A Divergent Polyene Cyclization for the Total Synthesis of Greenwayodendrines, Greenwaylactams, Polysin and Polyveoline

Abstract

We present a concise (5–8 steps) asymmetric total synthesis of nine sesquiterpenoid alkaloids featuring four different tetra-/pentacyclic scaffolds. To this end, a novel, bioinspired indole N-terminated cationic tricyclization has been developed enabling the divergent synthesis of greenwayodendrines and polysin. Subtle variation of the C2-substituted indole cyclization precursor allowed switching between indole N- and C-termination. For the latter, a subsequent Witkop oxidation enabled conversion of the cyclopentene fused indole to the eight-membered benzolactam directly furnishing the family of greenwaylactams. In addition, a diastereomeric C-termination product has been elaborated to provide access to polyveoline.

Polyene cyclizations represent one of the most powerful transformations of nature’s toolbox to construct polycyclic molecular architectures in a single step.^[1] Since the seminal work of Stork and Eschenmoser rationalizing the stereochemical outcome of these highly diastereoselective transformations, chemists have repeatedly attempted to mimic nature’s efficiency.^[2] To date, a plethora of initiating groups (e.g, epoxides, allylic alcohols, alkenes, and alkynes) are known and their application in natural product synthesis has been highly successful. Despite the impressive progress in this area, most of the reported cyclizations are limited to carbon or oxygen terminating groups (e.g., alkynes, alkenes, arenes, enol ethers, phenols, aliphatic alcohols, and carboxylic acids) (Scheme 1A).^[1] There are only few examples diverting from these canonical terminating groups, for instance sulfonyl amides. Those have been successfully employed by Knight and Gagné and, to the best of our knowledge, represent the first nitrogen terminated polyene cyclizations.^[3] Besides sulfonyl amides, the only other N-terminated polyene cyclization has been established by Knölker, in which substituted 2-hydroxy carbazoles underwent bicyclization.^[4] The fact that nitrogen terminated polyene cyclizations are yet underdeveloped is surprising as the proposed biosynthesis of several alkaloid natural products involves N-termination of a cationic polycyclization cascaded.^[5] Two representative alkaloids formed via such a pathway are the sesquiterpene indoles polysin (1) and greenwayodendrin-3β-ol (polyavolensinol, 2), which have been isolated together with closely related congeners from the stem of Greenwayodendron suaveolens (Scheme 1B).^[6] A bioactivity screen for polysin (1) and selected family members of the greenwayodendrines revealed antiparasitic activity against Trypanosoma brucei through inhibition of various glycolytic enzymes.^[6a] The putative biosynthetic N-termination pathway and the complex nature of these alkaloid sesquiterpenoids inspired us to devise an asymmetric strategy towards their synthesis. Based on our previous experience, we identified the dual nucleophilicity of the indole and its inherent selectivity for the C3 position as the key challenges of this transformation.^[7]

Scheme 1.

(A) State of the art of polyene cyclization terminations. (B) Retrosynthesis for targeted alkaloid sesquiterpenoids.

Herein, we report on the development of a divergent polyene cyclization to selectively effect either N- or C-terminated tricyclization that culminated in the total synthesis of greenwayodendrines, greenwaylactams, polysin (1) and polyveoline (3).

The retrosynthetic analysis of polysin (1) and greenwayodendrin-3β-ol (2) prompted us to develop a bioinspired, nitrogen terminated polyene cyclization from general indole cyclization precursor 5 (Scheme 1B). The epoxyfarnesyl chain at the C2 position of 5 was envisioned to be installed via nucleophilic substitution, Sharpless asymmetric dihydroxylation and conversion to the epoxide. Indole 5 features two potential nucleophilic positions — namely the indole N and C3 positions. A literature survey of related polyene cyclizations showed a high prevalence of C3-connection and only few examples for C2-connected indoles.^[8] However, this dual nucleophilicity opened the way to envision a complementary C-terminated polyene cyclization, which might also be observed in nature in an analogous fashion. Although the direct C-terminated cyclization products such as 6 have not been isolated from natural sources so far, they can be envisaged as synthetic intermediates in the biosynthesis of polyveoline (3) and greenwaylactam A (4).^[9] In the latter case, a Witkop oxidation was anticipated to reveal the eight-membered lactam motif of greenwaylactam A (4).^[10]

