3-Oxindole is a prominent structural motif found in numerous natural products (Fig. 1a)1-5 and exhibits extensive bioactivities6-9 and fluorescence property10. Recently, continuous endeavors have been devoted to synthesize the chiral 3-oxindoles through multistep reactions, involving oxidation and construction of chiral quaternary carbon starting from 2-substituted indoles (Fig. 1b)11-25. However, catalytic asymmetric de novo building chiral 3-oxindole has been far behind.
Direct C-H functionalization greatly reduces tedious manipulations and unwanted chemical wastes. Furthermore, enantioselective directed ortho-C-H functionalization has emerged as a powerful tool to construct chiral carbon center26. In the past decade, the enantioselective ortho C-H additions to alkene27, 28, imine29, 30 and aldehyde31, 32 have been well developed to access chiral alkane, amine and secondary alcohol in the presence of transition metal or chiral phosphoric acid (CPA) as the catalyst (Fig. 2a). However, the enantioselective ortho-C-H addition to ketones was mentioned scarcely due to its low reactivity and difficulties in enantiocontrol. In addition, β-carbon elimination33-35 of the unstrained tertiary alcohols is usually inclined to be occurred in the presence of transition metal as the catalyst. For ortho-Friedel-Crafts addition with the CPA catalyst, the para-selectivity is usually preferred36-40. Previous reports are only limited to 2-naphthylamine41, 42, para position hindered arylamine43 and 2-naphthols44 as substrates. Recently, the groups of Shibata45 and Yamamoto46, 47 developed the iridium(I) catalyzed intramolecular enantioselective ortho C-H addition to ketone independently (Fig. 2b). The carbonyl insertion was the rate-limiting step, rather than the C-H bond cleavage. The nucleophilicity of the aryl-iridium intermediate is an important factor in this procedure47. There is higher energy barrier for the intermolecular ortho-C-H addition to ketone carbonyl group. To the best of our knowledge, the catalytic intermolecular enantioselective ortho-C-H addition to ketone of para-C-H non-hindered aryl compounds has not been documented to date. We delude that an intermolecularly enantioselective addition of ortho-C-H of primary aryl amine to α-diketone could afford a chiral 3-hydroxyindolenine and a chiral 3-oxindole through followed by α-iminol rearrangement (Fig. 2c)48-53. The challenge to realize de novo construction of chiral 3-oxindoles mainly lies on: 1) the low reactivity of intermolecular ortho-C-H addition to construct tertiary alcohol; 2) the ortho-C-H selectivity with para- non-hindered primary aryl amine; 3) the enantioselective control of intermolecular ortho-C-H addition; 4) easily formed ketimine via dehydration condensation and its undesirable C-H addition side reaction. Herein, we disclose a CPA catalyzed intermolecularly enantioselective ortho-Friedel-Crafts addition of primary arylamines to 2,3-diketoesters, providing the chiral tertiary alcohol. The CPA with bifunctional basic center and a hydrogen bond donor group play key roles to activate both nucleophiles and electrophiles, reducing the energy barrier of ortho-Friedel-Crafts addition to ketone54. Meanwhile, the hydrogen bonding interaction of amine with O=P from CPA worked as noncovalent directing group to realize ortho-regioselectivity instead of the para-position attack. The following cyclization generated chiral 3-hydroxyindolenines. The final enantioselective [1,2]-ester migration fulfills the de novo construction of chiral 3‑oxindoles.
Results: Investigation of reaction conditions. To validate the feasibility of our hypothesis, the reaction of 1-naphthylamine 1a and 2,3-diketoester 2a was conducted with a family of chiral BINOL CPAs as the catalyst (Table 1, entries 1-4). The obtained results indicated that the bulky BINOL-(R)-CPAs and condensed aromatics substituted BINOL-(R)-CPAs produced opposite enantiomer 4aa, which indicated that different coordination patterns worked in the bulky and condensed aromatics substituted CPAs catalytic system. Firstly, the sterically bulky CPAs were surveyed. The bulkier BINOL-(R)-CPAs gave the product 4aa with better enantioselectivity. Especially, (R)-5c with 2,4,6-trisisopropylphenyl group at the para-position of the aryl ring in BINOL-(R)-CPA produced (S)-4aa with 81% ee (entry 7). The evaluation of different solvents showed that the chlorinated hydrocarbon gave better enantioselectivity with the bulky CPAs. The combination of CH2Cl2 and ODCB (1,2-dichlorobenzene) (4:1) improved the ee to 88% (entry 8). Increasing the amount of solvent (CH2Cl2:ODCB = 4:1) to 4 mL, 91% ee of (S)-4aa was obtained (entry 9). Thereafter, the further evaluation of spirocyclic-based CPAs (entries 10-12) showed that 9-anthryl substituent spirocyclic CPA (R)-6b gave good enantioselectivity (68% ee). Solvents were surveyed with (S)-6b as catalyst (entries 13-14) and AcOiPr gave the best enantioselectivity (88% ee). Increasing the amount of AcOiPr to 4 mL, 92% ee of (S)-4aa was obtained (entry 15).
