Sophie Racine

The first [4+2]-annulation between aminocyclobutanes and aldehydes to access tetrahydropyranyl amines is reported. With phthalimido cyclobutane dicarboxylates and aromatic aldehydes, tetrahydropyrans were obtained in 53-92% yield and 3:117:1 dr using scandium triflate or iron trichloride as catalyst. The use of thymineor fluorouracilsubstituted cyclobutanes gave direct access to six-membered ring nucleoside analogues. Finally, the [4+2]-annulation between aminocyclobutanes and enol ethers led to the corresponding cyclohexylamines. Six-membered nitrogen-substituted carboand heterocycles are among the most frequently encountered scaffolds in natural and synthetic bioactive compounds (Figure 1). A cyclohexylamine or a tetrahydropyranylamine ring for example can be found in the core of the natural alkaloids strychnine (1) and staurosporine (2) respectively. The synthetic antiviral drug Tamiflu (3) is constituted by a cyclohexenyldiamine core. Synthetic methods giving access to these important scaffolds with high efficiency and broad scope are desirable to accelerate the discovery of new bioactive compounds. Whereas the Diels-Alder reaction has emerged as a powerful method to synthesize cyclohexenylamines and dihydropyranylamines, 1 there is currently a lack of transformations giving straightforward access to saturated ring systems with high convergence. Figure 1. Bioactive compounds containing nitrogensubstituted six-membered rings. The use of annulation reactions of donor-acceptor substituted strained rings constitutes a valuable alternative for the synthesis of saturated carboor heterocycles. In the case of sixmembered rings, the [4+2]-annulation between donor-acceptor cyclobutanes and olefins or carbonyl compounds appears particularly attractive (Scheme 1, A). Nevertheless, the chemistry of donor-acceptor cyclobutanes has been much less developed than for the corresponding cyclopropanes. 2 It is only very recently that more general catalytic methods have been developed in the groups of Johnson, Christie and Pritchard, and Pagenkopf in particular (Scheme 1, B). 3 However, these works focused on the use of oxygen and carbon as electron-donating groups, and the scope of substituents on the cyclobutanes was often limited. In the case of nitrogen as donor, an important pioneering result has been reported by Saigo and co-workers in 1991. 4 Unfortunately, the precious nitrogen functionality could not be conserved in the final product, as hydrolysis occurred upon work-up. Scheme 1. [4+2]-Annulations for the Synthesis of SixMembered Rings. Recognizing the underexploited potential of nitrogensubstituted strained rings for the synthesis of bioactive compounds, 5 our group has initially focused on the discovery of new types of donor-acceptor systems which could be broadly applied in annulation reactions. In particular, we reported that imido-substituted cyclopropane dicarboxylates can be used in [3+2]-annulations with both enol ethers and carbonyl compounds under mild catalytic conditions. 6 In 2013, we reported a new method to access the corresponding imido substituted cyclobutane dicarboxylates. 7 A single example of [4+2]annulation between an aminocyclobutane and an enol ether was also described in this work. Herein, we report the first Lewis acid-catalyzed [4+2]-annulation reaction between donor-acceptor aminocyclobutanes and carbonyl compounds and a further extended scope of the reaction with enol ethers (Scheme 1, C). In the case of aldehydes, the reaction was also successful for multi-substituted aminocyclobutanes, leading to tetrahydrofurylamines bearing up to three distinct stereocenters. We stared our investigations by examining the [4+2]annulation of aminocyclobutane 4a, 7 and benzaldehyde 5a (Table 1). The reaction proceeded with 20 mol % FeCl3•Al2O3 as catalyst, which had been used in the corresponding reaction with aminocyclopropanes, 6c but the product was obtained only with low diastereoselectivity as part of a complex mixture (Table 1, entry 1). A higher diastereoselectivity (6:1) was observed with tin tetrachloride, previously used for the single example reported of [4+2]-annulation of enol ethers and aminocyclobutanes (Table 1, entry 2). 