Gold-catalyzed efficient synthesis of azepan-4-ones via a two-step [5 + 2] annulation

Abstract

A surprisingly efficient synthesis of azepan-4-ones via two-step [5 + 2] annulation is developed. This reaction involves a key gold catalysis and shows generally high regioselectivities and good to excellent diastereoselectivities.

Azepane, a seven-membered N-heterocycle, is a building block frequently found in various natural products¹ including stenine,^1b galanthamine,^1c and croomine^1d and a large array of compounds studied in medicinal chemistry.² Although functionalized azepanes can be prepared via various approaches,³ methods based on synthetically flexible and versatile cycloaddition or annulation approaches are still much needed, and only a few examples⁴ have been reported. Herein we report a gold-catalyzed efficient and flexible synthesis of azepan-4-ones via a two-step [5 + 2] annulation.

We previously reported a two-step synthesis of piperidin-4-ones in a two-step [4 + 2] manner.⁵ It is speculated that a similar [5 + 2] sequence might be possible, which would afford synthetically versatile azepan-4-ones (Scheme 1).⁶ This sequence might involve a gold carbene intermediate (i.e., A) via a gold-catalyzed intramolecular alkyne oxidation and require favoring a challenging formal 1,7-C(sp³)–H insertion by carbene A⁷ over a seemingly more feasible formal 1,5-C(sp³)–H insertion in the ring formation step.^8,9

Scheme 1.

A two-step [5 + 2] annulation toward the synthesis of azepan-4-ones?.

We tested the hypothesis, nevertheless, by subjecting N-(pent-4-yn-1-yl)piperidine (1) first to m-CPBA oxidation and then to Ph₃PAuNTf₂¹⁰ catalysis without isolating the N-oxide intermediate (Scheme 2). The major product was isolated in 53% yield and, to our delight and surprise, assigned as bicyclic azepan-4-one 2 instead of cyclopentanone 3. This structural assignment was initially supported by the lack of symmetry in the ¹³C NMR spectrum of deoxygenated 2 (i.e., compound 4) and later corroborated by X-ray crystallography (vide infra). No cyclopentanone 3 or its dehydroamination product, cyclopentenone, was observed. Different gold(i) catalysts as well as AuCl₃ and PtCl₂ were screened, and (2-biphenyl)Cy₂PAuNTf₂¹¹ was found to be the most efficient, furnishing 2 in 79% isolated yield. Of note, Tf₂NH did not catalyze this reaction.

Scheme 2.

Initial studies.

The scope of this surprising chemistry was studied in a two-step sequence: (1) alkylation with pent-4-yn-1-yl tosylate (2 equiv.) in refluxing MeCN for 12 h using K₂CO₃ as base; (2) m-CPBA (1 equiv.) oxidation followed by (2-biphenyl)-Cy₂PAuNTf₂ (5 mol%) catalysis at 0 °C. As shown in Table 1, the alkylation step was expectedly efficient and, to our delight, the one-pot oxidation/gold catalysis was in general efficient as well; together, these two steps constituted an efficient synthesis of azepan-4-ones via [5 + 2] annulation. For symmetric secondary amines (entries 1–4), the two-step sequence tolerated phenyl (entry 1) or TBSO (entry 2) groups and allowed the formation of bicyclic azepan-4-ones fused with either a pyrrolidine (entry 3) or another azepane ring (entry 4) in good yields.

Table 1.

Synthesis of azepan-4-ones via a two-step, 5 + 2 annulation process: scope study

			Yieldb (%)
Entrya	Substrate	Product	1st step	2nd step
1			99	87
2			98	51c
3			88	80
4			83	89
5			78	69
6			85	85
7			93	73d,e
8			91	71c
9			80	70
10			99	63
11			90	93d,g
12			87	74
13			84	76h
14			90	71d

^aThe reaction concentration was 1 M for the first step and 0.05 M for the second step.

^bRegioselectivity, if not indicated, is >20: 1.

^cReaction time: 8 h.

^dEt₃PAuNTf₂ (5 mol%) was used instead.

^eReaction time: 6 h.

^fAn inseparable mixture.

^gA ratio of 1.2/1 was observed using (2-biphenyl)Cy₂PAuNTf₂ as catalyst.

^hReaction time: 3.5 h.

Fortuitously, 6a was crystalline, and its azepane skeleton was confirmed by X-ray crystallography (Fig. 1).

Fig. 1.

ORTEP drawing of compound 6a.

For non-symmetric secondary amines, this chemistry was sensitive to steric difference and in most cases good to excellent regioselectivities were observed. For example, a Me group participated in the ring formation highly selectively over a primary alkyl (entry 5) or a benzyl (entry 6) group, yielding only one azepan-4-one product in each case. Comparing an n-butyl and a benzyl group was rather revealing. As shown in entry 7, the former was surprisingly favored over the latter albeit to a small extent, which, however, could be rationalized by the fact that phenyl is bigger than n-propyl. A further example substantiated this rationale: with a bulkier 2-bromophenyl group, the selectivity was increased to >10 : 1 (entry 8). Notably, the 2-bromobenzyl group is a removable group and can be employed in radical translocation reactions.¹² While clearly sterics outplayed electronics in determining regio-selectivity, under similar steric environment, however, electronic difference could provide significant regioselectivity. For example, 1,2,3,4-tetrahydroisoquinoline underwent exclusive formal insertion into its benzylic C–H bond, yielding tricyclic azepan-4-one 6i in 70% yield (entry 9). While no significant selectivity between PMB (p-MeOBn) and Bn was observed (entry 10), the low selectivity in the case of 5k (6k/6k′ = 1.2/1) was surprising and in contrast to what was previously observed in the synthesis of piperidin-4-ones (exclusive insertion into the butyl group).^5a The selectivity was improved to 5/1 using Et₃PAuNTf₂ as catalyst (entry 11). The sensitivity of this chemistry to sterics was further evident as the gold catalysis in the case of sterically demanding N-butyl-3-pentanamine proceeded sluggishly and no azepan-4-one product was isolated. In contrast, 2-methylpiperidine reacted smoothly to yield bicyclic azepan-4-one 6l in 74% yield (entry 12). Although the NH group of 2-methylpiperidine is similarly flanked by a secondary and a primary carbon centers, the ring structure likely alleviated steric hindrance for the gold catalysis. Importantly, this reaction showed good diastereo-selectivity and excellent regioselectivity. Even better diastereoselectivities were observed in the cases of 4-methylpiperidine (entry 13) and methyl prolinate hydrochloride (entry 14). While (2-biphenyl)Cy₂PAuNTf₂ was generally the catalyst to use, in some cases (entries 7, 11 and 14), Et₃PAuNTf₂ gave better results. Substitutions on the pent-4-ynyl chain were readily tolerated at the 1, 2 and 3 positions (eqn (1)-(3)), leading to azepan-4-ones with substitutions at the 5, 6 and 7 positions in fairly good yields. Notably, decreased regioselectivity was observed with substrate 11, indicating the subtlety of regiochemistry control.

(1)

(2)

(3)

We have discovered a gold-catalyzed efficient synthesis of azepan-4-ones via a two-step [5 + 2] annulation. This reaction is sensitive to steric differences and can in general be highly regioselective. Good to excellent diastereoselectivities can be achieved. This chemistry opens an easy, flexible, and efficient route to access various azepane derivatives. The detailed mechanism of this surprising reaction is currently probed and the results will be reported in due course.