A Torquoselective 6π Electrocyclization Approach to Reserpine Alkaloids

Abstract

A highly torquoselective thermal triene 6π electrocyclization controls the relative stereochemistry between the C3 and C18 stereocenters of the dodecahydroindolo[2,3-a]benzo[g]quinolizine skeleton of reserpine-type alkaloids. Employing a tandem cross-coupling/electrocyclization protocol allowed us to form the requisite triene and ensure its subsequent cyclization. A novel low-temperature dibromoketene acetal Claisen rearrangement established the requisite exocyclic dienylbromide precursor for the palladium-catalyzed cross-coupling reaction.

The reserpine-type alkaloids, possessing the dodecahydroindolo[2,3-a]benzo[g]quinolizine skeleton, are the most structurally complex of the indole alkaloids (Figure 1). These compounds exhibit significant CNS activity and are key components of herbal medicines that have been used for centuries. Sixteen naturally occurring reserpine alkaloids have been isolated; among them, several have been prepared, along with a variety of semi-synthetic derivatives.^1,2 The most common synthetic strategies toward reserpine alkaloids are “bottom-up” approaches, involving preparation of the fused D/E ring system with subsequent installation of the C3 stereocenter in the process of C2–C3 bond formation.^3a–h “Top-down” approaches, involving either C3–C14 or C14–C15 bond formation with concomitant generation of the C3 stereocenter and establishment of the C15 and C20 stereocenters, have been employed less frequently.^3i–l

Figure 1.

Reserpine-type alkaloids.

As an alternative to these strategies, we were intrigued by the idea of the C3 center exerting remote stereocontrol across the fused D/E ring system during the formation of the C18 stereocenter. Because C18 is the most remote stereocenter from C3 on the benzo[g]quinolizine scaffold, we were aware of the potential difficulty of this 1,6-relay of stereochemical information and sought to identify an appropriate reaction manifold for such a strategy.

The thermal 6π electrocyclization of all-carbon–containing trienes is an underutilized, yet highly powerful, reaction with the capacity to generate two new stereocenters. Furthermore, if the starting triene possesses a stereogenic center, the newly generated stereocenters could be formed in a highly diastereoselective manner. Nevertheless, the electrocyclization of all-carbon trienes has received relatively little attention in the synthesis of small molecules when compared with other pericyclic reactions, such as the Diels–Alder and Claisen reactions.⁴

As a part of our growing interest in novel 6π electrocyclizations⁵ and a research program in nucleophilic phosphine catalysis for the synthesis of heterocycles and carbocycles,⁶ we became interested in the potential of these transformations to meet the goals of our synthetic strategy for remote 1,6-stereoinduction. We envisioned that the presence of the C3 stereocenter would allow the C18 stereocenter to be formed in a stereocontrolled manner through a torquoselective 6π electrocyclization (Scheme 1). This unique approach stands in stark contrast to previous synthetic strategies toward the synthesis of reserpine alkaloids and hinges upon a seldom-explored remote 1,6-stereoselective 6π electrocyclization.^4d

Scheme 1.

Remote Stereochemical Induction via Thermal Torquoselective 6π Electrocyclization

Although the literature does not offer an extensive discussion of remote stereocontrol in the context of 6π electrocyclization, we reasoned that a suitably constrained scaffold, such as that of the triene 1, should allow the C3 stereocenter to influence the stereoselectivity of this transformation (Scheme 1).^4b While optimistic for the possibility of stereochemical control, we lacked any established models to predict the diastereoisomer favored upon electrocyclization. If successful, however, this synthetic strategy would provide the pentacycle 2 possessing the requisite ABCDE skeleton in a stereodefined manner. Upon removal of R¹, the exposed hydroxyl functionality could be inverted readily, if necessary, and the resultant α,β-unsaturated ester contained in the E ring could act as a handle for elaboration to (±)-reserpine and related alkaloids.

We envisioned efficient access to 1 through cross-coupling of a suitable pseudo-metal with the vinyl bromide 3, which could be derived from the appropriately functionalized ester 4 (Scheme 2). We planned to generate the α,α-dihaloester 4 from the allylic alcohol 5 through [3,3]-sigmatropic rearrangement of the corresponding dihaloketene acetal. Based upon our previous experience with heterocycle formation through phosphine catalysis, we expected a relatively straightforward elaboration of the allylic alcohol 5 from the [4 + 2] annulation product 6.^6b

Scheme 2.

Reterosynthesis of the Key Triene 1

The [4 + 2] annulation between the imine 7 and the butadienoate 8 in the presence of catalytic PBu₃ proceeded uneventfully (Scheme 3);⁷ we isolated the tetrahydropyridine 9 as a yellow crystalline solid after removal of the Boc group in toluene under reflux in the presence of silica gel. Next, in a one-pot procedure, we prepared the tryptophol 10 through treatment of the indole 9 with oxalyl chloride at ambient temperature and subsequent BH₃·SMe₂-mediated reduction of the intermediate glyoxylic acid chloride. The use of bases commonly employed under classical Fukuyama conditions for nosyl deprotection led to partial decomposition of 10 and low yields. We found, however, that addition of potassium thiophenoxide (PhSK) as a preformed salt led to clean removal of the o-nosyl protecting group in excellent yield. Formation of ring C in 11 was facilitated by the intramolecular alkylation of the resulting secondary amine in the presence of PPh₃ and I₂. Protection of the indole nitrogen atom with Boc₂O and subsequent DIBAL-mediated reduction of the ester gave the key allylic alcohol 5 in 88% yield. Overall, this route provided efficient access to 5 in eight steps from the imine 7 on a multigram scale.