7 However, a complex mixture was also observed in this case. As the use of titanium tetrachloride did not lead to any improvement (Table 1, entry 3), we then turned to well-established metal triflates as catalysts. Whereas no reaction was observed with ytterbium triflate (Table 1, entry 4) and a complex mixture was obtained with hafnium triflate (Table 1, entry 5), both indium and scandium triflates led to complete conversion without the formation of side products (Table 1, entries 6 and 7). A better diastereoselectivity (13:1 vs 9:1) was observed in the case of scandium triflate. Table 1. Optimization of the [4+2]-Annulation. entry 4, R catalyst time conversion dr 1 4a, H FeCl3•Al2O3 40 min >95% c 2:1 2 4a, H SnCl4 2.5 h >95% c 6:1 3 4a, H TiCl4 2.5 h 88% c 5:1 4 4a, H Yb(OTf)3 2.5 h <5% d 5 4a, H Hf(OTf)4 2.5 h >95% c 9:1 6 4a, H In(OTf)3 2.5 h >95% 9:1 7 4a, H Sc(OTf)3 2.5 h >95% 13:1 8 4b, Me Sc(OTf)3 24 h <5% d 9 4b, Me FeCl3•Al2O3 2.5 h 57% 1.5:1 10 4b, Me FeCl3•Al2O3 f 5 h >95% 5:1 Reaction conditions: 0.05 mmol 4, 0.075 mmol 5a, 20 mol % catalyst in 1.5 mL CH2Cl2 at rt. Conversion estimated by the ratio of product 6 to cyclobutane 4 on the H NMR of the crude mixture. Determined on the H NMR of the crude mixture. Complex mixture of products was observed by H NMR. No product observed. Temperature increased from rt to 40 °C after 7 h. 100 mol % catalyst loading. When the methyl-substituted aminocyclobutane 4b was examined, the reactivity dropped significantly and no conversion was observed with scandium triflate (Table 1, entry 8). However, product 6ba could be obtained with 57% conversion and 1.5:1 dr when using FeCl3•Al2O3 (Table 1, entry 9). Finally, increasing the amount of this cheap and non-toxic catalyst to 100 mol % allowed us to reach full conversion and 5:1 diastereoselectivity (Table 1, entry 10). 8 With optimized conditions in hand, we first examined the scope of the reaction of unsubstituted donor acceptor aminocyclobutanes (Table 2). On a preparative scale, tetrahydropyranyl amine 6aa could be isolated in 92% yield and 16:1 dr in favor of the cis diastereoisomer (Table 2, entry 1). 9 Electronwithdrawing and donating groups in para position of the aromatic ring were well tolerated (Table 2, entries 2 and 3), although a lower diastereoselectivity was observed in case of the methoxy substituent (Table 2, entry 3). Product 6ad bearing an ortho-methoxy substituent was obtained in 91% yield and 9:1 dr (Table 2, entry 4). The [4+2]-annulation with cinnamaldehyde 5e proceeded in nearly quantitative yield, but with only 2:1 dr. Under these reaction conditions, only low yields (<30%) and diastereoselectivities were obtained when using aliphatic aldehydes or ketones as partners (results not shown). We then turned to the more challenging use of substituted aminocyclobutanes. With cyclobutane 4b, the desired tetrahydropyranyl amine 6ba was obtained in 64% yield as a 5:1 mixture of only two diastereoisomers at the benzylic center (Table 2, entry 6). The main product obtained corresponded to the isomer with all substituents in equatorial position, which is probably the most stable. The reaction was also successful with para-chloro benzaldehyde as partner (Table 2, entry 7). When using phenyl-substituted cyclobutane 4c, the reaction became slower. Nevertheless, the desired product 6ca could still be obtained in 53% yield (Table 2, entry 8). Finally, the annulation of cyclobutane 4d bearing a substituent in 3 position relative to the phthalimide was examined (Table 2, entries 9-10). This class of cyclobutanes can only be obtained as a mixture of diastereoisomers using our previously reported [2+2]-cycloaddition method. 7 At room temperature, a significant amount of retro [2+2]-cycloaddition was observed, but this side reaction could be suppressed at 0 °C. The desired products could then be obtained in 64-81% yield as a 3:1 mixture of diastereoisomers starting from a 1:1.1 mixture of cyclobutanes. The annulation reaction is therefore not stereospecific. Interestingly, we observed that the cis cyclobutane reacted faster than the trans isomer in the [4+2]-annulation. This difference in rate could be used to do a resolution of the difficult to separate isomers of aminocyclobutane 4d: with the less reactive indium triflate catalyst, the trans isomer of aminocyclobutane 1b could be recovered in quantitative yield and 14:1 dr (Scheme 2). Scheme 2. Diastereospecific [4+2]-Annulation. During the investigation of the scope of the [4+2]annulation reaction of unsubstituted aminocyclobutanes, a relatively high catalyst loading of scandium triflate (20 mol %) has been used for practical reasons. Nevertheless, the catalyst loading could be decreased to 5 mol % when the reaction was run on a 1 mmol scale and product 3aa was obtained in 81% yield and 13:1 dr (Scheme 3). Scheme 3. [4+2]-Annulation at 1 mmol Scale. Table 2. Scope of the [4+2]-Annulation. entry aminocyclobutane product yield dr