Total Synthesis of Strempeliopidine and Non-Natural Stereoisomers Through a Convergent Petasis Borono–Mannich Reaction

Abstract

Strempeliopidine is a member of the monoterpenoid bisindole alkaloid family, a class of natural products that have been shown to elicit an array of biological responses including modulating protein-protein interactions in human cancer cells. Our synthesis of strempeliopidine leverages palladium-catalyzed decarboxylative asymmetric allylic alkylations to install the requisite all-carbon quaternary centers found in each of the two monomeric natural products, aspidospermidine and eburnamine. Initial studies employing a Suzuki–Miyaura cross-coupling followed by diastereoselective hydrogenation provided evidence for a structural reassignment of the natural product. Our final synthetic sequence employs a diastereoselective Petasis borono–Mannich reaction to couple eburnamine to a trifluoroborate aspidospermidine derivative. These convergent approaches enabled the synthesis of eight diastereomers of this heterodimer and offer support for the reassignment of the absolute configuration of strempeliopidine.

Graphical Abstract

Introduction:

Natural products possess a wealth of structural diversity and biological properties, which continue to motivate investigations into their therapeutic potential.¹ Monoterpenoid bisindole alkaloids, a class of natural products that include the FDA-approved anticancer drugs vinblastine (1) and vincristine (2), are hypothesized to modulate protein-protein interactions (PPIs) during cell mitosis, thus inducing cell death.^2–5 PPIs are essential for biological processes including immune response, nitric oxide synthase, and pre-programmed cell death, among many others.^6–7 Despite the importance of PPIs in biological function, small molecules that modulate PPIs are relatively rare and therefore understanding their structure-activity relationship is of great interest to the scientific community.^{4, 8} Notably, dimeric alkaloids tend to show greater bioactivity than their monomeric constituents, and thus have captured the interest of the scientific community over the past 60 years.⁴

The bisindole alkaloid family is comprised of dimeric compounds, generated from two independently biosynthesized monomeric alkaloids unified through at least one carbon–carbon bond (Figure 1A).^9–11 Stemming from their rich bioactivity and wealth of structural diversity, numerous syntheses of the requisite monomeric indole alkaloids have been pursued.^12–24 In stark contrast, total syntheses of complex bisindole alkaloids remains underexplored; as the parent heterodimers contain elements from aspidosperma, iboga, kopsia, corynanthe, clevamine, and eburna alkaloid families,^25–28 their syntheses often require two fundamentally different strategies. The direct assembly of heterodimeric bisindole alkaloids from their monomeric components through a convergent coupling is an appealing strategy, though it requires regio- and diastereoselective control of the unifying carbon–carbon bond forming event.^{12, 29} Biomimetic approaches to access numerous members of the bisindole alkaloid family have been attempted with varying levels of success, and often suffer from low yields, poor regioselectivities, and limited extension to analog syntheses.^30–32 We envisioned that developing a convergent, modular coupling strategy would enable access to several members of this class of natural products, as well as non-natural analogs, and facilitate the exploration of their biological properties.

Figure 1.

(A) Representative members of the monoterpenoid bisindole alkaloid family, (B) structural revision of strempeliopidine, (C) previously proposed structures of strempeliopidine

Strempeliopidine (6) (Figure 1B) was first isolated from the leaves of Strempeliopsis strempelioides in 1984 by Novotný and coworkers.^33–34 This molecule is comprised of two monoterpenoid indole alkaloids: a western pentacyclic fragment derived from aspidospermidine and an eastern pentacyclic fragment derived from eburnamine. Like other bisindole alkaloids, strempeliopidine is proposed to arise from an enzyme-mediated Friedel–Crafts type reaction between aspidospermidine and an iminium ion derived from eburnamine.^35–36

Synthetic studies by the isolation chemists, as well as ¹H NMR data, were used to establish the proposed structure of compound 7. However, there is some ambiguity regarding the true stereochemistry of strempeliopidine as highlighted by a proposed structural revision (8) from Kam and Choo in 2006 (Figure 1C).¹⁰ While the authors did not reisolate strempeliopidine, they proposed the stereochemical revision based on biological data from the Strempeliopsis plant and coupling constant analysis of the originally published ¹H NMR data, though the absolute stereochemistry of strempeliopidine has thus far not been substantiated.¹⁰