Scope. With the optimized reaction conditions in hand, the reaction scope was next examined (Table 2). The 2,3-diketoesters 2 with alkyl (R) substituents, including methyl, n-butyl, benzyl and n-heptyl, reacted smoothly with 1a to generate 4ab-4ae in high yields and enantioselectivity. The phenyl (Ar) substituents of 2, with para-substituted halogen (2f-2h), electron-withdrawing groups (2i-2k) and methyl (2l), gave the desired products with excellent yields (88%-99%) and enantioselectivity (94-96%). Electron-donating substituted 2 (-OMe, 2m) afforded 4am with 78% ee and 97% yield using 0.5 mL of solvent. The substrates 2 with meta-substituted group could give product with excellent yields and ee (4an, 90% yield, 94% ee). The ortho-substituents gave product with good yield and low ee (4ao, 80% yield, 20% ee). Gratifyingly, 2 with 2-furyl or 2-thienyl groups produced 4ap and 4aq with excellent ee (96% and 97%). Then, the tolerance of substituents on the 1-naphthylamine were evaluated. Cyclopropyl (1b), phenyl (1c), 1-naphthyl (1d), 3-thienyl (1e), bromo (1f, 1g), chloride (1h) and methoxy (1i) substituents at the different position were tolerated well and the corresponding products were provided smoothly with good to excellent yields (61-99%) and excellent enantioselectivity (94-99% ee) using 2q as reaction partner. In addition, 1-aminoanthracene (1j) worked well, providing 4jq in excellent yield (93%) and enantioselectivity (97%). Interestingly, the reaction of 3-fluoranthenamine (1k) and 2q produced opposite enantiomer product 4kq in 99% yield and 96% ee. Surprisingly, 1-dibenzofuranamine (1l) also produced the desired product 4lq with excellent ee of 96% and moderate yield. Moreover, the reaction of 4-aminoindole (1m) and 2q gave the product 4mq with 81% yield and 97% ee. 7-Aminoindole (1n) only provided 22% yield of 4nq with 62% ee.
Mechanistic investigation. To validate if the 3-hydroxyindolenine 3 is a key intermediate and the enantioselective formation step of this reaction, the control experiment of 1a and 2q with (S)-6b as catalyst was conduct (Fig. 3). 3aq was obtained in 99% yield and 98% ee. The result implied that this methodology could also be used for catalytic enantioselective synthesis of 3-hydroxyindolenine. Besides, the ortho-Friedel-Crafts addition to carbonyl group might be the enantioselectivity-determining step. With the diphenyl phosphate as catalyst, only 47% yield of rac-3aq was obtained along with ketimine 5 as the by-product.
To finally determine the enantioselective ortho addition as turnover-limiting step of this reaction, a Hammett analysis was performed for a series of 2,3-diketoesters (Fig. 4). A positive value (0.77) was observed, implying that the reaction rate of the 2,3-diketoester with electron-withdrawing substituents at the para position to the central carbonyl is faster than that of substrates with an electron-donating groups, indicating that the ortho-Friedel-Crafts addition to carbonyl group might be the rate-limiting step.
In view of the catalysts of (R)-5c and (S)-6b with the same orientational intersection angel resulted opposite enantiomer products, the two different possible chiral control models were proposed (Fig. 5). The repulsive steric interaction mode with (R)-5c as catalyst via Si-face attack result in (S)-3aa and (S)-4aa. A stereo-control mode with (S)-6b as catalyst involving π-π stacking resulted in (R)-3aa and (R)-4aa.
Based on the above results, a plausible mechanism was proposed (Fig. 6). By virtue of its multiple coordinating sites and the highly reactive electrophilic character of the central carbonyl group55, 56, 2,3-diketoesters 257-62 and CPA form a chiral pocket via hydrogen bonding63. Meanwhile, the other hydrogen bonding is formed between the amine 1 and P=O moiety in CPA. The dual activations of both the nucleophile and the electrophile substrates facilitate the enantioselective ortho-Friedel-Crafts addition to the central ketone carbonyl and the chiral tertiary alcohol is formed. The following dehydration/heteroannulation produce 3. The finally [1,2]-ester migration gives the product 4.
Synthetic application. A successful gram-scale synthesis of 4aq demonstrated the practicality of this reaction (Fig. 7). Further transformations of enantioenriched 3-oxindoles 4aq were performed to illustrate the synthetic potential of this reaction. Reduction of 4aq with BH3·DMS resulted in a chiral amino alcohol 6 with 90% ee. The following cyclization with 1,1’-carbonyldiimidazole (CDI) under DMAP catalysis afforded the polycyclic compound 7 in 98% yield and 92% ee. All these examples showed the advantage of the current methodology as it is difficultly accessible by other means.