Asymmetric synthesis of N,O-heterocycles via enantioselective iridium-catalysed intramolecular allylic amidation

Chiral N,O-heterocycles are important structural motifs found in a variety of natural products which exhibit signicant biological activities including anti-tumor, antifungal and antibacterial properties. Among these heterocycles, oxazolines, oxazines, benzoxazines and their derivatives are of particular importance. Efavirenz (Sustiva), which features a 4H-3,1-benzoxazin-2-one core, is a non-nucleoside reverse transcriptase inhibitor and is used for the treatment of human immunodeciency virus (HIV) type 1. 4H-3,1-benzoxazin-4-one derivatives are able to inhibit C1r serine protease. Furthermore, N,Oheterocycles play a prominent role in asymmetric synthesis as they are core structures of several chiral ligands used in asymmetric catalysis. The preparation of this class of heterocycles and their derivatives usually relies on the use of chiral amino alcohols. However, direct catalytic asymmetric syntheses of these chiral heterocycles remain challenging. As far as we know, it was not until very recently that the rst catalytic methods were reported, which are all based on asymmetric halocyclization of amides (Scheme 1a). In order to expand chiral space in heterocyclic chemistry, the development of complementary asymmetric routes to N,O-heterocycles with diverse functional groups is highly desirable. Asymmetric allylic substitution (AAS) represents a powerful transformation for the assembly of congested chiral architectures. In particular, Ir-catalysed AAS features high reactivity and allows stereoselective construction of chiral branched


Introduction
Chiral N,O-heterocycles are important structural motifs found in a variety of natural products which exhibit signicant biological activities including anti-tumor, antifungal and antibacterial properties. 1 Among these heterocycles, oxazolines, oxazines, benzoxazines and their derivatives are of particular importance. Efavirenz (Sustiva), which features a 4H-3,1-benzoxazin-2-one core, is a non-nucleoside reverse transcriptase inhibitor and is used for the treatment of human immunode-ciency virus (HIV) type 1. 2 4H-3,1-benzoxazin-4-one derivatives are able to inhibit C1r serine protease. 2b Furthermore, N,Oheterocycles play a prominent role in asymmetric synthesis as they are core structures of several chiral ligands used in asymmetric catalysis. 3 The preparation of this class of heterocycles and their derivatives usually relies on the use of chiral amino alcohols. 2,4 However, direct catalytic asymmetric syntheses of these chiral heterocycles remain challenging. As far as we know, it was not until very recently that the rst catalytic methods were reported, which are all based on asymmetric halocyclization of amides (Scheme 1a). 5 In order to expand chiral space in heterocyclic chemistry, the development of complementary asymmetric routes to N,O-heterocycles with diverse functional groups is highly desirable.
Asymmetric allylic substitution (AAS) represents a powerful transformation for the assembly of congested chiral architectures. 6 In particular, Ir-catalysed AAS features high reactivity and allows stereoselective construction of chiral branched allylic compounds. 7 Since the pioneering work of Helmchen 8 and Hartwig, 9 several iridium-catalysed asymmetric allylic substitution reactions with phosphoramidites 10 as chiral ligands have been developed. To date, a variety of nucleophiles including C-, 11 N-, 12 O-13 and S-nucleophiles 14 have been used in this type of reaction. We recently developed the rst intramolecular Ir-catalysed asymmetric allylic amidation. 15 Through nucleophilic attack by the amide nitrogen atom, chiral tetrahydroisoquinolines and saturated N-heterocycles could be obtained with high enantioselectivities. However, to the best of our knowledge, transition metal-catalysed asymmetric allylic substitution with nucleophilic attack by the amide oxygen atom is unprecedented. 16 In view of our continuing interest in developing new catalytic methodologies for the construction of chiral heterocycles, 17 we decided to study the asymmetric synthesis of N,O-heterocyclic derivatives. We envisioned that reversing the selectivity of the ambidentate amide nucleophile in the iridium-catalysed intramolecular AAS could provide a versatile method to access chiral oxazolines, oxazines and benzoxazines (Scheme 1b). Herein, we report the successful realization of this catalytic transformation i.e. iridium-catalysed intramolecular allylic substitution with nucleophilic attack by the amide oxygen atom. The resulting N,O-heterocyclic derivatives can be readily obtained in high yields and excellent enantioselectivities (up to 97% ee).

