Expedient Synthesis of Fused Azepine Derivatives using a Sequential Rh(II)-Catalyzed Cyclopropanation/1-Aza-Cope Rearrangement of Dienyltriazoles

Abstract

A general method for the formation of fused dihydroazepine derivatives from 1-sulfonyl-1,2,3-triazoles bearing a tethered diene is reported. The process involves an intramolecular cyclopropanation of an α-imino Rh(II)-carbenoid, leading to a transient 1-imino-2-vinylcyclopropane intermediate which rapidly undergoes a 1-aza-Cope rearrangement to generate fused dihydroazepine derivatives in moderate to excellent yields. The reaction proceeds with similar efficiency on gram-scale. The use of catalyst-free conditions leads to the formation of a novel [4.4.0] bicyclic heterocycle.

A zepine and azepane derivatives are found in numerous bioactive natural products and other compounds of pharmaceutical interest (Figure 1).^[1–2] While a vastarray of methods have been developed through the years for their synthesis,^[1] only a few approaches exist for the preparation of their ring-fused analogs.^[1b] The ubiquity of these units in a variety of biologically relevant compounds such as alkaloids (see Figure 1) makes the development of new stereoselective strategies to access polycyclic, substitute dazepines and azepanes in a single operation from readily available acyclic precursors highly desirable. The [3,3]-sigmatropic rearrangement of 1,2-divinylcyclopropane derivatives is a well-established strategy to access 7-membered ring compounds in a stereospecific manner.^[3] The analogous transformation where one of the vinyl groups is replaced by a C=N group is known to lead to azepine derivatives through a similar mechanism. While 2-aza-Cope rearrangements of this type are quite common for 2-azepinone synthesis using an isocyanate intermediate (formed in situ),^[4] examples of the corresponding 1-aza-Cope rearrangement, which directly leads to 2,5-dihydro[1H] azepines, remain scarce.^[5] Indeed, the preparation of the cis-1-imino-2-vinylcyclopropanes (IVC) required for such a rearrangement to occur is hampered by the multiple steps needed for their synthesis, discouraging the use of this approach for the formation of azepine derivatives.^[5b,6]

Figure 1.

Natural products containing fused azepine derivatives.

In recent years, 1-sulfonyl-1,2,3-triazoles have emerged as stable and readily available azavinyl carbene equivalents for a variety of useful transformations.^[7–8] We have recently shown that these carbenes can be utilized in the synthesis of 3,4-fused pyrroles upon reaction with a tethered allene functionality, through a Rh-TMM intermediate (Scheme 1a).^[9] To access fused azepine derivatives through an analogous approach, we envisioned the reaction of an α-iminometallocarbene with a tethered E,E-1,3-diene would instead generate a cis-1-imino-2-vinylcyclopropane (IVC), ideally substituted for a subsequent 1-aza-Cope rearrangement (Scheme 1b). Herein, we report our studies on the synthesis of 3,4-fused dihydroazepines, which is complementary to a recent report by Tang et al. that appeared during the completion of this work.^[10] In their report, Tang et al. achieve an elegant divergent synthesis of nitrogen heterocycles by intermolecular reaction of triazoles with 1,3-dienes to provide access to azepine or pyrroline derivatives, with the pyrroline products being favored over extended reaction times.

Scheme 1.

Synthesis of fused heterocycles using azavinyl carbene equivalents

In this Communication, we report a general method for the expedient synthesis of fused azepine derivatives as the sole products from dienyltriazoles. Our mechanistic studies strongly suggest that the reaction proceeds by a sequential intramolecular Rh(II)-catalyzed cyclopropanation / 1-aza-Cope rearrangement. The stereospecific nature of each step in this transformation is critical and allows for a highly diastereoselective process, leading to fused 2,5-dihydro [1H] azepines in good to excellent yields.

Applying the conditions we previously developed for the formation of 3,4-fused pyrroles from allenyltriazoles to dienyltriazole 1a (see Table 1),^[9a] we were delighted to observe the formation of dihydroazepine 2a in 54% yield (entry 1). A major byproduct observed in the reaction was α,β-unsaturated N-tosylimine 3, resulting from a competing 1,2-hydride shift from the Rh(II)-carbenoid intermediate.^[11] To combat this challenge, we investigated a variety of more sterically encumbered Rh(II) complexes (entries 3–6).^[12] Gratifyingly, we found that Rh₂(Adc)₄ affords the desired dihydroazepine in increased yield (entry 6). Varying the nature of the solvent and the concentration of the reactants did not significantly improve the yield.^[13] Decreasing the temperature to 60 °C and increasing the reaction time to 16 h afforded dihydroazepine 2a in 74% isolated yield (entry 7), along with a minimal amount of the α,β-unsaturated N-tosylimine (3).

