Ligand-accelerated enantioselective methylene C(sp3)–H bond activation

See allHide authors and affiliations

Science  02 Sep 2016:
Vol. 353, Issue 6303, pp. 1023-1027
DOI: 10.1126/science.aaf4434


Effective differentiation of prochiral carbon–hydrogen (C–H) bonds on a single methylene carbon via asymmetric metal insertion remains a challenge. Here, we report the discovery of chiral acetyl-protected aminoethyl quinoline ligands that enable asymmetric palladium insertion into prochiral C–H bonds on a single methylene carbon center. We apply these palladium complexes to catalytic enantioselective functionalization of β-methylene C–H bonds in aliphatic amides. Using bidentate ligands to accelerate C–H activation of otherwise unreactive monodentate substrates is crucial for outcompeting the background reaction driven by substrate-directed cyclopalladation, thereby avoiding erosion of enantioselectivity. The potential of ligand acceleration in C–H activation is also demonstrated by enantioselective β-C–H arylation of simple carboxylic acids without installing directing groups.

Enantioselective functionalization of prochiral C–H bonds can potentially lead to a broad range of efficient routes to chiral compounds. Despite extensive efforts, the scope and efficiency of enantioselective C(sp3)–H activation reactions are far from adequate for broad applications in asymmetric synthesis (1, 2). Enantioselective carbene and nitrene insertions into C(sp3)–H bonds have been demonstrated in both diastereoselective and enantioselective fashion (37). However, asymmetric C(sp3)–H activation reactions via metal insertion are limited to the desymmetrization of two prochiral carbon centers (815) (Fig. 1A). For example, desymmetrizations of relatively reactive cyclopropyl and cyclobutyl C–H bonds have been achieved with Pd(II) catalysts and chiral monoprotected amino acid (MPAA) ligands (811). Desymmetrization of two carbon centers has also been achieved through a Pd(0)-catalyzed intramolecular C–H arylation, as demonstrated in a series of pioneering studies (1215). Thus far, development of an efficient chiral metal catalyst that can differentiate prochiral C–H bonds residing on a single methylene carbon center via metal insertion remains a challenge. In terms of synthetic disconnection, such a process is also distinct from the desymmetrization, as the newly created chiral center of amide substrates resides at the β-methylene carbon instead of the α-carbon center. Recently, a transient chiral directing group has also been shown to perform enantioselective C–H arylation of benzylic C–H bonds (16). However, the transient amino acid directing group does not promote alkyl methylene C–H activation. Furthermore, the transient directing group is also incompatible with substrates derived from carboxylic acids.

Fig. 1 Enantioselective methylene C–H activation reactions.

(A) Enantioselective C–H activation via desymmetrization of two carbon centers. (B) Two synthetic disconnections. (C) Differentiating prochiral C–H bond on a single methylene carbon center. DG, directing group; PG, protecting group; OTf, trifluoromethanesulfonate; Ar, aryl group; Ac, acetyl group; Et, ethyl group; Bu, butyl group; R, alkyl or aryl group.

The use of a bidentate 8-aminoquinoline directing group and a chiral phosphoric amide ligand has afforded moderate enantiomeric ratios (er), ranging from 74:26 to 91:9 with benzyl C–H bonds, though this method is much less successful with alkyl C–H bonds (63:37 er) (17). In general, such strongly coordinating directing groups promote ligandless C–H activation reactions, which could be detrimental for asymmetric catalysis, as these background reactions erode enantioselectivity. Bidentate coordination from substrates also prevents the exploitation of a wide range of potentially powerful chiral bidentate ligands in palladium catalysis due to a lack of vacant coordination sites. Practically, the requirement for bidentate coordination from substrates precludes the use of a variety of simple monodentate directing groups and native functional groups to direct C–H activation, an important goal of the field.

Despite the aforementioned challenges, enantioselective β-C–H functionalization has long been the focus of our research efforts due to the importance of constructing β-chiral centers in asymmetric synthesis. Current retrosynthetic disconnections for the asymmetric synthesis of β-functionalized chiral carboxylic acids or amides extensively use conjugate addition reactions of the corresponding olefins. Rh(I)-catalyzed asymmetric conjugate addition of α,β-unsaturated ketones with aryl boronic acids has afforded a useful method for the preparation of chiral β-arylated compounds (18, 19). However, when a given substrate or synthetic intermediate contains a saturated aliphatic acid chain without double bonds, direct enantioselective arylation of the methylene C–H bonds at the β position of amides through palladium insertion provides a solution (Fig. 1B). In our early efforts, we adopted a chiral auxiliary approach to gain insight into stereoselective palladium insertion into β-C(sp3)–H bonds (20). However, development of an enantioselective version of these diastereoselective β-C–H iodination and acetoxylation reactions has not been successful, owing to the lack of an appropriate ligand that can match the strongly coordinating oxazoline directing group (21). Employing a weakly coordinating amide directing group in combination with chiral MPAA ligands has led to desymmetrization of methyl, cyclopropyl, and cyclobutyl C–H bonds (Fig. 1A) at two different carbon centers (9, 10). Unfortunately, MPAA ligands have proven ineffective in promoting palladium insertion into acyclic methylene C–H bonds.

