Kinetic resolution of indolines by asymmetric hydroxylamine formation

Catalytic kinetic resolution of amines represents a longstanding challenge in chemical synthesis. Here, we described a kinetic resolution of secondary amines through oxygenation to produce enantiopure hydroxylamines involving N–O bond formation. The economic and practical titanium-catalyzed asymmetric oxygenation with environmentally benign hydrogen peroxide as oxidant is applicable to a range of racemic indolines with multiple stereocenters and diverse substituent patterns in high efficiency with efficient chemoselectivity and enantio-discrimination. Late-stage asymmetric oxygenation of bioactive molecules that are otherwise difficult to synthesize was also explored.

E nantiopure cyclic secondary amines are key constituents of natural products, pharmaceuticals, and agricultural chemicals 1 . Catalytic kinetic resolution (KR) of racemic amines represents a practical and robust approach to access optically pure targets, especially in cases where the racemates are readily available and the enantiopure materials are not [2][3][4][5][6][7][8] , though the theoretical yield of the expected optically pure target can never exceed a limit of 50% [9][10][11] . The current nonenzymatic KR of secondary amines predominantly relies on asymmetric N-acylation strategy involving N-C bond formation, which typically requires the use of stoichiometric pre-prepared acylating agents involving lengthy reagent synthesis (Fig. 1a) [12][13][14][15][16][17][18] . Development of a catalytic KR of secondary amines based on other elementary reaction bearing an economy-oriented mind-set would be highly desirable.
To the best of our knowledge, asymmetric oxygenation of secondary amines to produce hydroxylamines through either enzymatic or non-enzymatic catalysis has never been established to date, which might be ascribed to three key challenges. First, the inherent high reactivity of amines results in easy nonselective oxygenation without the intervention by a catalyst. Second, competitive dehydrogenation of secondary amines to imines usually accompanies the expected OAT process [37][38][39] . Third, chirality in hydroxylamine products could be facilely destroyed through further oxidation to nitrones 40 . On the other hand, aqueous hydrogen peroxide is a desirable oxidant from the viewpoints of atom efficiency (48%), easy-to-handle, and ecological benignity 32,41 . Given the significance of optically pure indolines in modern pharmacology, we herein report a titanium-catalyzed KR of indolines using H 2 O 2 as the oxidant through N-O bond formation. Late-stage asymmetric oxygenation of bioactive molecules that are otherwise difficult to synthesize was also explored.

Results
Reaction condition optimization. Initially, asymmetric oxygenation of racemic indoline 1a was selected as a reference reaction using aqueous H 2 O 2 as the oxo-transfer agent to search for a suitable chiral catalyst (Table 1). Chiral monomeric (salen)titanium(IV) C1 exhibited no oxidation catalysis reactivity (entry 1). We then explored dimeric metallosalen complex as catalyst. Delightedly, di-μ-oxo titanium(salen) C2 effected the expected asymmetric oxygenation, though poor chiral recognition and notable over-oxidation were observed (entry 2). The substituent patterns on 1,2-ethanediamine proved to be crucial to catalytic reactivity and asymmetric induction. Replacing the 1,2-cyclohexanediamine moiety in C2 with 1,2-diphenylethylenediamine one (C3) significantly reduced the oxidation catalytic reactivity (entry 3). Di-μ-oxo titanium(salalen) C4 was prepared by reducing one of the process (entry 4). Displacing the phenyl group on salicylaldehydes with other two imine bonds of C2, was beneficial for suppressing undesired over-oxidation substituents afforded inferior results, which prompted us to introduce another chiral element at C 3 site of the basal salalen ligand to enhance the enantio-differentiating ability of the catalyst. The "hybrid" titanium(salalen) C5 bearing a (R a )-binaphthyl unit on the imine side was not an effective catalyst (entry 5). Promising chiral recognition was observed when di-μ-oxo titanium(salalen) catalyst C6 bearing two (R a )-binaphthyl units on both amine and imine sides were used (entry 6). The absolute configuration of (R)-1a was assigned to be R by comparing the optical rotation and HPLC analysis with reported data. See the Supplementary Information for details. Reversing the absolute configuration of 1,2-cyclohexanediamine in catalyst C7 provided a mismatch recognition (entry 7). Extensive optimization of the solvent identified CHCl 3 to be optimal (entries 8-10). The level of chiral discrimination was further enhanced by lowering the loading of titanium(salalen) C6 from 2.5 mol % to 1.0 mol %, though a slightly prolonged time period was required (entry 11). Asymmetric oxygenation of indolines bearing one stereocenter. The scope of di-μ-oxo titanium(salalen)-catalyzed asymmetric oxygenation of racemic indolines was explored (Fig. 2) 40 . Substrates 1a-1k bearing a wide range of electronically varied aryl and heteroaryl groups at C 2 position with different substituent patterns proceeded with excellent chemoselectivity and chiral recognition (Fig. 2a). Indolines 1l-1s bearing diverse C 2 -alkyl substituents were suitable components with excellent selectivity factors (Fig. 2b). Diverse functional groups including phenyl motif (1o and 1p), silyl ether (1q), carboxylic acid ester (1r), and terminal alkyne (1s) were tolerated as additional functional handle. Spirocyclic 1r and 1s containing the variant of the geminal disubstitution at C 3 -position were also tolerated (Fig. 2c). Simple indoline 1t without C 3 -substituent was competent substrate, though the oxidized hydroxylamine 2t was unstable and underwent decomposition during purification probably due to the existence of reactive benzylic C 3 -H bonds.
The substituent effects on the indoline arene were then investigated (Fig. 3). A range of electron-withdrawing and -donating substituents at either C 4 , C 5 , or C 6 position of substrates 4a-4j were tolerated with high level of chiral discrimination (Fig. 3). Bromo (4a-4c), chloro (4d), and fluoro (4e) substituents were compatible with the oxidation system for further diversification.

