Synthesis of Alkaloid (−)-205B via Stereoselective Reductive Cross-Coupling and Intramolecular [3+2] Annulation

Abstract

An asymmetric synthesis of alkaloid (−)-205B, a tricyclic member of the architecturally diverse family of natural products isolated from the skin of neotropical poison frogs, is described that proceeds through two recently developed stereoselective synthetic methods: (1) Ti-mediated allylic alcohol–imine reductive cross-coupling, and (2) intramolecular [3+2] cycloaddition of a glyoxylate-based homoallylic nitrone. The utility of this latter cycloaddition process for the assembly of the stereochemically dense piperidine core of 205B is noteworthy, as this method enables direct [3+2] annulation of an intermediate homoallylic (E)-nitrone via a pathway that is stereochemically unscathed by competitive [3,3] sigmatropic rearrangement processes. Overall, the synthesis is asymmetric, concise, and highly stereoselective – features which point to the potential future utility of these chemical methods in natural product synthesis and medicinal chemistry.

Due to the pioneering and focused efforts of John Daly at the National Institutes of Health, it has become clear that neotropical frogs are a rich source of architecturally unique alkaloids.¹ To date, over 800 heterocycles spanning more than 20 structural classes have been discovered from anuran skin, many of which are sequestered unchanged from dietary sources (ants, mites, beetles, and millipedes). Despite the rich structural diversity associated with this class, the majority of these natural products lack pharmacological characterization due, in part, to their low natural abundance. As a result, chemical synthesis stands as an enabling tool for future evaluation of this potentially important class of natural products. In this regard, the azatricyclododecene (−)- 205B, isolated from Dendrobated pumilio,² has risen as a target of interest due to its intriguing structure and the synthesis-driven discovery that the unnatural enantiomer blocks the α7 nicotinic acetylcholine receptor in a selective fashion (Figure 1).³ Since Toyooka’s pioneering synthesis of (+)-205B,⁴ Smith and Comins have independently reported syntheses of the natural antipode through chemical pathways that have secured the utility of anion relay chemistry and dihyropyridone-based functionalization processes to access the heterocyclic core of this target.⁵ Here, we report a conceptually unique synthesis of (−)-205B that proceeds by two stereoselective synthetic methods that have recently emerged from our laboratories: (1) Ti-mediated reductive crosscoupling between an aldehyde and an allylic alcohol (via the intermediacy of a TMS-imine),⁶ and (2) path-selective intramolecular [3+2] annulation of a glyoxylate-based homoallylic nitrone.⁶ The current studies document the first demonstrations of these reaction processes in natural product synthesis.

Figure 1.

Structurally diverse alkaloids from the skin of the neotropical poison-frog Dendrobates pumilio.

While related to the structure of coccinelline (Figure 2A), the stereochemically dense piperidine core of 205B presents a challenge not easily addressed with the elegant two-directional synthesis that Stevens demonstrated for the former⁷ – a strategy undoubtedly inspired by Robinson’s classic tropinone synthesis,⁸ and one that explored the stereochemical course of an intermolecular Mannich reaction on a six-membered ring-containing iminium ion (Figure 2B). We suspected that a related approach to 205B, while potentially elegant, would likely be substantially less effective for a number of reasons. As illustrated in Figure 2C, the primary amine required for 205B is not symmetric and can engage in iminium ion formation with either of the terminal acetals (paths A and B) – each path presenting different obstacles. Adding to the mounting uncertainty surrounding a coccinelline-inspired annulation en route to 205B, the likely multi-step nature of any modern asymmetric synthesis of the requisite chiral primary amine starting material dissuaded us from the pursuit of such an annulation process.

Figure 2.

Strategic considerations and retrosynthesis of (−)-205B.