Here we report the discovery of chiral acetyl-protected aminoethyl quinoline (APAQ) ligands that enable Pd(II)-catalyzed enantioselective arylation of β-methylene C–H bonds of aliphatic amides, with enantiomeric ratios reaching up to 96:4 and yields as high as 89% (Fig. 1C). These APAQ ligands containing quinoline and acetyl-protected amino coordinating moieties form six-membered bis-chelating rings with palladium, which drastically accelerates methylene C–H activation, thereby controlling the stereoselectivity. In contrast, the acetyl-protected aminomethyl quinoline coordinating with Pd(II) via five-membered bis-chelation is completely inactive in this reaction. Conceptually, the combination of weakly coordinating monodentate substrates and ligand acceleration opens the possibility for using a variety of simple coordinating groups, including native functional groups, to direct enantioselective C–H activation, as demonstrated with free carboxylic acid substrates.

Guided by our overarching goal of developing ligand-accelerated enantioselective C–H activation of weakly coordinating substrates, we set out to use the electron-deficient amide substrate 1 and evaluate the effects of chiral ligands on the extensively studied C–H arylation reaction (2224). Following our previous finding that quinoline and pyridine ligands can accelerate C(sp3)–H activation (25, 26), we prepared a number of corresponding chiral ligands (including L4 and L5) and examined their activity under standard reaction conditions (table S1). Unfortunately, these monodentate chiral ligands do not exert substantial influence on the stereochemistry of the palladium insertion step. Considering the effectiveness of bidentate MPAA ligands in controlling the stereochemistry of Pd-catalyzed desymmetrization of prochiral cyclopropyl and cyclobutyl C–H bonds on two carbon centers, we began to develop bidentate ligands incorporating structural motifs from both quinoline and MPAA ligands. The crucial role of the NHAc (Ac, acetyl) moiety of MPAA ligands in the C–H cleavage step, identified by experimental (8, 27) and computational studies (28, 29), prompted us to develop acetyl-protected aminomethyl quinoline ligands that incorporate this coordinating moiety. Disappointingly, such ligands (L6 to L8) resulted in a complete loss of reactivity (Fig. 2A and table S1). We reasoned that the five-membered bidentate chelation with Pd(II) could result in the formation of a stable but inactive palladium complex tetra-coordinated with two ligands. As such, we prepared APAQ and aminopropyl quinoline ligands (L9 and L10) that would coordinate with Pd(II) via six- and seven-membered chelate structures, respectively, both of which should have markedly reduced binding constants compared to the corresponding five-membered chelate (L6). Such subtle modification restored the reactivity with L9 and L10, thus offering a bidentate ligand scaffold for further development.

Fig. 2 Ligand optimization for enantioselective methylene C–H arylation.

(A) Yields were determined by 1H nuclear magnetic resonance analysis of the crude product, using CH2Br2 as an internal standard. Enantiomeric ratios were determined by chiral high-performance liquid chromatography. The absolute configuration of L35 was determined by x-ray crystallography (fig. S4). p-Tol-I, para-tolyl iodide; equiv., equivalent; HFIP, hexafluoro-2-propanol; Me, methyl group; Ph, phenyl group; Pr, propyl group; Bn, benzyl group; N.R., no reaction. The asterisk indicates the presence of a chiral center at the atom. (B) DFT-optimized structures and relative free energies of the two enantiomeric C–H metalation-deprotonation transition states.