Discussion
To get a preliminary understanding of the catalysis role of titanium(salalen) C6 in asymmetric oxygenation of indolines, conversion (%) of rac-1a was plotted against time (min) for reaction under standard conditions and that with pre-mixed catalyst and H 2 O 2 , respectively (Fig. 6A). For the standard reaction, conversion followed a sigmoidal curve, and an induction period of about 30 min was observed. Such induction period was not observed in the reaction using pre-mixed C6 and H 2 O 2 , suggesting that the generation of the real active species for oxygenation reaction is a slow process. Moreover, the latter reaction exhibited an apparent steady-state regime, resulting in linear ln([S S ]/[S S ] 0 ) vs. time dependence (Fig. 6B).
Control experiments were performed to further understand the identity of the generated species by mixing C6 with H 2 O 2 . The relationship between the excesses of 2a and the catalyst C6 was investigated, and a positive nonlinear relationship was observed, suggesting the possible intervention of a dimeric complex (Fig. 6C) 54,55 . ESI mass-spectrometry analysis of the mixture of diμ-oxo titanium(salalen) C6 and 30% aqueous H 2 O 2 after 30 min showed the peak of 1797.6, which is equal to [M C6 + O + H] + , implying the formation of a μ-oxo-μ-peroxo species S (Fig. 7a) 56 . However, a lower intensity of this peak was observed for mixing C6 and H 2 O 2 after 15 min, implying that the formation of S might be a slow process. No reaction was observed when mixing one equiv of S with rac-1a (Fig. 7b). However, the addition of aqueous hydrogen peroxide rendered S catalytically active with comparable results to C6 (Fig. 7c). These results clearly indicate that S is neither the real active species nor a dead-end species but an active intermediate. Crossover experiments involving two different catalysts C4 and C6 were next performed. Mass spectra analysis of equivalent C4 and C6 in the presence of 30% aqueous H 2 O 2 showed the peak of 1445.5, which is equal to [M C4 /2 + M C6 /2 + O + H] + , suggesting the generation of a crossover dimerization peroxo complex (Fig. 7d). No such peak was detected during the mass spectra analysis of the mixture of C4 and C6 in the absence of H 2 O 2 (Fig. 7e). These observations indicated that a disassembly and reassembly process might occur after the formation of μ-oxoμ-peroxo species S. We envisioned that the real active species involved in the catalytic cycle might be the disassembled monomeric peroxo Ti(salalen) complex 57 .
Asymmetric oxygenation of racemic indolines with various kinds of C 5 -substituents (X) on indoline arene were performed and the reaction rates were dependent on the electronic effects of the substituents (see the Supplementary Information for details) 58,59 . The Hammett plot (log(k X /k H ) versus σ) for the competitive oxidation of 1a and respective variants was exhibited in Fig. 6D. The observed plot displayed linear correlation with a ρ value of −1.347 (R 2 = 0.99). Good linearity suggests that the oxidation proceeds through a single mechanism. The negative value of ρ indicates a positive charge build-up on nitrogen in the transition state 60 . The use of Hammett parameter σ + gave a relatively poorer correlation (R 2 = 0.97, see the Supplementary Information for details). These data are consistent with a concerted mechanism, in which an electrophilic active oxygen species might be directly attacked by a nucleophilic nitrogen 61 .
Based on the above studies and literature survey, a plausible mechanistic pathway for asymmetric oxygenation of racemic indolines was suggested (Fig. 8). Treating di-μ-oxo titanium(salalen) C6 with hydrogen peroxide gave μ-oxo-μ-peroxo species S, which is the incubation period for the oxygenation process. In the presence of hydrogen peroxide, relatively stable S underwent a disassembly process providing monomeric peroxo Ti(salalen) complex 14, which was proposed to be the real active species involved in the catalytic cycle 62 reacted with hydrogen peroxide providing hydroxylamine 2a together with 14 for the catalytic cycle. According to the absolute configuration of recovered indolines, we envisioned that amine (S)-1a was oxidized more preferentially than (R)-1a, and therefore unreacted (R)-1a was isolated with high enantioselectivity. The stereochemical induction model was suggested in Fig. 9. The structure of monomeric peroxo Ti(salalen) complex 14 is proposed based on the crystal structure of di-μ-oxo titanium (salalen) C6 64,63 . The quadrant diagrams elaborate the origins of the chiral recognition. The steric repulsion between two (R)binaphthyl unites and indoline 1a prevents the nucleophilic attack of the nitrogen from the third and fourth quadrants. In addition, locating the less sterically demanding phenyl moiety of indoline skeleton in the first quadrant is preferred to avoid the steric repulsion between the (R)-binaphthyl unite and the αsubstituent of 1a. The transition state TS2 for (R)-1a has obvious steric repulsion between the α-phenyl substituent of indoline with the cyclohexane moiety of salalen ligand. Such steric repulsion is absent in the transition state TS1 for (S)-1a. Therefore, (S)-1a was oxidized more preferentially than (R)-1a, and unreacted (R)-1a was recovered with high enantioselectivity.
In summary, an oxidative KR of secondary amines based on N-O bond formation is reported. The practical titanium(salalen) catalyzed asymmetric oxygenation with environmentally benign hydrogen peroxide as oxidant is applicable to a range of indolines with multiple stereocenters and diverse substituent patterns in high efficiency with efficient chemoselectivity and enantiodiscrimination. Late-stage asymmetric oxygenation of bioactive molecules that are otherwise difficult to synthesize was further explored. The KR of secondary amines based on N-O bond formation described herein represents an advance in the field of asymmetric oxidation.