Aiming to avoid these issues, we abandoned a strategy for annulation that targets the simultaneous generation of the C2a and C5a stereocenters, favoring a late stage annulation of the relatively unfunctionalized indolizidine onto a pre-existing piperidine core (1 → 205B; Figure 2D).⁹ This piperidine was inturn envisioned to derive from functional group manipulation of the 1-aza- 7-oxabicyclo[2.2.1]heptane 2, itself generated by a highly stereoselective [3+2] annulation of an intermediate homoallylic nitrone generated by condensation of 3 with butylglyoxylate (4). In this manner, the C5a–C6 bond was targeted as a key site for assembly of the densely functionalized piperidine core, rather than relying on C5-C5a bond construction via intermediacy of a six-membered ring iminium ion intermediate. Finally, the stereodefined acyclic intermediate 3 could derive from enantioselective Ti-mediated reductive cross-coupling of a TMS-imine (generated in situ from the aliphatic aldehyde 5) with a simple chiral allylic alcohol 6.¹⁰

Our pursuits began with attempted reductive cross-coupling of aldehyde 5 with allylic alcohol 6 and quickly led to the identification of a limitation in our Ti-mediated reductive cross-coupling technology.⁶ In short, allylic alcohol 6 is not a robust substrate for Ti-mediated coupling, with all attempts leading to low conversion and poor stereoselection (alkene geometry). Given the documented success of related coupling reactions with allylic alcohols that bear groups larger than Me- at the allylic position,^6,11 we opted to pursue reductive cross-coupling with an allylic alcohol that was functionally equivalent to 6 but contained a suitable substituent at the allylic position to facilitate reductive cross-coupling. As illustrated in Figure 3, reaction of aldehyde 7 with the TMS-substituted allylic alcohol 8¹² proceeds effectively by sequential treatment of the aldehyde (2 equiv) with LiHMDS (2 equiv) (to generate the TMS-imine in situ), Ti(Oi-Pr)⁴/s-BuLi (to form the azatitanacyclopropane from the TMS-imine), and the allylic alcohol 8 (1 equiv). In contrast to previous reports that employ the Li-alkoxide of the allylic alcohol in Ti-mediated reductive crosscoupling reactions, use of the free alcohol in this procedure avoided potential Peterson elimination.¹³ Overall, the primary homoallylic amine 9 was produced as a single isomer in 57% yield, with complete transfer of absolute stereochemical information.

Figure 3.

Asymmetric synthesis of (−)-205B via Ti-mediated reductive cross-coupling and stereoselective intramolecular [3+2] cycloaddition.

The stereochemical course of this coupling reaction is consistent with the empirical model depicted in Figure 4. Directed carbometalation is proposed to occur through a boat-like transition state geometry (A) that allows for alignment of the σ_C–Ti bond with π_C=C bond, where minimization of allylic-1,3 strain (A1,3)¹⁴ and non-bonded steric interactions about the developing C–C bond combine to define a highly stereoselective C–C bond forming process. Subsequent syn-elimination of the oxatitanacyclobutane intermediate 17,¹⁵ and hydrolysis of the resulting inorganic complex leads to the homoallylic amine product 9. Competing transition state geometries for this alkoxide-directed Ti-mediated coupling are thought to be destabilized by substantial non-bonded steric interactions (A1,3 and/or eclipsing interactions about the developing C–C bond; highlighted in B–D of Figure 4).

Figure 4.

Stereochemical course of the reductive cross-coupling reaction between aldehyde 7 and allylic alcohol 8.