Results and discussion
Our preliminary studies started with the intramolecular reaction of 2-benzamido cinnamyl carbonate 1a with 5 mol% of [Ir(cod)Cl] 2 in the presence of 10 mol% of phosphoramidite L1 at 50 C. We were pleased to nd that the anticipated reaction proceeded smoothly to provide the benzoxazine product 2a with 31% ee, without formation of any S N 2 by-products resulting from attack by either N or O atoms (Table 1, entry 1).
Encouraged by this result, we examined a series of related phosphoramidite ligands L2-L6 (Table 1, entries 2-6). It turned out that the reaction with L2 (ref. 18) at RT led to a promising increase in ee, however, only modest conversion was achieved (entry 7, 79% ee). It has been reported that in allylic substitution with the iridium catalysts derived from commonly used ligands L1 and L2, ortho-substituted cinnamyl substrates in general are not tolerated, with much decreased enantioselectivities observed in many cases. 9a,11h,13d,19 The use of tetrahydroquinoline-derived phosphoramidite ligand L7 gave rise to full conversion and furnished the cyclization product with 88% ee at 50 C and 95% ee at RT (entries 8,9). Related ligand L8, which features an additional stereogenic center and was reported to give good results in AAS of cinnamyl substrates, 20 provided a slightly lower ee in the present transformation (entry 10). It should be noted that the active catalyst generated from L1 or L2 features an Ir-C(sp 3 ) bond in accordance with the structure shown by Hartwig, 9 while the active catalyst derived from L7 or L8, introduced by You's group, incorporates an Ir-C(sp 2 ) bond instead. 20 With a highly selective catalyst based on ligand L7 in hand, we further optimized the reaction conditions (Table 2). Evaluation of several bases indicated that the reaction was slower with inorganic bases such as Cs 2 CO 3 and K 2 CO 3 , probably due to poor solubility (Table 2, entries 2 and 3), whereas the use of DABCO as a base led to a slightly higher ee (97% ee vs. 95% ee) and a similar reaction rate as compared with DBU (entries 1 and 4). With DABCO as the optimal base, the reaction was then performed with 2.5 mol% of [Ir(cod)Cl] 2 at room temperature (entry 5). However, the reaction turned out to be very slow and did not reach full conversion under these conditions. Gratifyingly, when the reaction was performed at 35 C with 5 mol% (2.5 mol% of [Ir(cod)Cl] 2 ) catalyst, the ee was not affected and the reaction was complete when 3 equiv. of DABCO were used ( Table 2, entries 6 and 7). Importantly, the enantioselectivity could be maintained and the reaction was completed in a shorter time when carried out at 50 C (Table 2, entry 8, 81% yield, 97% ee). As shown in Table 2, the enantioselectivity and reaction rate can both be enhanced with additional equivalents of DABCO. It seems that this catalytic conversion tolerates higher temperature when DABCO is used as the base. In contrast, the use of DBU as the base under the same conditions led to 2a with only 89% ee (Table 2, entry 9). Under the optimized reaction conditions ( Table 2, entry 8), we explored the substrate scope of this new catalytic allylic cyclization (Table 3). High tolerance for substituents at the paraposition of the benzamide aryl group was observed, irrespective of the electronic and steric properties (Table 3, 2b-2d and 2g-2i). Substrates 1j and 1k with substituents at the meta-and ortho-position of the benzamide aryl group were also tested, and it was shown that 2j with meta-Cl gave a reduced ee, while ortho-Cl substituted substrate 1k provided 2k with 97% ee. Heteroarylamides 1e and 1f were also suitable substrates for the present cyclization reaction. Benzoxazine 2n with a t-butyl substituent at the 2-position could be obtained under standard reaction conditions in 63% yield and 86% ee. Notably, substrates 1l and 1m with substituents at the C4 and C5 position of the cinnamyl carbonate gave the desired products 2l and 2m in good yields and enantioselectivities.
To further demonstrate the versatility of the reaction, we examined some aliphatic substrates to provide the corresponding chiral oxazoline and oxazine. Under the optimized reaction conditions for the enantioselective synthesis of benzoxazines 2, the reaction of aliphatic substrates 1o and 1p gave rise to a complex mixture of products. However, through a slight modication of the reaction conditions, i.e. running the reaction at room temperature and using DBU (0.5 equiv.) as the base instead of DABCO, the corresponding chiral oxazoline 2o and oxazine 2p were obtained with excellent ee (Scheme 2).
In order to evaluate the synthetic applicability of this new catalytic method, a number of transformations of the benzoxazine products were carried out (Scheme 3). The allylation products of this transformation feature a terminal double bond which allows for further functionalization. As an example, the vinyl moiety of 2a underwent cross-metathesis with ethyl acrylate in the presence of the Grubbs-Hoveyda II catalyst without compromising the enantiomeric excess of product 3. Table 3 Ir-catalysed synthesis of chiral benzoxazines 2a-n a a Reaction conditions: 2.5 mol % [Ir(cod)Cl] 2 , 5.0 mol % of ligand, 0.2 mmol of 1, 0.6 mmol DABCO in 2.0 mL THF at 50 C for 24 h. Yields of isolated products aer column chromatography; enantiomeric excess determined by HPLC or GC analysis (see the ESI †). b The absolute conguration of 2m was determined as R by X-ray crystallographic analysis (Fig. 1). 21 In view of the synthetic challenges to prepare chiral hemiaminals, 22 we also explored the 1,2-addition of organolithium reagents to chiral benzoxazine 2a. 23 As shown in Scheme 3, a series of linear and branched organolithium reagents were successfully employed, furnishing the chiral N,O-ketals 4a-4d bearing both a tertiary and a quaternary stereogenic center with excellent diastereoselectivities. The relative congurations of 4a and 4d were determined by NOESY studies. 24 Surprisingly, the X-ray analysis of 4a (ref. 21) conrmed that the methyl group is installed cis to the vinyl group (Fig. 1). It is expected that under kinetic control, the alkyl nucleophile would prefer to attack from the opposite side of the vinyl group due to steric hindrance. 23,25 While the detailed mechanism is as yet unclear, on the basis of control experiments, 26 we propose that there is an equilibrium between the closed acetal form and open imine form of the product formed in the reaction (Scheme 4). Through the ring opening and subsequent closing, the kinetically favored intermediate epimerizes in situ to the thermodynamically more stable intermediate, although alternative pathways cannot be excluded.

Conclusions
In conclusion, we have developed the rst enantioselective iridium-catalyzed intramolecular allylic substitution reaction with nucleophilic attack by the amide oxygen atom. A series of N,O-heterocycles, i.e., oxazolines, oxazines and benzoxazine derivatives, could be readily obtained with high enantioselectivities. Furthermore, the resulting benzoxazine derivatives were transformed into chiral N,O-ketals bearing tertiary and quaternary centers with excellent diastereoselectivities. This methodology offers excellent opportunities to explore the "3 rd dimension" in common planar N,O-heterocycles.