The reported structures of strempeliopidine consists of ten rings, seven stereogenic carbons (three of which are all-carbon quaternary centers), and a challenging C(sp²)–C(sp³) bond that unites the two monomeric natural products. Furthermore, the complexity of these two fragments, as well as the challenges associated with forging the unifying C(sp²)–C(sp³) bond, has precluded synthetic access to this natural product. Although both eburnamine^{29–30, 37–44} and aspidospermidine^{13, 22, 45–48} have been synthesized, there are no reported syntheses of strempeliopidine to date. Conflicting reports in the literature pertaining to the true structure of strempeliopidine,¹⁰ as well as the unique biological properties of bisindole alkaloids, further motivated our synthetic investigations.

Based on the conflicting reports of strempeliopidine’s structure and the potential to access non-natural dimeric analogs, we devised a modular approach to both reported structures of strempeliopidine from enantioenriched monomeric indole alkaloid constituents (Scheme 1). Specifically, noting the differences between Novotný’s (i.e., 7) and Kam’s (8) proposed structures, a route that would allow precise control over each stereocenter on the eburnamine fragment was necessary. Both the (20’β, 21’⍺) and (20’β, 21’β) ring junctions of the eastern eburna fragment would need to be prepared, in addition to designing complementary strategies to control the stereocenter where the monomers are linked (C16’). To prepare the original structure proposed by Novotný (7), we envisioned using a Suzuki–Miyaura/reduction approach to prepare the 16’ β stereocenter from epi-eburnamine triflate 9 and aspidospermidine-derived boronic ester 12 (Scheme 1A). To synthesize the requisite eburnamonine vinyl triflate 9, a key Bischler–Napieralski reaction would be employed following trans-selective hydrogenation to afford the trans-20’β, 21’⍺-ring fusion.²⁹ Lactam (–)-10 would be derived from δ-valerolactam (11) using our lab’s previously established scalable route exploiting a palladium-catalyzed asymmetric decarboxylative alkylation.⁴⁹ To access the western fragment 12, we would apply a hydroamination/Pictet–Spengler sequence to build up the pentacyclic framework of dehydro-aspidospermidine derivative (12) from acyl indole 13.^50–51 Another decarboxylative allylic alkylation would be used to set the C3 all-carbon quaternary stereocenter, starting from 5-bromoindole (14).

Scheme 1.

(A) Retrosynthetic analysis of the proposed structure by the Novotný and coworkers, and (B) retrosynthetic analysis of the structure proposed by Kam.

Results and Discussion:

To prepare the revised structure reported by Kam (i.e., 8) (16’⍺, 20’β, 21’β), we designed a highly modular synthesis in which several stereoisomers of eburnamine ((–)-15) could be united with aspidospermidine using a 1,2-addition into the eburnamine-derived iminium via a Petasis borono–Mannich reaction to prepare the 16’⍺ stereocenter from eburnamine ((–)-15) and aspidospermidine-dervied trifluoroborate salt (+)-16 (Scheme 1B). We posited the diastereoselectivity of this transformation would rely on inherent substrate preference that has been demonstrated in the syntheses of related bisindole alkaloids containing an eburna monomer.^{29, 31} Eburnamine (15) would be accessed by a Bischler–Napieralski reaction followed by cis-selective hydrogenation to afford the cis-20’β, 21’ β-ring fusion from lactam (–)-10.³⁸ The western fragment (+)-16 would be prepared from an analogous sequence from indole (18) via lactam (+)-17. Through application of these two coupling strategies, we expected to be able to answer the question about the structural assignment of strempeliopidine. Additionally, this would provide access to additional bisindole alkaloids bearing natural and non-natural stereochemistry to explore the biological properties of this class of molecules.