Table 1.

Catalyst optimization for the intramolecular Rh(II)-catalyzed dihydroazepine formation^[a]

Entry	Rh2L4	Temp.(°C)	Time	Yield 2a(%)[b],[c]
1	Rh2(Ooct)4	140[d]	0.25 h	54 (34)
2	Rh2(OAc)4	140[d]	0.25 h	47 (48)
3	Rh2(tpa)4	140[d]	0.25 h	18 (70)
4	Rh2(OPiv)4	140[d]	0.25 h	58 (18)
5	Rh2(esp)2	140[d]	0.25 h	66 (33)
6	Rh2(Adc)4	140[d]	0.25 h	68 (17)
7	Rh2(Adc)4	60	16 h	74 (14)

^[a]Only one diastereomer of the dihydroazepine product 2a was observed by ¹H NMR in all cases.

^[b]Isolated yield.

^[c]NMR yield of imine 3 in parentheses.

^[d]Reaction was performed in a microwave apparatus.

Using the optimized conditions, a range of dienyltriazole substrates is transformed into the corresponding 3,4-fused dihydroazepines (Table 2). A variety of aryl-substituted substrates, including phenyl, electron-rich and electron-poor dienyls are tolerated (entries 2–4). In addition, internal substitution of the dieneportion is compatible in the transformation, albeit proceeding in slightly lower yield to the dihydroazepine (entry 2 vs 5). Furthermore, dienyltriazole substrates bearing N-tosylamine or diester groups instead of the ether tether furnish the corresponding 3,4-fused dihydroazepines in moderate to good yields (entries 6–8). In the case of 1h, the observed yield of the corresponding product (2h) was significantly lower as a result of side-reactions, presumably arising from intramolecular attack of the proximal ester groups on the Rh-carbene intermediate. Importantly, all-carbon tethered dihydroazepine 2i, corresponding to a portion of tetrapetalone A (see Figure 1), could be accessed in excellent yield using the optimized reaction conditions (entry 9). In all cases, only one diastereomer of the dihydroazepine product is observed. Notably, dienyltriazole substrates that would yield a 6–7 fused dihydroazepine product afford only the corresponding 1,2-hydride shift product. It is noteworthy that the 1-aza-Cope rearrangement was found to proceed significantly more slowly for substrates 1d and 1e, where the IVC intermediates were observed by NMR after 0.5 h, and complete conversion was only achieved after 16 h.

Table 2.

Substrate scope of the intramolecular Rh₂(Adc)₄-catalyzed dihydroazepine formation^[a],^[b]

Entry	Substrate	Product	Time	Yield(%)[c]
1	1a	2a	16 h	74
2	1b	2b	0.5 h	92
3	1c	2c	0.5 h	61
4	1d	2d	16 h	69
5	1e	2e	16 h	63
6	1f	2f	0.5 h	72
7	1g	2g	0.5 h	72
8	1h	2h	0.5 h	45(60)[d]
9	1i	2i	0.5 h	92

^[a]PMP=4-MeO-C₆H₄-, PNP=4-O₂N-C₆H₄-, E=CO₂Me.

^[b]Only one diastereomer of the dihydroazepine product was observed by ¹H NMR in all cases.

^[c]Isolated yields.

^[d]Yield in parentheses obtained when Rh₂(Ooct)₄ was used at 140 °C in the microwave for 0.25 h.

Finally, the reaction is amenable to a gram-scale synthesis of dihydroazepines, as exemplified with 2a (Scheme 2a). Interestingly, we also found that in the absence of a Rh(II) catalyst under more forcing conditions, a distinct heterocyclic product (4) was formed in 50% yield, presumably arising from aketenimine intermediate (Scheme 2b).^[14] For both types of products (i.e., 2a and 4), the structure and relative configuration was confirmed by X-ray analysis.^[15]

Scheme 2.

