Synthetic Access to Bent Polycycles by Cation-π Cyclization

Abstract

The presence of an ether oxygen within a chain undergoing cation–polyene cyclization has a profound influence on the stereochemistry of this important construction, apparently due to nucleophilic participation of oxygen in the cyclization process and formation of an oxonium intermediate, leading to bent fused ring systems.

The carbocation-initiated polycyclization of appropriate polyunsaturated substrates is one of the most powerful molecular constructions in synthetic chemistry. It is also an extremely important biosynthetic process, for instance, in the one-step tetracyclization of (S)-2,3-oxidosqualene¹ to a sterol (Figure 1, 1 → 2). The carbocation–π-cyclization process is also the dominating pathway for the biosynthesis of cyclic terpenoids. Cationic cyclization has served well for the chemical syntheses of many complex terpenoids, e.g., pentacyclosqualene^2a (an early example), dammarenediol,^2b lanostenol,^2c onocerin,^2d serratenediol,^2e β-amyrins, and lupeol.2f,g However, there is still a large gap in efficiency between biosynthetic polycyclizations and present chemical syntheses. Unlike the enzymic cyclizations which proceed with an efficiency of ca. 99% per ring formed, the chemical cyclizations are, at best, only 70–80% efficient per ring formed. This difference arises partly because the cyclization substrate is held by the enzyme in a prefolded conformational arrangement which selects the proper π-face of a double bond for attack by the propagating carbocation, minimizing the decrease in the entropy of activation of enzymic cyclization.

Figure 1.

Biosynthetic conversion of (S)-2,3-oxidosqualene (1) to protosteryl cation (2).

Although there is no known way to mimic this template-guided cyclization, the entropic problem may be diminished by conducting cyclization at the lowest possible temperature (to minimize the (positive) TΔS‡ component of the free energy of activation (ΔG‡)) in CH₂Cl₂ with MeAlCl₂ as catalyst. Functional groups more basic than epoxide are not tolerated. The studies described herein were undertaken to gain a better understanding of the relationship between substrate structure and π-facial selectivity and to extend the scope of cyclization. One motivation for this study was the observation that the common stereochemical pathway for cyclization, exemplified by the conversion of 3 to the A/B/C/trans–anti–trans product 4 (Figure 2),³ is not universal since cases such as 5 → 6 are known (Figure 3)⁴ which would seem to involve cations 7 and 8 and a different π-facial selectivity at the second olefinic linkage. This difference in π-facial selectivity at the double bond involved in closure of the second ring is of great interest since it is clearly a branch point in the biosynthetic cyclizations which lead to sterols or plant triterpenes.⁵

Figure 2.

Trans–anti–trans cyclization of 3 to 4 producing 9α-H stereochemistry.

Figure 3.

Cyclization of 5 to 6 through cation 8, which possesses 9β-H stereochemistry.

Our initial research of the relationship between the structure of the substrate and the stereochemistry of closure of the second ring in epoxide-initiated cation–olefin cyclizations was conducted with simple epoxyfarnesol derivatives. In the discussion which follows, the two modes of bicyclization will be referred to as the 9α-H or 9β-H pathway, as structurally indicated with formulas 9–11 (Figure 4). This phase of our work was carried out with CH₂Cl₂ as solvent, and RAlCl₂ or R₂AlCl (3 equiv, e.g., EtAlCl₂) as catalytic Lewis acid at −78 °C, conditions which generally are most favorable for epoxide-initiated cationic polycyclization reactions.

Figure 4.

Two modes of cyclization for the formation of ring B: the 9α-H or 9β-H pathways.