Our synthesis commenced with carboxylactam (±)-19, which could be accessed from δ-valerolactam in three steps (Scheme 2).⁴⁹ Applying our previously developed asymmetric decarboxylative alkylation technology, we could access decagrams of enantioenriched lactam (–)-10 in 87% yield and 96% ee using 1.5 mol% of Pd(OAc)₂ and (R)-(CF₃)₃-t-BuPHOX.²⁹ Using chemistry previously developed in our laboratory, lactam (–)-10 could be converted to iminium salt (–)-20, which we identified as a suitable precursor for both (–)-epi-eburnamonine ((–)-21) and (+)-eburnamonine ((+)-22). Following a procedure outlined by Wenkert, we were able to form the eastern pentacyclic ring system ((–)-21) by treatment of iminium (–)-20 with KOAc.³⁸ This allowed for a diastereoselective hydrogenation of the iminium salt using Pd/C to yield (–)-epi-eburnamonine ((–)-21) in 74% yield and >20:1 d.r. Reversing the order of ring closure and hydrogenation, iminium (–)-20 was exposed to Pd/C and H₂, followed by DBU, to deliver the pentacycle in 99% yield and 6.6:1 d.r. favoring (+)-eburnamonine ((+)-22).⁵² Treatment of either (–)-21 or (+)-22 with lithium aluminum hydride provided access to (–)-eburnamine ((–)-15) and (+)-epi-eburnamine ((+)-23). Additionally, (+)-eburnamine ((+)-15) and (–)-epi-eburnamine ((–)-23) were synthesized through this synthetic route by using (S)-(CF₃)₃-t-BuPHOX, providing access all four stereoisomers of eburnamonine in ten steps from commercially available material.

Scheme 2.

Preparation of (–)-eburnamine and (+)-epi-eburnamine.

The synthesis of (+)-aspidospermidine ((+)-26) began with an enantioselective decarboxylative alkylation of acyl indole (±)-24, which can be accessed in six steps from indole (Scheme 3).⁵⁰ Utilizing 1 mol% of a palladium(II) precatalyst and (S)-(CF₃)₃-t-BuPHOX provided enantioenriched allyl indole (+)-25 in 97% yield and 94% ee. Using conditions employed in our synthesis of a related alkaloid, limaspermidine, a 4-step sequence of hydrozirconation/amination,⁵³ reductive cyclization, alkylation, and annulation provided sufficient quantities of (+)-aspidospermidine ((+)-26) for further evaluation of our divergent coupling.⁵⁰ To install a functional handle for the coupling of aspidospermidine ((+)-26) and (–)-epi-eburnamonine ((–)-21), we opted to pursue a bromination/borylation approach. Selective monobromination of aspidospermidine ((+)-26) could be achieved through benzyl protection of aspidospermidine under reductive amination conditions (94% yield), followed by treatment with NBS. Miyaura borylation of bromoindole (+)-28 produced pinacol ester (+)-29 in 95% yield, setting the stage for our convergent coupling.

Scheme 3.

Preparation of functionalized (+)-aspidospermidine derivative.

With pinacol ester (+)-29 in hand, preparation of the eburna fragment for the key coupling reaction was achieved by enolization/triflation of (+)-epi-eburnamonine ((+)-21) to provide triflate 9 in 80% yield (Scheme 4). To our delight, subjecting 9 and (+)-29 to canonical Suzuki–Miyaura conditions formed the desired C–C bond in 93% yield (30). We next turned our attention to the challenging diastereoselective reduction of the sterically hindered trisubstituted olefin of 30 and reductive cleavage of the benzyl group. Unfortunately, mild hydride sources under acidic conditions did not lead to any productive reaction. We next explored several homogeneous catalysts (Pfaltz, Crabtree, and Brown catalysts) that are designed for reducing styrenyl tri- and tetra-substituted olefins;^54–56 disappointingly, no reaction was observed even at elevated pressures. Turning to heterogenous catalysis, it was found that subjecting dimer 30 to palladium on carbon under an atmosphere of hydrogen resulted in adsorption of the substrate to the solid support, preventing material recovery. Other heterogenous catalysts typically used for hydrogenation either failed to provide the desired product or resulted in over-reduction of the substrate.

Scheme 4.

Cross-coupling of epi-eburnamonine derivative with aspidospermidine derivative.

