Stereoselective synthesis of isoquinuclidines through an aza-[4 + 2] cycloaddition of chiral cyclic 2-amidodienes.

A highly stereoselective aza-[4 + 2] cycloaddition of chiral cyclic 2-amidodienes with N-sulfonyl aldimines is described. While this Lewis acid promoted heterocycloaddition provides an efficient strategy for constructing optically enriched isoquinuclidines, it is mechanistically intriguing. The cycloaddition favored the endo-II pathway in the absence of a viable bidentate coordination. This represents an unexpected switch from the anticipated endo-I selectivity obtained in the all-carbon cycloaddition.

The isoquinuclidine [or n class="Chemical">2-azabicyclo[2.2.2]octane] core is a prevalent motif among biologically active natural products.[1] Catharanthine, an iboga-alkaloid, is a biosynthetic precursor of dimeric vinca alkaloids such as vinblastin and vincristine, which are being used as antitumor agents for treatment of a number of human cancers (Figure 1).[2] Moreover, Daphniphyllum alkaloids such as caldaphnidine D have shown a variety of pharmacological activities such as cytotoxicity and antioxidative activity.[3] Xestocyclamine A, a polycyclic alkaloid isolated from a marine sponge, exhibits potent inhibitory property of PKCβ (PKC = protein kinase C).[4] Given such potential significance in cancer therapeutic development, designing efficient methods to access the isoquinuclidine core represents an important endeavor not only for constructing these natural products but also advancing biological studies that can lead to possible drug discovery.[2−4]

Figure 1

Bioactive natural products with isoquinuclidine core.

Recently, we[5] reported a highly regio- and stereoselective Diels–Alder cycloaddition featuring de novo chiral cyclic 2-amidodienes 1 (Scheme 1) derived from n class="Chemical">allenamides.[6] This cycloaddition provides a facial entry to optically enriched [2.2.2]bicyclic manifolds [1→2 in Scheme 1]. Despite a rich history of amino and amido dienes,[7,8] cyclic 2-amidodienes are rare,[9] and there certainly is a lack of overall systematic exploration of their reactivities.[10] Given the significance of isoquinuclidine, we envisioned that an aza-[4 + 2] cycloaddition[11,12] employing imines as heterodienophiles could prove to be an efficient strategy for stereoselective constructions of optically enriched isoquinuclidines that have mostly been synthesized through [4 + 2] cycloadditions of 1,2-dihyropyridines.[13−17] We wish to report here our success in developing a highly stereoselective aza-[4 + 2] cycloaddition of cyclic 2-amidodienes with N-sulfonyl aldimines and finding of an unexpected switch in the selectivity.

Scheme 1

Cyclic 2-Amidodienes in [4 + 2] Cycloadditions

We commenced our investigation by screening Lewis acid condition class="Chemical">ns using cyclic 2-amidodiene 1(18) with N-Ts aldimine 4 serving as the dienophile. As shown in Table 1, although aluminum- and boron-based Lewis acids were less effective (entries 1–3) and TMSOTf is only marginal (entry 4), the best promoter for this reaction appeared to be SnCl4, leading to the desired cycloadduct 5(19) in 60% yield (entry 5). There are three isomers found in this reaction with the major isomer being the endo-II product (5b) and the two minor isomers being exo-I and exo-II products 5c and 5d, respectively. The actual stereochemical assignments were made using another cycloadduct. While TiCl4 afforded comparable diastereoselectivity but gave a lower yield (entry 6), MgBr2 and Zn(OTf)2 were ineffective in promoting the cycloaddition (entries 7 and 8). Neither were rare earth metal triflates such as Yb(OTf)3 and La(OTf)3 useful in this context (entries 9 and 10), although they have been successfully utilized in aza-[4 + 2] reactions.[20]

Table 1

Identifying of a Suitable Lewis Acid

entry	LA	solvent	temp (°C)	time (h)	yielda (%)	dr ratiob
1	EtAlCl2	CH2Cl2	–78	12	14	N.D.c
2	AlCl3	CH2Cl2	–78	12	38	60:(20:20)
3	BF3–OEt2	CH2Cl2	–78	12	33	N.D.
4	TMSOTf	CH2Cl2	–78	12	48	34:(33:33)
5	SnCl4	CH2Cl2	–78	12	60	68:(16:16)
6	TiCl4	CH2Cl2	–78	12	52	68:(16:16)
7	MgBr2	THF	rt	14	no rxnd
8	Zn(OTf)2	CH2Cl2	rt	18	no rxn
9	Yb(OTf)3	CH2Cl2	rt	15	no rxn
10	La(OTf)3	CH2Cl2	rt	15	no rxn

Isolated yields; in all cases, 1.0 equiv of Lewis acid was used.

Ratios denote endo:exo with exo-I:exo-II ratios shown in parentheses. All ratios determined using 1H and/or 13C NMR.

N.D. = not determined.

Complete recovery of the starting diene 1.

Complete recovery of the starting n class="Chemical">diene 1.