Returning to the synthesis, oxidation of the primary homoallylic amine (10) was straightforward by sequential exposure to (BzO)₂ and hydrazine. Heating a solution of this species with butylglyoxylate resulted in a highly stereoselective annulation process that likely proceeds by formation of the (E)-homoallylic nitrone and direct intramolecular [3+2] annulation to deliver the stereodefined 1-aza-7-oxabicyclo[2.2.1]heptane 11 in 67% yield (dr ≥ 20:1). The complexity of this annulation process deserves comment as this glyoxylate-based nitrone annulation is rather unique among related intramolecular cycloaddition processes. Homoallylic nitrones are well known to undergo two competive reaction processes: (1) direct [3+2] annulation with the pendant alkene, or (2) sequential aza-Cope rearrangement/intramolecular [3+2] cycloaddition.^{6, 16} This latter pathway can provide a stereochemically distinct product to that derived from direct [3+2] annulation. Unlike homoallylic nitrones generated from aliphatic aldehydes, those derived from glyoxylates have been observed to preferentially react by way of a direct [3+2] annulation through the (E)-isomer. Overall, bicyclic products are typically observed with outstanding levels of stereoselection, unscathed by potential competitive [3,3] sigmatropic rearrangement chemistry typical of aliphatic nitrones.⁶

With the densely functionalized heterocycle 11 in hand, the butyl ester was homologated by a simple three-step sequence to the α-trimethylsilylmethyl-substituted enoate 13. Reduction to the allylic alcohol and deoxygenation with allylic transposition (NBSH, PPh₃, DIAD, NMM)¹⁷ then delivered the allylic silane 14 in 55% yield (over the two-step sequence). Reductive cleavage of the N–O bond (Zn, HOAc), followed by stirring in 3N HCl/THF resulted in iminium ion formation (E) and cyclization to deliver tricyclic product 15 in 82% yield.

While this acid-promoted cyclization reaction proved to be an effective tandem process to establish the tricyclic core of the natural product, the success of this sequence may speak to the relative rates of aza-Sakurai ring closure and aza-Cope rearrangement of the intermediate bicyclic iminium ion E. As illustrated in Figure 5, direct aza-Sakurai-based ring closure would lead to the observed product 15, while competitive aza-Cope rearrangement could deliver an isomeric iminium ion F that has the potential to proceed through a variety of reaction pathways capable of scrambling the C6 stereochemistry. For example, loss of a proton would provide a means to generate an enamine G that could be in equilibrium with the isomeric iminium ion H. Alternatively, F could undergo thermodynamic equilibration by retro aldol/aldol (F → I → J). While the lack of an observed isomeric product in this cyclization does not rigorously exclude the possibility of competitive aza-Cope rearrangement of E, the results certainly support the conclusion that the 1-aza-7-oxabicyclo[2.2.1]heptane 14 is a viable intermediate en route to the azatricyclododecene core of 205B, and that this cyclization process is not effected by the potential pitfalls delineated in Figure 5.

Figure 5.

Success of the cationic annulation may reflect an increased rate for aza-Sakurai vs. [3,3] rearrangement and resulting competitive fragmentation and epimerization pathways.

To complete the asymmetric synthesis of (−)-205B, a means to remove the remaining TMS-substituent, accomplish deoxygenation at C7, and selectively isomerize the alkene was required. The first of these goals was met by treatment with KOt-Bu/18-crown-6 – a procedure known to effect hydroxyl-directed desilylation.¹⁸ Next, treatment with thiocarbonyldiimidazole furnished the functionalized azatricyclododecene 16 in 59% yield (over two steps).¹⁸ Deoxygenation by treatment with tributyltin hydride and AIBN¹⁹ was followed by acid-mediated alkene isomerization to deliver (−) -205B in 61% yield.

In conclusion, we report a concise asymmetric synthesis of (−)- 205B that demonstrates the utility of two stereoselective synthetic methods that have recently emerged from our laboratory: (1) Ti-mediated reductive cross-coupling of allylic alcohols with aldehydes through the intermediacy of a TMS-imine, and (2) path-selective intramolecular [3+2] annulation of glyoxylate-based homoallylic nitrones.⁶ Overall, the combined use of these methods defines a highly stereoselective synthesis of a rare alkaloid that belongs to a family of natural products that likely possess potent and diverse neurological functions. We look forward to exploring the pharmacological profile of synthetic (−)-205B, and establishing the scope and limitations of these powerful chemical methods in stereoselective synthesis.