Hypothesizing that using a heterogeneous catalyst that could be degraded upon completion of the reaction would enable recovery of the substrate, dimer 30 was exposed to Raney nickel under an atmosphere of H₂ (Table 1, Entry 1). Excitingly, partial hydrogenation of the trisubstituted olefin was observed after 3.5 hours (32). Using conditions described by Okimoto, exposure of 30 to Raney nickel and H₂SO₄ allowed for removal of the benzyl group, leaving the olefin intact (31) (Entry 2).⁵⁷ Unfortunately, combining these two procedures either in a two-step or one pot procedure only led to intractable decomposition (Entry 3). The challenges regarding benzyl group removal, in addition to olefin reduction, led us to pursue an alternative strategy that avoided the use of protecting groups.

Table 1.

Attempted small scale hydrogenation and deprotection of benzyl heterodimer (+)-30

Entry	Conditions	7	31	32	30
1	Raney nickel, H2 (balloon)	trace	–	Yes	Yes
2	Raney nickel, H2SO4	trace	Yes	–	Yes
3	Raney nickel, H2 (balloon), then H2SO4	trace	–	–	No

To this end, bromoindole (±)-33 was subjected to palladium-catalyzed asymmetric decarboxylative alkylation to provide allylindole (–)-13 in 97% yield and 92% ee (Scheme 5). Hydrozirconation and subsequent Pictet–Spengler led to tetracycle (–)-34 in 53% yield. Acylation and annulation were accomplished over three steps in 83% yield to provide imine (–)-35. Imine (–)-35 was then used in a Miyaura borylation to produce a coupling partner ((–)-12) that was devoid of protecting groups. Suzuki cross-coupling of (–)-12 with triflate 9 was next conducted to afford imine dimer 36 in 82% yield, the structure of which was unambiguously confirmed by single-crystal X-ray analysis. Gratifyingly, reduction of the imine and amide was simultaneously achieved using LiAlH₄, avoiding the difficult benzyl deprotection from our initial route. Lastly, subjection to Ra/Ni at 4 bar H₂ produced 7, the structure of Novotný’s proposed assignment (16’ β, 20’β, 21’⍺) in the original isolation paper.

Scheme 5.

Completion of the synthesis of the originally proposed structure of Strempeliopidine (7).

When comparing the NMR spectra of this compound to the list of peaks given by Novotný and coworkers, we noticed several discrepancies. Firstly, this synthetic compound was isolated as a mixture of atropisomers, which is inconsistent with the isolation chemists’ report. Secondly, dimer 7 exhibited instability in CDCl₃, thus precluding characterization by ¹³C NMR. Furthermore, as ¹³C NMR data for this compound was not included in the initial report of strempeliopidine, ¹H NMR, optical rotation, and HRMS were used to compare our compound to that reported by the isolation. When evaluating 7 by ¹H NMR, several key peaks, including that of the stereocenter formed through the Suzuki cross-coupling/hydrogenation (C16’) and of C21’, were significantly shifted and possessed different coupling constants than those reported by Novotný and coworkers.³⁴ As previously reported, the ABX spin system of C16’ in related bisindole alkaloids norpleiomutine^{31, 58} and pleiomutine⁵⁹ was observed at δ 4.95 ppm (dd, J= 12, 5 Hz) and δ 5.0 ppm (dd, J= 11, 5 Hz) respectively. This is comparable to the reported values for strempeliopidine (1H, δ 4.94 ppm, dd, J= 11.5, 4.6 Hz). However, for synthesized 7, we observed 1H, δ 5.51 ppm, d, J= 7.8 Hz.^{28, 60–62} Our NMR studies and comparison to other related bisindole alkaloids suggest that the initial structural assignment of strempeliopidine from 1984 requires reassessment, in agreement with the assignment by Kam in 2006 (see Supporting Information for comparison).¹⁰

Before proceeding, we reexamined the synthetic studies initially reported by Novotný and coworkers. In addition to ¹H NMR and MS, the isolation chemists used degradation studies and semi-synthesis to assist in identifying the structure of strempeliopidine. It was reported that strempeliopidine could be synthesized by stirring aspidospermidine and eburnamine isolated from Strempeliopsis strempelioides in 2 N HCl at 20 °C for 48 hours.³⁴ However, in our hands, when exposing either (–)-eburnamine (–)-15 or (+)-epi-eburnamine (+)-23 and (+)-aspidospermidine (+)-26 to the reported reaction conditions the desired product was not observed; instead, each of the monomers could be quantitatively reisolated. Based on coupling constant analysis in Kam’s proposed structural revision which is substantiated by comparing the data for our synthesized dimer 7 with the primary data from Novotný and coworkers, we began targeting Kam’s revised structure of strempeliopidine (8).