Although aminopropyl quinoline L10 is more reactive than aminoethyl quinoline L9, we chose to focus on the latter scaffold because of its synthetic accessibility. We used Ellman’s highly efficient asymmetric imine addition reaction (30) to prepare a series of chiral APAQ ligands from 2-methylquinoline and optically pure sulfinyl imines . We initially found that ligand L11 (table S1), containing an α-methyl group at the chiral center, enhanced the reactivity considerably (75% yield), albeit with poor enantioselectivity (47:53 er). The α-methyl group was then replaced with various alkyl groups: Only the sterically bulky isopropyl group was found to produce a substantially improved enantiomeric ratio (27:73), but it also led to diminished yield (L16, table S1). Although further tuning of the alkyl substitution proved less promising, the result obtained with the α-phenyl substitution in L17 provided us with an encouraging lead for ligand optimization (76% yield, 29:71 er). With L17 in hand, we examined the effect of the protecting groups on the amino group (table S2). Replacing the acetyl-protecting group by more sterically hindered analogous motifs decreased the yields considerably. Other types of protecting groups, such as carbamates and sulfonyls, proved completely inactive. We thus prepared a number of APAQ ligands [L18 to L33 (table S1)] with a range of steric and electronic variation on the α-phenyl ring. We found that the steric effect is predominant, as indicated by the markedly improved yield and enantioselectivity obtained with ligand L32 bearing the sterically hindered 3,5-di-tert-butylphenyl group (85% yield, 19:81 er). At this point of optimization, we introduced a second chiral center at the benzylic position, hoping to further improve the enantioselectivity. Attributing the origin of the stereoselectivity to differentially hindered faces on the square planar palladium complex (811), we focused on the variations of syn-APAQ ligands in which both substituents would point up- or downward upon chelating with Pd(II). The introduction of a methyl group at the benzylic position (L34) afforded a substantial improvement in enantioselectivity (90:10 er) while maintaining the high yield. A slightly more bulky ethyl group (L35) further improved the enantioselectivity to 92.5:7.5 er. Further increasing steric hindrance at the benzylic position decreased both yield and enantioselectivity (L36 to L39). To obtain insight into the stereochemical model of this enantioselective palladium-insertion process, we also tested the anti-APAQ ligands (L40 and L41). Although both yield and enantioselectivity dropped considerably with these two anti-ligands, the reversal of chiral induction by altering the absolute configuration at the α position suggests that the chiral center adjacent to the amino group dictates enantioselection (table S3).

Building on results from previous computational studies (28, 29), we explored the enantioselectivity in the C–H metalation-deprotonation step with density functional theory (DFT), using ligand L34 and substrate 1. The optimized structures and relative free energies of the most favorable transition states for the two enantiomers are shown in Fig. 2B. In both transition states, the palladium(II) center is coordinated with the quinoline nitrogen and the amide nitrogen of the deprotonated APAQ ligand, as well as one nitrogen from the amide substrate formed by deprotonation. The oxygen of the amide of the APAQ ligand acts as an intramolecular base to deprotonate the methylene C–H and facilitate the Pd–C bond formation. TS_R is more favorable than TS_S by 1.2 kcal/mol, in excellent agreement with the experimental selectivity (90:10 er). The distinct six-membered bis-chelate design and the bulky α-aryl substituent orient the amide in such a way that the β-C–H bond is activated. The terminal methyl group of the substrate is oriented differently in the R and S transition states, with more severe substrate distortion in TS_S, where the methyl group of the substrate methylene is on the same face as the bulky α-aryl substituent of the ligand and the peri-hydrogen of the quinoline. This discrepancy underlies the enantioselectivity (see supplementary materials).

With the optimal ligand L35 in hand, we further optimized the reaction conditions for the arylation of 1 and improved the enantioselectivity to 95:5 er (table S4, entry 21). We next surveyed the scope of aryl iodides for this enantioselective β-C–H arylation (Fig. 3). Arylation of 1 with simple iodobenzene afforded the desired product in 89% yield and 95:5 er (2b). A wide range of para-substituted aryl iodides was employed as coupling partners. Electron-donating groups afforded excellent enantioselectivity and good yields (2a, 2c, and 2d). Electron-deficient aryl iodides bearing trifluoromethoxy, fluoro, chloro, bromo, ketone, and ester substituents were also compatible, providing consistently high enantioselectivity (2e to 2h, 2j, 2k), although the yield dropped to 45% with trifluoromethyl substitution (2i). Similar results were obtained with metasubstituted aryl iodides (2l to 2p). Although slightly lower enantioselectivity was obtained with ortho-methoxylphenyl iodide (2q), the ortho-ester group was well tolerated (2r). An aryl iodide containing phosphonate moiety also afforded synthetically useful yield and enantioselectivity (2s). Disubstituted aryl iodides also proved to be suitable coupling partners (2t to 2v).

Fig. 3 Scope of aryl iodides in enantioselective methylene C–H arylation.

Data are reported as isolated yield of purified compound. The absolute configurations of 2a, 2e, and 2k were determined by x-ray crystallography (figs. S5 to S7).

