Syntheses of Aristotelia Alkaloids: Reflections in the Chiral Pool

Abstract

The Aristotelia alkaloids are a family of monoterpene indole alkaloids possessing a characteristic azabicyclononane scaffold, which has been assembled by several synthetic methods. Herein we review those approaches that have adopted a biomimetic approach to unite heterocyclic synthons with chiral pool monoterpenes. Throughout this discussion, the tendency of monoterpenes like α-pinene and limonene to undergo racemization is highlighted, revealing the challenges in developing stereospecific syntheses of these alkaloids. Finally, we provide a brief discussion of how these synthetic efforts have enabled the structural confirmation and elucidation of the Aristotelia alkaloids’ absolute configurations, including our own recent efforts to employ bioactivity data to deduce the naturally occurring configuration of the quinoline alkaloid aristoquinoline.

Graphical Abstract

1. Introduction

The Aristotelia alkaloids are a family of ~50 monoterpene indole alkaloids produced by the plants of the Aristotelia genus.¹ Whereas most indole monoterpene alkaloids are biosynthesized through the condensation of tryptamine with the iridoid glycoside secologanin,² the terpene portion of the Aristotelia alkaloids is derived from a simple, non-rearranged monoterpene precursor. As exemplified by hobartine (1) and makomakine (2), the tryptamine and monoterpene units combine to form the hallmark azabicyclo[3.3.1]nonane core found in most of the Aristotelia alkaloids (Figure 1).³ The remaining members of the family of alkaloids are thought to be biosynthetically derived from 1 and 2 through various oxidations and ring fusions between the indole and monoterpene core.⁴ For instance, aristoteline (3) is derived from an intramolecular Friedel-Crafts alkylation of 1 or 2.⁵ In addition to these relatively simple peripheral changes, more elaborate structural rearrangements give rise to alkaloids like peduncularine (4) and aristoquinoline (5).

Figure 1.

Representative members of the Aristotelia alkaloids highlighting the characteristic azabicyclic scaffolds biosynthesized from monoterpene (red) and tryptamine (blue) units.

Over the years, the unique azabicyclic architecture of the Aristotelia alkaloids has attracted the attention of synthetic chemists. Although elegant syntheses involving cycloaddition reactions and ring-closing metathesis have been reported, the majority of these synthetic approaches have mirrored the alkaloids’ biosynthesis, uniting a monoterpene with an indole-containing synthon.^{6, 7} In addition to providing highly convergent routes, these biomimetic approaches also take advantage of various chiral pool monoterpenes, thereby allowing stereocontrolled syntheses of the alkaloids. While appealing, as we will discuss here, these monoterpene building blocks possess multiple mechanisms for racemization. Nevertheless, by overcoming these challenges, these stereocontrolled syntheses have aided in defining the relative and absolute configurations of the Aristotelia alkaloids. Included in these efforts are our recent use of bioactivity to identify the more active, and presumably naturally occurring, enantiomer of 5.

2. Mercury-mediated Ritter-like reactions

In 1978, Delpeche and Khuong-Hu reported a remarkably efficient synthesis of the azabicyclo[3.3.1]nonane core by employing a mercury nitrate-mediated Ritter-like reaction between α-pinene (6) and acetonitrile to synthesize imine 7, which in turn, undergoes a completely diastereoselective in situ reduction to the amine 8 (Scheme 1A).⁸ Later, Lévy et al. adapted this approach to report the first synthesis of an Aristotelia alkaloid, by condensing 7 with isatin to install the indole nucleus, which upon a series of reductions yielded 1.⁹ Applying this reaction sequence to (−)-β-pinene (9), furnished the isomeric 2 bearing an exocyclic olefin.⁹ Stevens and Kenney further simplified the approach by conducting analogous mercury-mediated Ritter reactions between indolyl-3-acetonitrile and 6 or 9 to produce imines 10 and 11, respectively, which can be similarly reduced in situ to 1 and 2 (Scheme 1).¹⁰ Interestingly, in addition to controlling whether the olefin of the products was endo or exocyclic, the use of α- or β-pinene also dictated the enantiospecificity of the reaction: whereas 9 produced enantioenriched (+)-2, 6 yielded (±)-1. This racemization is explained by the mechanism depicted in Scheme 1A. Mercuriation of the olefin and concomitant opening of the cyclobutane ring of 6 initially produces carbocation intermediate 12. Attack of the carbocation by the indolyl-3-acetonitrile in a Ritter-like fashion forms the nitrilium 13, which undergoes intramolecular cyclization and demercuration to furnish 7 or 10. However, if intermediates 12 or 13 undergo a [1,3]-sigmatropic rearrangement to their respective enantiomers prior to cyclization, ent-7 or ent-10 will be formed.⁸ The fact that 7 and 10 formed from this reaction are racemic mixtures indicates the interconversion of 12 and ent-12 and/or 13 and ent-13 must occur faster than cyclization. Conversely, intermediates 14 and 15 formed from the mercuriation of 9 can only undergo cyclization at C16, thereby preserving the stereochemistry of the reactant (Scheme 1B).