Having identified a suitable Lewis acid, a series of n class="Chemical">N-Ts-aldimines were examined as summarized in Table 2: (1) The minor isomers of cycloadducts 9 were cleanly isolated as a mixture and could be concisely assigned as a pair of exo-cycloadducts, more specifically 9c and 9d, after its hydrolysis.[21] (2) The tert-butyl-substituted imine was unfortunately not tolerated likely due to the steric hindrance of the t-Bu group, and aryl aldimines appear to be inferior substrates in terms of selectivity (see 13). (3) With respect to nitrogen substitutions, methane sulfonyl and Ns groups were not ideal (see 14 and 15, respectively). In contrast, Ts- and PMP-sulfonyl groups were very effective (see 16 and 17). (4) Lastly and most importantly, the single-crystal X-ray structure the major isomer of cycloadducts 12 reveals that it is 12b, therefore unambiguously confirming the endo-II selectivity (Figure 2). Cycloadditions of other chiral cyclic 2-amidodienes with N-sulfonyl aliphatic aldimines were also examined (Table 3). Depending on the imine, in general, good selectivities as well as yields were obtained including the usage of Sibi’s auxiliary (see diene 8).[22] These results demonstrate that this cycloaddition can be useful for constructing optically enriched isoquinuclidines.

Table 2

Examining the Scope of aza-Dienophiles

Reaction conditions followed those described for entry 5 in Table 1. All are isolated yields.

Ratios denote endo:exo [a/b:c/d] with the exo-I:exo-II ratio [c:d] shown in parentheses. They are determined using 1H and/or 13C NMR.

PMP = p-methoxyphenyl.

Stereochemistry of each isomer was not unambiguously assigned.

This ratio implies a single isomer is observed here.

Only one exo-cycloadduct was seen but unassigned whether it is exo-I or exo-II.

Decomposition of starting diene 1 with no observable desired cycloadduct.

Figure 2

X-ray structure of endo-II cycloadduct 12b.

Table 3

Cycloadditions of Other Cyclic 2-Amidodienes

Reaction conditions followed those described for entry 5 in Table 1. All are isolated yields.

Ratios denote endo:exo [a/b:c/d] with the exo-I:exo-II ratio [c:d] shown in parentheses. They are determined using 1H and/or 13C NMR.

PMP = p-methoxyphenyl.

This ratio implies a single isomer is observed here.

Only one exo-cycloadduct was seen but unassigned whether it is exo-I or exo-II.

Mechanistically, we became intrigued because the stereochemical outcome is opposite from what we had anticipated. As shown in Scheme 2, there are two possible low energy rotameric conformations for these n class="Chemical">cyclic 2-amidodienes, syn and anti, which can be defined by the position of the carbamate carbonyl relative to the blue-colored olefin with respect to the red C–N bond (see wavy arrows in Scheme 2). In both rotamers, the nitrogen lone pair is coplanar and fully delocalized into the diene motif. In our previous all-carbon cycloaddition,[5] the observed endo-I selectivity suggests that the syn rotamer is operative because its substituent on the chiral auxiliary shields the top endo-II face. Although Spartan B3LYP/6-31G* calculations reveal that the anti rotamer is more favored, the energetic difference is sufficiently small (i.e., ΔE = −0.28 kcal mol–1 for diene 8) that it should not preclude the possibility of the syn rotamer being the more reactive conformer.

Scheme 2

Two Rotameric Conformations of the Diene

Consequently, we expected endo-I selectivity also for the current n class="Chemical">aza-[4 + 2] cycloadditions, but instead, these are endo-II selective processes, thereby implying a facial preference switch with the anti rotamer being operative. Furthermore, as shown in Scheme 3, Spartan B3LYP/6-31G* calculations demonstrate that TS-endo-II is actually lower in energy than TS-endo-I (1.41 kcal mol–1 for diene 1).[23] This energetic different in the transition states agrees quite well with our observed stereochemical outcome. To reconcile these differences, we suspect that the previous cycloaddition involved a highly organized transition-state through a long distance chelation with the tin metal coordinating to both carbonyl oxygen atoms (right-hand side in Scheme 3). This bidentate coordination is likely only feasible with TS-endo-I (d = 5.1 Å) given the much longer distance between the same two oxygen atoms in TS-endo-II (d = 6.0 Å, not shown).

Scheme 3

Unexpected Switch in the Facial Preference

To provide support for this proposal, we examined the all-carbon cycloaddition in the absence of any n class="Chemical">Lewis acids. We had previously avoided this condition because these cyclic 2-amidodienes tend to undergo isomerization via a 1,5-H shift at high temperatures over an extended period of time, leading to the more stable 1-amidodienes.[18] With some trepidation, we carefully carried out cycloadditions of diene 6 with two enones at high temperature as shown in Scheme 4. Although yields were not particularly high, we managed to isolate the desired cycloadducts 34a/b and 35a/b with complete reversal of selectivity from those obtained under Lewis acid conditions. These results provide rather strong support for our assertion that the cycloaddition prefers the energetically more favored TS-endo-II (0.28 kcal mol–1 for diene 6)[23] through the anti rotamer and that, in the presence of a viable bidentate coordination,[24] this selectivity can be altered and/or completely reversed.

Scheme 4

Reversal to the Endo-II Selectivity

In summary, we have developed a highly stereoselective aza-[4 + 2] cycloaddition of n class="Chemical">2-amidodienes and N-sulfonyl aldimines. This Lewis acid promoted heterocycloaddition provides an efficient strategy for constructing optically enriched isoquinuclidines. Mechanistically, we uncovered an unexpected switch from the anticipated endo-I selectivity. In the absence of a chelation-dictated transition state, these aza-[4 + 2] cycloadditions favored the endo-II pathway through the anti rotamer. Further mechanistic study through calculation and its applications in total synthesis of relevant alkaloids are currently underway.