Following the synthesis of 7, we next set out to establish the relative configuration at C16’ and C21’ of strempeliopidine. With a reliable route to all four stereoisomers of eburnamonine (see Scheme 2, (+)-21, (–)-21, (+)-22, and (–)-22), we focused on developing a method to forge the linkage between the two indole monoterpenoid alkaloids, aspidospermidine ((+)-26) and eburnamine ((–)-15), with ⍺-selectivity at C16’ (see Scheme 1B, 8). Given the strong bias in facial selectivity for hydrogenation of this class of molecules, we elected to explore addition of aryl nucleophiles to iminium ions as a means to directly forge the requisite stereocenter at C16’. Inspired by previous reports of Petasis borono–Mannich additions of boronic acids to iminium salts, and hypothesizing that aryl nucleophile addition to an iminium ion derived from eburnamine might have a similar facial preference to that observed for hydrogenation, we opted to pursue this strategy.^63–65

Preliminary investigations into the Petasis borono–Mannich reaction were conducted on a model system using aryl boronic acid 37 and tertiary amine-containing hydroxy-indole 38. Using several solvents commonly employed in Petasis borono–Mannich reactions, including MeOH, CH₂Cl₂, or CH₃CN, we did not observe an appreciable amount of the desired product (Table 2, entry 1). Positing that ionization of the hydroxy-indole was slow, we turned to more acidic solvents and found that trifluoroethanol (TFE) and hexafluoroisopropanol (HFIP) promoted the reaction in 16% and 79% yield, respectively (entries 2 and 3). Additionally, we found that changing the solvent to i-PrOH (entry 4), modifying the equivalency (entry 5), increasing the temperature (entry 6), or changing the identity of the aryl boron species (entries 7 and 8) had detrimental effects on the yield of this reaction. Lastly, when subjecting boronic acid 37 and (–)-eburnamine ((–)-15) or (+)-epi-eburnamine ((+)-23) to the reaction conditions, we were pleased to observe a single diastereomer bearing the same C16’⍺ stereochemistry as Kam’s proposed structure of strempeliopidine (8) (see Supporting Information).

Table 2.

Development of a Petasis borono–Mannich coupling between boronic acids and hydroxy-indoles.

Entry	Conditions	Yield of 39a
1	MeOH, CH2Cl2, or CH3CN	–
2	TFE	16%b
3	HFIP	79%, 9:1 d.r. (64% isolated)
4	i-PrOH	–
5	1 equiv ArB(OH)2, HFIP	31% b
6	60 °C, HFIP	72% b
7	ArBpin, HFIP	<5% b
8	ArBF3K, HFIP	34% b

^aYields determined by ¹H NMR analysis of crude reaction mixture relative to Et₄Si internal standard.

^bDiastereomeric ratio not determined.

With this new method in hand, we began by synthesizing the boronic acid coupling partner from aspidospermidine (Scheme 6). Due to stability issues during purification, this boronic acid ((+)-40) was generated in situ and used immediately in the Petasis borono–Mannich reaction with (–)-eburnamine ((–)-15). Unfortunately, upon mixing these two coupling partners, only products resulting from proto-deborylation ((+)-27) and dehydration ((–)-42) were observed. Hypothesizing that proto-deborylation was more facile than in the initial model system, we explored other, more stable boron nucleophiles.^66–67

Scheme 6.

Attempted Petasis borono–Mannich with aspidospermidine boronic acid derivative.