When the benzoate of 10,11-epoxyfarnesol was examined under standard conditions for cyclization only monocyclic reaction products were obtained. The most obvious explanation for this result is that the Lewis acid coordinates more strongly with the benzoate carbonyl, and essentially completely (with 3 equiv of EtAlCl₂), and that coordination removes electron density from the terminating double bond which prevents formation of the second ring. Because of this result, we next examined the two bulky silyl ethers of 10,11-epoxyfarnesol 12a and 12b (Scheme 1), the assumption being that Lewis acid complexation with the sterically hindered silyloxy oxygen would not be a complication. Indeed, in each case bicyclization was the principal reaction pathway (70–85% isolated yield of 13 and 14 after desilylation and column chromatography on silica gel). For each of these substrates there was a near balance between 9α-H and 9β-H pathways for cyclization, as indicated in Scheme 1. The 9α-H and 9β-H products each consisted of a mixture of one endo- and one exocyclic olefinic species.⁶

Scheme 1.

Cyclization of Silyl Ethers 12a and 12b To Produce Equimolar Mixtures of 9α-H and 9β-H Products^a

^a Conditions: (1) EtAlCl₂ (3 equiv added slowly), CH₂Cl₂ (0.01 M), −78 °C, 3 h; (2) TBAF (1.5 equiv), THF (0.5 M), 25 °C, 6 h.

The close similarity of ratios for the 9α-H and 9β-H cyclization pathways came as a surprise, given the considerable number of reported examples in which the 9α-H product is strongly preferred. One possible reason for the exceptional behavior of the substrates 12a and 12b becomes apparent when the reported sequence 5 → 6 (presumably via 7 and 8) is recalled. If it were generally true that the 9β-H pathway is favored for reactions that proceed to bicyclic 6/6-fused product via a bicyclic 6/5-fused intermediate, it is logical to explain the formation of the 9β-H product 14 via the oxonium intermediate 15, which then converts to 16 (Figure 5). To test whether the 9β-H pathway is made more favorable if the incipient ring B is generated via a 5-membered-like precursor, cyclization of the corresponding allylsilane 17 was examined (Scheme 2).

Figure 5.

Intermediacy of an oxygen-stabilized secondary cation 15 may explain the formation of 9β-H products.

Scheme 2.

Cyclization of Silane 17 Giving Rise To Hydrindane 18, Which Possesses 9β-H sStereochemistry^a

^a Conditions: (1) EtAlCl₂ (3 equiv added slowly), CH₂Cl₂ (0.01 M), −78 °C, 3 h; (2) TBAF (1.5 equiv), THF (0.5 M), 25 °C, 6 h; (3) DCC (1.1 equiv), 3,5-dibromobenzoic acid (22) (1.1 equiv), DMAP (0.05 equiv), CH₂Cl₂ (0.1 M), 25 °C, 1 h.

The substrate 17 was synthesized from 9,10-epoxyfarnesyl benzoate by coupling with the reagent derived from dimethylphenylsilyllithium and copper(I) cyanide. Cyclization of 17 under the conditions developed for 12a and 12b and subsequent desilylation with TBAF afforded an equimolar mixture of two products which were chromatographically separated after conversion to their 3,5-dibromobenzoates 20 and 21 (chromatography on AgNO₃-impregnated silica gel). The stereochemistry of the decalin product 19 corresponds to the usual 9α-H cyclization (confirmed by X-ray crystallographic analysis of 21, mp 144–146 °C), whereas that of the hydrindane 18 corresponds to the π-facial selection of the 9β-H geometry (confirmed by NOE studies of 20). This result for 18 is consistent with the proposal that cyclization via a 5-membered B-ring structure can involve the opposite π-face selectivity to that which is generally found for 6-membered B-ring formation.

Further studies uncovered a surprising effect of π-bond basicity on the stereochemistry of B-ring formation. The more π-basic tri-n-butylstannyl analogue 23 of the allylic silane 17 underwent cyclization under the standard conditions to produce a 1:1 mixture of the two diastereomeric bicyclic structures 24a and 24b (Scheme 3). This complete lack of π-face selectivity in the formation of ring B may well be a consequence of a much earlier transition state for ring B closure, i.e., a much longer incipient C–C bond in the transition state.

Scheme 3.