Scheme 3.

Synthesis of the Cross-Coupling Partner 3 Through Phosphine-Catalyzed [4 + 2] Annulation and Dibromoketene Acetal Claisen Rearrangement.

We prepared a variety of dimethylorthoacetates containing α-halogen atoms with the intention of exploiting a Johnson–Claisen rearrangement for the preparation of the α,α-dibromoester 4.⁸ Despite our best efforts, we were unable to facilitate the desired [3,3]-sigmatropic rearrangement employing these substituted orthoesters.⁹ Fortunately, access to 4 could be achieved through [3,3]-sigmatropic rearrangement of the dibromoketene acetal generated through dehydrobromination of the acetal 19.¹⁰ Remarkably, rearrangement occurred spontaneously after elimination of HBr at −78 °C, providing the desired α,α-dibromoester 4 in 88% yield. To the best of our knowledge, this example is the first [3,3]-sigmatropic rearrangement leading directly to highly functionalized α,α-dibromoesters.¹¹ A number of amine and alkoxide bases facilitated β-elimination from the α,α-dibromoester 4 to give the vinyl bromides 3 in approximately a 2:1 dr, but with very low mass recoveries (<40%). Ultimately, we found that potassium hexafluoroisopropoxide, prepared in situ through the addition of 1 M ^tBuOK to a solution of excess hexafluoroisopropanol (HFIP) and [18]crown-6 at 0 °C, cleanly afforded the vinyl bromides E-3 and Z-3 as a separable 2:1 mixture in 85% yield.¹²

For the formation of the requisite triene 1 for our planned 6π electrocyclization, we initially speculated that only Z-3 would be suitable for stereospecific cross-coupling reaction to form the E-triene 1. We found, however, that we could employ both isomers under an optimized sequence of Suzuki cross-coupling, photochemical isomerization of the undesired triene, and electrocyclization at 80 °C in benzene.¹³ Following this protocol, we obtained the desired pentacycle in 54% yield as a single diastereoisomer, starting from a mixture of the two vinyl bromides E/Z-3. X-ray crystallographic analysis revealed the relative configuration of 2a. We suspect that the reaction selectivity arises from an allylic interaction between the methyl ester group and the adjacent D-ring methylene unit in the transition state.^4a,e Such interactions can be sufficiently large to provide high levels of stereocontrol; studies are currently ongoing in our laboratories to fully assess the controlling element in this reaction.

Although we could functionalize 2a through conjugate addition to the α,β-unsaturated ester, the yields and diastereoselectivities of these reactions were generally very poor, providing all four possible diastereoisomers. We reasoned that deprotection of the benzyl group would allow a hydroxyl-directed reaction to provide better control over the stereoselectivity. Unfortunately, our efforts at removing this protecting group under a variety of conditions gave undesired products, including those from reduction of the α,β-unsaturated ester and elimination of the benzyloxy substituent resulting in aromatization of the E ring. Considering the difficulty we encountered installing the requisite functionality onto 2a, we sought to install all of the E-ring alkoxy groups of reserpine in a single step through our tandem cross-coupling/ 6π electrocyclization. We envisioned that this approach would require a 1,2-dialkoxyvinyl pseudo-metal to undergo the cross-coupling reaction. From an exploration of the literature, we identified the dioxole 21 as a potentially suitable candidate (Scheme 5).¹⁴

Scheme 5.

Torquoselective Negishi/6π Electrocyclization

Although the conversion of 21 into a variety of vinyl boronic esters (including pinacol and diisopropyl) proceeded well, their instability during purification limited our options for optimization of the Suzuki/6π electrocyclization cascade. Instead, following treatment of 21 with nBuLi and exposure to anhydrous ZnCl₂, we could facilitate a Negishi/6π electrocyclization cascade upon heating E/Z-3 under reflux in THF in the presence of Pd(PPh₃)₄. To the best of our knowledge, this transformation marks the first use of the dioxole 21 as a highly oxidized partner in cross-coupling chemistry. The optimized conditions provided the acetonide 22 as the sole product in 55% yield and 20:1 dr. Surprisingly, photoisomerization was unnecessary in this case: we obtained the cyclization product in similar yield directly from the cross-coupling reaction.¹⁵

Deprotection of the acetonide with InCl₃ and water in MeCN provided an oxidatively sensitive and unstable hydroxy ketoester.¹⁶ Reduction of the ketone under H₂ in the presence of PtO₂ provided the trans diol 23 in 70% yield as a single diastereoisomer. Advancing the diol 23 to (±)-reserpine would require selective benzoylation of the α-hydroxyl group, deprotection of the Boc group, reduction (from the α face) of the olefinic bond at the D/E ring junction,¹⁸ inversion of the relative configuration of the vicinal diol through a benzoate walk,¹⁹ and O-methylation.

In conclusion, we have obtained highly functionalized pentacyclic intermediates en route to the preparation of reserpine-type alkaloids. Our synthetic sequence relies on a phosphine-catalyzed [4 + 2] annulation, a novel dibromoketene acetal Claisen rearrangement, and a cross-coupling/ torquoselective 6π electrocyclization cascade. In particular, we have demonstrated that the torquoselective triene 6π electrocyclization is an extremely efficient method for remote 1,6-transfer of chirality from C3 to C18 in the synthesis of reserpine alkaloids. We are currently assessing the stereochemistry-controlling elements in these reactions and their potential for further synthetic applications.

Scheme 4.

Torquoselective Suzuki/6π Electrocyclization