Our synthetic studies commenced with the asymmetric preparation of indole cyclization precursors 12–15 (Scheme 2). Lithiation of commercially available 1-(phenylsulfonyl)indole (8) with n-butyllithium (n-BuLi) and trapping with farnesyl bromide (7) afforded 9 in 80% yield.^[8k,11] Sharpless asymmetric dihydroxylation employing the Corey–Zhang ligand (10), which was required to achieve selective dihydroxylation of the terminal double bond, furnished terminal diol 11 in 73% with only trace amounts of the internal diols.^[12] It is of note that the Corey–Zhang ligand (10) could be recovered and reused without loss in efficiency. A one-pot procedure involving mesylation of the sterically less encumbered secondary alcohol using mesyl chloride (MsCl) and pyridine (py) followed by intramolecular substitution (K₂CO₃, MeOH) gave epoxide 12 in excellent yield.^[13] Free indole 13 was obtained through desulfonylation at 60 °C using potassium hydroxide in degassed methanol to prevent oxidative decomposition. Upon exposure of 13 to either N-chlorosuccinimide (NCS) or N-bromosuccinimide (NBS) in the presence of triethylamine, clean formation of the C3-halogenated indoles 14 and 15 was observed.

Scheme 2.

(A) Enantioselective synthesis of cyclization precursors. (B) Initial screening of regioselectivity for polyene cyclization terminations. See the Supporting Information for detailed procedures and characterization data.

With the cyclization precursors 12–15 in hand, we moved forward to investigate the envisioned N-terminated cyclization. At first, free indole 13 was treated with iron(III)chloride hexahydrate to exclusively deliver the C-terminated product 6a in 25% NMR yield along with a complex mixture of incomplete cyclization products (Scheme 2B, entry 1). Free indole pentacycle 6a turned out to be unstable upon exposure to air and showed decomposition during flash column chromatography.^[14] To circumvent the instability of 6a, N-sulfonylated indole 12 was subjected to boron trifluoride etherate giving C-terminated pentacycle 6b in 22% NMR yield (entry 2).^[8k] The inherent substrate selectivity of 13 for the C3 position prompted us to investigate the C3 halogenated derivatives to enable cyclization via the less nucleophilic indole nitrogen. When 3-chloroindole 14 (entry 3) was submitted to iron(III)chloride hexahydrate, we were delighted to observe N-terminated product 16a in 19% isolated yield. Unfortunately, the C–Cl bond turned out to be reluctant to undergo reduction under a variety of reaction conditions.

Therefore, we turned our attention to the C3-brominated analogue 15, which underwent the cyclization in similar yield (entry 4). Based on these preliminary results, we set out to further optimize the N-termination of 3-bromoindole 15. After careful optimization and a survey of several Lewis acids (see the Supporting Information), the use of boron trifluoride etherate at cryogenic temperatures (-78 °C) in dichloromethane (17 mM) followed by addition of triethylamine turned out to be optimal. A telescoped reduction step with palladium on carbon under hydrogen atmosphere (1 bar) was found to be crucial since attempted isolation of 16b suffered from decomposition. This procedure delivered greenwayodendrin-3β-ol (2) and cis-pentacycle 17 in 31% combined yield as a 2.4:1 diastereomeric mixture (Scheme 3A). The stereochemical outcome of this transformation is in good agreement with the literature^[5,8a] and can be rationalized via two possible scenarios: a chair-boat-like conformation leading to cis-fused pentacycle 17 and a chair-chair-like conformation furnishing trans-fused greenwayodendrin-3β-ol (2). Initial oxidation attempts of the secondary alcohol in greenwayodendrin-3β-ol (2) using common oxidants (e.g., Dess-Martin periodinane (DMP), pyridinium chlorochromate, Bobbitt’s salt) only led to decomposition (see the Supporting Information). However, exposure to acetic anhydride and dimethyl sulfoxide (DMSO) (Albright-Goldman oxidation) tolerated the electron-rich C2- and N-dialkylated indole moiety and furnished greenwayodendrin-3-one (polyavolensinone, 18) in 83% yield.^[15] Applying the same reaction conditions to cis-fused pentacycle 17 afforded the antiparasitic alkaloid polysin (1)^[6a] in 81% yield. For 1, a chemoenzymatic synthesis has been previously reported by Renata.^[14b] Subjecting greenwayodendrin-3-one (18) to potassium tri-sec-butylborohydride (K-selectride) gave greenwayodendrin-3α-ol (19) together with greenwayodendrin-3β-ol (2) as a 4.8:1 diastereomeric mixture (58% combined yield). The structure of 19 was confirmed by single crystal X-ray analysis.^[16] Acetylation of the equatorial alcohol of 2 using acetic anhydride in the presence of 4-dimethylaminopyridine (DMAP) and triethylamine gave 3β-acetoxy-greenwayodendrine (polyavolensin, 20) in 78% yield.