Gram-scale synthesis of dihydroazepine 2a and metal-free access to [4.4.0] bicycle 4

The Rh(II)-catalyzed dihydroazepine formation can, in principle, occur through three distinct mechanistic pathways (Scheme 3). First, the azavinyl-substituted Rh-carbenoid species formed by reaction of the dienyltriazole with the Rh catalyst could undergo a [2+1] cycloaddition with the proximal alkenyl group to generate a cis-1-imino-2-vinylcyclopropane (IVC), which could undergo a [3,3] sigmatropic rearrangement to directly afford the dihydroazepine product (Path A). Alternatively, the IVC intermediate could undergo ring-opening to a zwitterionic intermediate, generating an allylcation capable of ring-closure by an intramolecular N-attack of the azanide thus formed on the distal alkenyl moiety (Path B). Similarly, the Rh-carbenoid that is formed initially could be attacked by the dienyl moiety to generate a Rh-bound zwitterion, able to cyclize via an analogous mechanism (Path C). While the stereospecificity of Path A should lead to a single diastereomer of the product bearing the predicted stereochemistry as shown, Path B or C could give rise to a mixture of both isomers. Notably, the single diastereomer2a obtained from the reaction, as determined by X-ray analysis, is consistent with the [2+1]/[3,3] sequence depicted in Path A (see Scheme 2a).

Scheme 3.

Different possible mechanistic pathways for the dihydroazepine formation

To obtain further insight, several other triazoles were synthesized and evaluated (Scheme 4). To gain support for the intermediacy of iminocyclopropanes, alkenyltriazole 1j (Scheme 4a), which lacks the distal alkenyl group required for the subsequent 1-aza-Cope to occur, was submitted to the standard reaction conditions. Iminocyclopropane 5a was obtained as a single diastereomer in 66% yield (determined by NMR). Although 1-sulfonyl-1,2,3-triazoles are known to lead to iminocyclopropanes by intermolecular cyclopropanation with alkenes,^[7–8a,b] to our knowledge, the analogous intramolecular cyclopropanation has never been reported. Thus the viability and stereo selectivity of such a mechanistic step is confirmed by this observation. Sterically encumbered dienyltriazoles 1k and 1l (Scheme 4b) lead to the formation of IVC intermediates (5b and 5c, respectively), which do not undergo the subsequent 1-aza-Cope rearrangement. This is likely due to an unfavorable interaction with the cis-R group in the transition state of the [3,3] rearrangement, significantly slowing the rate of this pathway.^[16–17] The fact that none of the dihydroazepine was observed in this particular case strongly suggests that a concerted mechanism is operative for the rearrangement, as the increased flexibility of a ring-opened zwitterionic intermediate (as shown in Path B or C) should still allow for the cyclization to occur. Notably, the formation of such zwitterionic intermediates should not be hampered by the steric hinderance of the distal alkenyl moiety in 1k or 1l. Possibly due to a similar steric effect, it is noteworthy that 1,1-disubstituted dienes, which would generate a quartenarycenter in the product, only reacted sluggishly. Finally, Z,E-dienyltriazole 1m (Scheme 4c), which leads to a trans-IVC (5d), was also found to be unreactive toward dihydroazepine formation. This observation can be attributed to the inability of such an intermediate to engage in a concerted 1-aza-Cope rearrangement. The corresponding zwitterionic intermediate formed from 1m should be identical to the case of E,E-dienyltriazole 1b, which in contrast to 1m, leads to dihydroazepine formation in excellent yield in only 0.5 h (see Table 2, entry 2). These results strongly support a sequential intramolecular Rh(II)-catalyzed cyclopropanation / 1-aza-Cope rearrangement as the operative pathway for dihydroazepine formation (see Scheme 3, Path A). Moreover, if a zwitterionic intermediate is involved in the process, the formation of the corresponding 5-membered heterocycle (pyrroline) would be expected to be competitive, as the generation of 5-membered ring compounds from this type of zwitterionic intermediate is typically a fast process.^[3f,8g,10] Notably, this type of product was not observed in any case throughout this study, again lending support to Path A.

Scheme 4.

Experimental mechanistic insights: [3,3] sigmatropic rearrangement vs zwitterionic pathways

In summary, a general approach for the synthesis of fused dihydroazepines from dienyltriazoles is reported. A range of substrates have been found to participate in the transformation, and several mechanistic investigations strongly support a sequential intramolecular Rh(II)-catalyzed cyclopropanation / 1-aza-Cope rearrangement as the operative mechanistic pathway. The reaction can be scaled with similar efficiency, and the use of catalyst-free conditions provides access to a novel [4.4.0] bicyclic heterocycle. Given the ubiquity of these scaffolds in biologically relevant compounds, this work should prove valuable for the synthesis of new and useful fused azepine-based building blocks.