Scheme 1.

Ritter-like reactions between nitriles and α-pinene (A) or β-pinene (B) assemble the Aristotelia alkaloid core in a convergent fashion. The depicted mechanism rationalizes the observation that racemic product is formed with α-pinene is employed, while β-pinene produces enantioenriched (+)-hobartine.

3. Brønsted acid-mediated Ritter-like reactions

To deviate from the use of toxic mercury nitrate, subsequent investigations into the Ritter-like reactions explored the use of Brønsted acids. Marschoff et. al. successfully used perchloric acid to generate carbocation intermediate 16 from 6 and 9, which undergoes a Ritter-like reaction with alkyl- and aryl-nitriles to yield bicyclic imines (17, Scheme 2).^11–13 Because of the large excess of nitrile used in these reaction, the penultimate carbocation intermediate is intercepted by a second equivalent of the arylnitrile, which upon hydrolysis of the resulting nitrilium is converted to the amide in 17. In addition to 6 and 9, other monocyclic terpenes like (+)-limonene (18), α-terpineol (19), and terpinolene (20) gave rise to 17. Furthermore, even acyclic monoterpenes including geraniol and nerol are also viable substrates for the reaction, presumably by first undergoing an acid-mediated cyclization to generate 16. Interestingly, unlike the mercury-mediated reactions, when chiral, non-racemic terpenes were employed in the reactions, the products of these perchloric acid-mediated Ritter reactions are optically active. And while the enantiopurities of the products were not reported, this does suggest complete racemization of the products does not occur.

Scheme 2.

An array of monoterpenes and nitriles undergo Brønsted acid mediated Ritter-like reactions without complete erosion of the monoterpene’s enantiopurity. Values under each monoterpene represent the [α]_D (MeOH, 0.9) of the products formed with benzonitrile (R=Ph).

4. Synthesis of aristoquinoline: Ritter-like reaction approach

Inspired by these Ritter-like reactions to construct the Aristotelia alkaloids, we recently employed a sulfuric acid-mediated Ritter reaction for the synthesis of aristoquinoline (5).¹⁴ Based on the precedent with other Brønsted acid, we anticipated this approach would provide facile access to both enantiomers of 5 by employing either (−)- or (+)-6. Indeed, the reaction between 4-cyanoquinoline (21) and (−)-6 with sulfuric acid affords the bicyclic imine 22, which could be diastereoselectively reduced to 5 with NaB(OAc)₃H (Scheme 3). Notably, because the nitriles were employed as the limiting reagent, no products arising from a second Ritter reaction were observed. However, 22 and 5 from these reactions were racemic mixtures. Similar racemization was encountered when other chiral, non-racemic terpenes such as 9, 18 and 19 was used in lieu of 6. Along with the desired product 22, we also isolated and characterized a positional isomer 23, which can also be easily reduced to iso-aristoquinoline (24). Although previous Brønsted acid-mediated Ritter reactions have reported the formation of norbornyl side-products, 23 and 24 represent unprecedented scaffolds.

Scheme 3.

Synthesis of racemic aristoquinoline and iso-aristoquinoline using a Ritter-like reaction. Hydrogens in red in the product depict sites of deuterium incorporation when the reaction is conducted with D₂SO₄. R = 4-quinoline.