Having previously established the ability of trifluoroborate salts (RBF₃K) to form the desired C–C bond, albeit in diminished yields in the model system (Table 2), we next synthesized trifluoroborate salt (+)-43 (Scheme 7). Upon exposure of (+)-43 to eburnamine ((–)-15) in HFIP, the desired product ((–)-41) was formed in 98% yield as a single diastereomer. It is possible that the aryl–BF₃K salts can be hydrolyzed in solution, serving as a reservoir for the slow release of boronic acid. Alternatively, aryl–BF₂ has also been proposed to be the active arylation species in similar reactions.^68–69 Whatever the rationale, the Petasis borono–Mannich coupling of (+)-43 and (–)-15 produced (–)-41 as a single diastereomer and in excellent yield. Finally, deprotection of the N-Bn indoline under acidic conditions using Raney nickel gave Kam’s strempeliopidine 8, the structure of which was unambiguously confirmed after treatment with 4-bromobenzenesulfonyl chloride by X-ray diffraction. Comparison of the ¹H NMR data for the natural material to this synthetic sample suggests that 8 indeed possesses the correct relative configuration at C16’and C21’, as proposed by Kam; however, while the optical rotation was of similar magnitude to the isolated material ([α]²¹_D = –64.0 °), it was of opposite sign ([α]²⁰_D = +100 ° lit.).

Scheme 7.

Petasis borono–Mannich with aspidospermidine trifluoroborate derivative and determination of the relative configuration of strempeliopidine.

It is well known that members of the eburna-family of natural products have been isolated as enantiopure compounds and racemic mixtures⁷⁰ possessing all potential stereochemical permutations. With a route that allows facile access to each of the four stereoisomers of eburnamonine (vide supra, Scheme 2), we applied our newly developed Petasis borono–Mannich reaction to synthesize non-natural stereoisomers of strempeliopidine (Scheme 8). Gratifyingly, the Petasis borono–Mannich coupling followed by debenzylation provided access to analogs of strempeliopidine possessing all four C20’and C21’ diastereomeric combinations in good yields (45–47). In each case, the Petasis borono–Mannich reaction produced diastereomerically pure material with the newly formed C16’ stereocenter having a proton anti to the ethyl group at C20’. This method, in parallel with cross-coupling/hydrogenation approach, allowed for divergent access to different stereochemical outcomes from the same starting materials in an efficient and facile manner.

Scheme 8.

Synthesis of heterodimeric stereoisomers and the determination of the absolute configuration of strempeliopidine.

Unfortunately, none of these diastereomers matched the ¹H NMR and optical rotation reported by Novotný and coworkers. Since the coupling of (+)-aspidospermidine ((+)-26) with (–)-eburnamine ((–)-15) gave heterodimer 8 that most closely matched the original report by Novotný and coworkers, we elected to synthesize the enantiomer of this compound. To our delight, upon coupling of (–)-aspidospermidine ((–)-26) and (+)-eburnamine ((+)-15) and deprotection of the N-Bn indoline, strempeliopidine stereoisomer 6 (ent-Kam’s assignment, i.e., 8) was isolated, which most closely matched the ¹H NMR and optical rotation of the natural product ([α]²⁴_D = +71.0 °) (Scheme 8). Taken as a whole, these data suggest a revision of the structure of strempeliopidine to 6, the enantiomer of Kam’s proposed assignment; however, as efforts to secure an authentic sample of the natural product have been to date unfruitful, we cannot say with complete certainty that 6 has the correct absolute configuration or if there was an error or impurity in the original measurement of the natural product.⁷¹ Given the circumstances, the final determination may never be known.

In conclusion, eight stereoisomers of the bisindole alkaloid, strempeliopidine, were synthesized via a modular strategy that features a palladium-catalyzed decarboxylative asymmetric allylic alkylation and a Petasis borono–Mannich reaction. Essential to this approach was the identification of conditions that allowed for the convergent coupling of aspidospermidine and eburnamine monomers. A traditional palladium- catalyzed cross-coupling approach, followed by hydrogenation, gave complimentary access to an epimer of the natural product, while arene addition to a masked iminium ion by a Petasis borono–Mannich cross-coupling provided the configuration found in strempeliopidine. Analysis of J-couplings, analog synthesis, and X-ray diffraction allowed determination of the relative configuration of previously misassigned stereocenters, while the optical rotation provided by the isolation chemists was used to determine the absolute configuration of the natural product. The general coupling strategy reported herein to access strempeliopidine, and its stereoisomers, serves as a foundation for future studies into the synthesis of other dimeric compounds in the bisindole monoterpenoid alkaloid family.