Cyclization of Allylstannane 23 To Form a Mixture of Diastereomers 24a and 24b

Behavior similar to that just described for 23 was observed in the cyclization of epoxide 25 which leads to a 1:1 mixture of 26a and 26b (Scheme 4). In this case also, there is no π-face selectivity in the formation of ring B, probably for the same reason: a longer C–C bond in the transition state for B-ring closure and a small energy difference between the diastereomeric transition states.⁷

Scheme 4.

Cyclization of Allylsilane 25 To Form a Mixture of Diastereomers 26a and 26b

It was found that 9,10-epoxyfarnesyl phenyl ether 27 underwent efficient conversion to the tetracyclic product 28a in a single step (Scheme 5). Evidently, the phenolic oxygen is sufficiently nonbasic to allow monocoordination of EtAlCl₂ to the more basic epoxide function, in contrast to the benzyl and cinnamyl ethers, which gave essentially no polycyclic products under identical reaction conditions. Of great interest is the fact that cyclization of 27 favored the 9β-H pathway in preference to the 9α-H path, since 28a predominated over the C9 diastereomer 28b by 2:1 (see the Supporting Information). The major tetracyclic product was isolated by chromatography after conversion to the 3,5-dibromobenzoate (51% overall yield from 27). The assignment of stereochemistry to 28a and 28b was made unambiguously by ¹H NMR measurements at 500 MHz and NOE correlations. It should be noted that decalin cation 29 must undergo a conformational change⁸ from initially formed boatlike conformer 29 to the chair form 30 before closure of ring C to produce tetracycle 28a.

Scheme 5.

Cyclization of Phenyl Ether 27 To Produce Predominantly 9β-H Product 28a

The contrast between the cyclization pathways 27 → 28a and 3 → 4 (Figure 6) provides another indication of the importance of the lone pairs of oxygen in influencing the balance between the 9α-H and 9β-H modes of reaction.

Figure 6.

Cyclization of 27 → 28a compared to the cyclization of 3 → 4.

Cyclization of the analogous 1-naphthyl ether 31 also proceeded cleanly and afforded the pentacycle 32 as the major product (Scheme 6). A slight increase in the diastereoselectivity of cyclization (9β-H/9α-H, 2.4:1) was observed relative to the cyclization of phenyl ether 27 (9β-H/9α-H, 2:1).

Scheme 6.

Cyclization of 31 To Produce Pentacycle 32

Additional oxygen substitution on the aryl moiety is also tolerated in the cationic polycyclization process and as such can provide useful reactive centers for elaboration to more complex structures (Scheme 7). The farnesyl ether derived from hydroquinone (33) underwent efficient cyclization to tetracycle 35 in good yield and in 2.3:1 excess over the 9α-H epimer (70% combined yield of diastereomers). Similarly, the phloroglucinol-derived bis-triisopropylsilyl (TIPS) ether 34 underwent cyclization to the corresponding tetracycle 36 with a 9β-H to 9α-H ratio of 3:1 (total yield 62%).

Scheme 7.

Oxygen Substitution on the Arene Is Tolerated under Cyclization Conditions

We also investigated the cyclization of the difarnesyl 1,5-naphthyl ether 37 (Scheme 8) which was synthesized in one step from 1,5-dihydroxynaphthalene and (S)-10,11-epoxyfarnesyl bromide. Treatment of 37 with EtAlCl₂ (6 equiv) in CH₂Cl₂ at −78 °C led to a mixture of diols, which were converted to the corresponding 3,5-dibromobenzoates and purified by chromatography to give octacycle 38 (31% from 37), the structure of which was confirmed by NMR and single-crystal X-ray analysis.⁸

Scheme 8.

Bidirectional Cyclization of 37 To Octacycle 38

The unique twisted, rigid octacyclic structure of 38, prepared in just a few steps from farnesol and commercially available 1,5-dihydroxynapthalene, would be difficult to make by other methods.⁹ It further demonstrates the power of cationic polycyclization methodology to produce a wide range of interesting structures.