Scheme 3.

(A) N-terminated polyene cyclization and divergent synthesis of indole sesquiterpene alkaloids. (B) C-terminated polyene cyclization and divergent synthesis of greenwaylactam A-C and polyveoline (3). See the Supporting Information for detailed procedures and characterization data.

Having successfully prepared all members of the greenwayodendrine family of natural products, we focused our attention on the total synthesis of greenwaylactam A (4), B (23) and C (24). These secondary metabolites have been isolated in 2021 by Kouam and Tchamgoue from the Cameroonian medicinal plant Greenwayodendron oliveri and were shown to possess antibacterial activity.^[9a] Continuation of our previous screening campaign (Scheme 2B) showed that exposure of epoxide 12 to different Lewis acids at 0 °C or –78 °C affords trans-pentacycle 6b^[16] in only 22–27% NMR yield together with a complex product mixture. Initial attempts to perform an alternative radical cyclization using the Nugent–RajanBabu reagent (Cp₂Ti^IIICl) only led to bicyclization.^[17] At high temperatures (115–145 °C) the desired tricyclization product 6a was obtained, albeit in low yields (<10%). Finally, we found that the use of catalytic amounts of methanesulfonic acid (MsOH, 10 mol%) in 1,1,1,3,3,3-hexafluoroisopropanol (HFIP, 19 mM) efficiently promotes cationic tricyclization of 12 with only small amounts of interrupted cyclization products formed (see the Supporting Information).^[7c] Notably, HFIP has emerged as a versatile solvent due to its low nucleophilicity, comparably high acidity, high hydrogen bond donor strength and stabilization ability of cations. Additionally, it has been proposed to enable prearrangement of acyclic polyenes to promote more selective and higher yielding cyclizations.^[18] Under these conditions, the trans- and cis-fused pentacycles 6b and 21 were obtained in 67% combined yield in a 5.7:1 diastereomeric ratio (Scheme 3B). To reveal the greenwaylactam scaffold, we investigated the Witkop oxidation of the indole moiety.^[10] Deprotection of 6b with potassium hydroxide in ethanol provided unstable indole 6a, which underwent partial Witkop oxidation to greenwaylactam A (4) upon exposure to atmospheric oxygen in dichloromethane. In light of this finding, we developed a superior one-pot protocol that involved addition of an aqueous solution of sodium periodate after deprotection of 6b. Under these conditions, greenwaylactam A (4) was directly obtained in 63% yield.^[19] The structure of 4 was unambiguously confirmed by single crystal X-ray analysis.^[16] Attempts to directly oxidize greenwaylactam A (4) to greenwaylactam B (23) were surprisingly challenging. Therefore, we opted for the oxidation of protected indole 6b with DMP to give ketone 22 in 91% yield. Removal of the sulfonyl protecting group was achieved under basic conditions (KOH, EtOH) and treatment of the crude product with sodium periodate yielded greenwaylactam B (23) in 51% over two steps. Acetylation of greenwaylactam A (4) using acetic anhydride and p-toluenesulfonic acid (p-TsOH) provided greenwaylactam C (24) in 36% yield. Finally, reduction of cis-pentacycle 21 to the indoline was accomplished under mild conditions (Pd/C, 1 bar H₂) and subsequent DMP oxidation afforded ketone 25. A highly diastereoselective reduction of the ketone in 25 to an axial alcohol with K-selectride (95% yield) and consecutive desulfonylation employing a solution of sodium naphthalenide at –78 °C furnished the antiparasitic alkaloid polyveoline (3) in 92% over two steps.^[6a,9b,c] Spectroscopic data for all synthesized natural products were identical in all respects to those reported in the literature (see the Supporting Information). In summary, we have disclosed the first indole nitrogen terminated polyene cyclization furnishing cis- as well as trans-fused pentacyclic frameworks, respectively. Both diastereomers were elaborated to achieve the divergent synthesis of polysin and the family of greenwayodendrines (6–8 steps, 3–10% overall yield). Additionally, we found a combination of catalytic Brønsted acid in HFIP to be essential for high yielding C-terminated tricyclization. This gave access to polyveoline and enabled a bioinspired late-stage Witkop oxidation to greenwaylactams A–C (5–8 steps, 5–21% overall yield). Ultimately, we believe that the key findings of this study will stimulate further research regarding N-terminated polyene cyclizations to access structurally related alkaloids.