generation of R-configuration product 4a is favorable. The calculated results overestimate the level of enantioinduction in this reaction process but are consistent with predicting the observed experimental product configuration. The geometry of those two transition states is also given in Fig. 6b. After the absorption of indole reactant onto the NHC catalyst, a strong π-π stacking between indolyl moiety and the aryl in the NHC catalyst can significantly stabilize the deprotonated indolyl moiety. The π-π attraction is clearly shown in calculated noncovalent interaction (NCI) maps. In addition, kinetic experiments revealed that electron-rich indoles or electron-deficient aryl ketones reacted more quickly, which partially elucidated the potential π−π interaction. When the nucleophilic attack occurs, trifluoromethyl of trifluoroacetophenone appears at the more bulky inner side in transition state TS (II-III)R. It is more favorable than the case in transition state TS (II-III)S that the phenyl group is set to the inner side. The NCI map of transition state TS(II-III)R clearly reveals that the repulsion between phenyl group of trifluoroacetophenone and the NHC catalyst leads to instability of transition state TS(III-IV)S, while this repulsion is absent in transition state TS(II-III)R.
In order to figure out whether the process from II to III would be concerted or stepwise, the intrinsic reaction coordinate calculation (IRC) of transition state TS(II)R has been performed (Fig. 6c). The result clearly shows the C1 of trifluoroacetophenone and N1 of indole would form the covalent bond firstly. Along with the decreasing distance of C1-N1, the bond of oxygen atom O1-C2 gradually formed until the intermediate III generate. Hence, we speculate that the process tends to be a concerted asynchronous process 59,69 .
Synthetic transformations and applications. Our protocol is amenable to large-scale preparation. For example, the use of standard conditions was sufficient to produce 4d (1.29 g) in 92% yield and with 90% ee (Fig. 7a). A facile Pd-catalyzed Suzuki coupling of 3d with 4-methoxyphenylboronic acid 5 led to product 6 in a 72% yield and with a remained enantioselectivity (Fig. 7b).
In summary, a unique NHC-catalyzed enantioselective hetero-[10 + 2] annulation of indole-2-carbaldehydes with trifluoromethyl ketone derivatives has been developed. This process generates a new NHC-bounded aza-benzofulvene as a key intermediate. This new protocol allows the rapid assembly of enantioenriched polycycles from readily available starting materials under mild conditions. DFT calculations elucidated the origins of the [10 + 2] process. Further investigations on new NHCbounded aza-arylfulvene as an active intermediate in asymmetric synthesis are currently ongoing in our laboratory.

Methods
Synthesis of 3/4. To a flame-dried Schlenk reaction tube equipped with a magnetic stir bar, was added the precatalyst F (15.4 mg, 0.03 mmol), DQ (90.0 mg, 0.22 mmol), additive H (5.0 mg, 0.01 mmol), PhCO 2 Na (28.8 mg, 0.20 mmol), 1 (0.20 mmol) and 4 Å MS (60 mg). The Schlenk tube was closed with a septum, evacuated, and refilled with an argon atmosphere. Hexane (2.0 mL) and 2 (0.24 mmol) was added. The mixture was then stirred at 25°C and monitored by TLC until 1 was consumed. The mixture was concentrated under reduced pressure and purified by column chromatography on silica gel (hexane/EtOAc = 100:1) to afford the desired product 3 or 4. Full experimental details can be found in the Supplementary Methods.

Data availability
For 1 H NMR, 13