This protocol for enantioselective arylation of methylene C–H bonds is also applicable to other aliphatic amides (Fig. 4). Aliphatic amides with various chain lengths were well tolerated, providing products 4a to 4d with excellent enantioselectivity and in high yields. The direct enantioselective β-C–H arylation of amides derived from naturally occurring caprylic acid and myristic acid demonstrates the utility of this method when a double bond is not available for conjugate addition (4c and 4d). Substrates containing sterically hindered alkyl groups at the β position (cyclopentyl, clyclohexyl) provided good enantioselectivity but lower yields (4e and 4f). Isopropyl, cyclopentyl, cyclohexyl, and cyclohexylmethyl moieties at the γ positions were well tolerated, affording satisfactory yields and enantioselectivity (4g to 4j). Phenyl, ester, amino, ether, and ketone functionalities at the δ and ε positions consistently afforded high enantioselectivity (4k to 4p). However, lower yields were obtained with the ether and ketone substrates (4o and 4p). Piperidine at the γ position afforded good yield and high enantioselectivity (4q), whereas the presence of piperidine at the β position gave lower yield (4r). The presence of a tetrahydropyran motif at the γ position was also well tolerated, affording synthetically useful yield and enantioselectivity (4s). The enolate of the corresponding ester of product 4s reacted with di-tert-butyl azodicarboxylate to give the chiral α-amino ester in >20:1 diastereoselectivity, providing a route for the preparation of complex chiral amino acids (see supplementary materials). Arylation of benzylic C–H with 3t using ligand L35 provided poor yield and enantioselectivity (38% yield, 68:32 er). Switching to ligand L32 resulted in a marked improvement in both yield and enantioselectivity (4t). β-phenyl groups containing both electron-withdrawing and -donating groups were also compatible with this reaction (4u and 4v). To demonstrate the advantage of using weakly coordinating monodentate substrates and chiral bidentate ligands, we applied the enantioselective β-C–H arylation protocol to two free carboxylic acids to provide 4w and 4x as cis-diastereomers exclusively. These products are not accessible by asymmetric conjugate addition.

Fig. 4 Scope of amides and acids in enantioselective methylene C–H arylation.

Data are reported as isolated yield of purified compound. Asterisks in the structures indicate the presence of a chiral center at the atom. The * symbol following a compound number indicates 1.5 equiv. of Ag2CO3, 2.5 equiv. aryl iodide, and 12 mole % (mol %) L32. The absolute configuration of 4u was determined by x-ray crystallography (fig. S8). The † symbol indicates 1.0 equiv. K2HPO4 as additive, 20 mol % L16, and 100°C. The ‡ symbol indicates 1.0 equiv. Na2HPO4·7H2O as additive, 2.0 equiv. AgOAc, 2.0 equiv. aryl iodide, and 12 mol % L32. The § symbol indicates that yields and er values were determined from corresponding methyl esters (see supplementary materials for experimental details). Phth, phthalimido group; Ts, tosyl group.

In summary, a chiral bidentate APAQ ligand scaffold was found to enable Pd-catalyzed enantioselective arylation of prochiral β-C–H bonds on a single methylene carbon center. The feasibility of asymmetric palladium insertion into ubiquitous methylene C–H bonds with monodentate substrates opens a new avenue for developing a wide range of synthetically useful enantioselective C–H activation reactions.

Supplementary Materials

Materials and Methods

Figs. S1 to S9

Tables S1 to S14

NMR Spectra

HPLC Traces

References (3155)

References and Notes

Acknowledgments: We thank The Scripps Research Institute and the NIH (National Institute of General Medical Sciences grant 2R01GM084019) for financial support. We also thank the Shanghai Institute of Organic Chemistry, Zhejiang Medicine, and Pharmaron for a postdoctoral fellowship (G.C.) and the Deutsche Forschungsgemeinschaft for a research fellowship (M.S.A.). We acknowledge earlier computational studies that enhanced our understanding of catalyst development, conducted in collaboration with J. Musaev and K. Houk within the NSF Center for Chemical Innovation: Center for Selective C–H Functionalization (grant CHE-1205646). Author contributions: J.-Q.Y. conceived the concept. G.C. and W.G. developed the chiral ligands and optimized the reaction conditions. Z.Z., M.S.A., Y.-Q.C., and T.L. optimized the chiral ligands and surveyed the substrate scope. X.H., Y.-F.Y., and K.N.H. performed the DFT calculation. J.-Q.Y. directed the project. J.-Q.Y. and The Scripps Research Institute have filed a provisional patent application (62311039). Metrical parameters for the structures of L13, L21, L40b, L35, 2a, 2e, 2k, and 4u (see supplementary materials) are available free of charge from the Cambridge Crystallographic Data Centre under reference number CCDC-14522020, CCDC-14522021, CCDC-14522023, CCDC-14522022, CCDC-14522016, CCDC-14522017, CCDC-14522018, and CCDC-14522019, respectively.
View Abstract

Stay Connected to Science

Navigate This Article