Based on a pair of isotopic labeling studies, the mechanism in Scheme 3 was proposed that accounts for the formation of 23 and the racemic products. As in the perchloric acid-mediated reactions, in the presence of sulfuric acid, 6, 9, 18, and 19, all form the terpinyl carbocation 16 that undergoes nucleophilic attack by a nitrile. A subsequent intramolecular attack by the olefin on the resulting nitrilium (25) and deprotonation ultimately affords 22. In support of this mechanism, Gujarati and Reber recently demonstrated intermediate 25, synthesized by dehydrating the corresponding amide, also produces 22.¹⁵ The isomeric imine 23 is formed from the deprotonation of 16 to 20, followed by re-protonation to either 16 or 26. Importantly, 20 and 26 are achiral intermediates, suggesting their formation is responsible for the racemic product mixtures. Interestingly, when performing the reaction using D₂SO₄, it became apparent that deuterium was also incorporated at the C11 and C15 positions of the products, suggesting a second mechanism for racemization, whereby the endocyclic olefin is reversibly protonated to an achiral intermediate, may also be operating.

Although the Ritter-like reaction with monoterpenes has provided general, rapid access to the Aristotelia alkaloid scaffold, this approach is not without its shortcomings.^16–18 Both the mercury- and Brønsted-mediated reactions typically employ a large excess of the nitrile, which may not always be economically or synthetically feasible. Also, as highlighted in Figure 3, terpenes are notoriously prone to acid-catalysed rearrangements giving rise to various isomeric by-products.

5. Aza-Prins-type reaction in the Synthesis of Aristotelia Alkaloids

As an alternative to the Ritter-like reactions detailed above, aza Prins-type cyclizations have been used to synthesize the Aristotelia-like scaffold, which effectively switches the nucleophilc and electrophilic roles played by the two synthons. In the first instance of this approach, Borschberg et al. condensed enantioenriched (−)-terpinylamine (27) with indol-3-ylacetaldehyde resulting in imine 28, which in the presence of formic acid undergoes an aza-Prins cyclization to afford (−)-1 (Scheme 4).¹⁹ Notably, 27 can be accessed through the azidation of 19 with hydrazoic acid and BF₃•Et₂O and subsequent reduction. However, this approach results in racemization, presumably through the protonation of the cyclic olefin by the hydrazoic acid. To prevent this enantio-erosion, it was necessary to dibrominate the olefin of (−)-19 prior to azidation. Conveniently, treatment of the dibromoazide with LiAlH₄ results in azide reduction and reformation of the olefin to deliver (−)-27 in good enantiopurity (e.r. =93:7). Following this work, in a series of elegant publications, Borschberg also leveraged the aza-Prins cyclization to access a number of other Aristotelia alkaloids either by directly transforming (−)-1 or using pre-functionalized versions of 27.^20–25

Scheme 4.

Aza-prins cyclization to synthesize the Aristotelia alkaloids 1 and 5. The ability to access both enantiomers of 27 from chiral pool materials enables the enantioselective synthesis of both enantiomers of 5.

To circumvent the problems associated with the Ritter-like reactions, we recently applied a similar aza-Prins reaction to the synthesis of 5. In our approach, the conversion of (−)-19 to (−)-27 was accomplished with a nitro-meditated, Lewis acid-catalyzed azidation and subsequent reduction, which deliver (−)-27 with complete stereoretention.²⁶ Employing slightly modified conditions, (−)-27 and 4-quinolinecarboxaldehyde were reacted in the presence of TFA to yield (−)-5 in similar enantiopurity as the starting material (−)-19 (e.r.=95:5). Additionally, by accessing (+)-27 from (+)-19,²⁷ an identical approach provided access to (+)-5 in excellent enantiopurity (e.r.>99:1). Both (−)-5 and (+)-5 possessed identical ¹H and ¹³C NMR spectra to the compound isolated from A. chilensis, indicating we had synthesized the natural product and its enantiomer. Furthermore, an X-ray crystal structure of the hydrochloride salt of (−)-5 provided unambiguous proof of the relative configuration of the C9 stereocenter, which was previously misreported by Arias et. al.²⁸

6. Determination of Naturally Occurring Absolute Stereochemistry

Since its inception, one of the primary motivations for conducting natural product synthesis has been the elucidation and confirmation of chemical structure, particularly with regard to defining the relative and absolute stereochemical configuration of chiral compounds. Indeed, the first synthesis of (+)-2 by Lévy provided definitive proof of its absolute configuration by comparing the specific rotations of the synthetic and isolated natural products. When a stereoselective route to (−)-1 was developed, similar comparisons provided unambiguous proof that it possessed the same absolute configuration as (+)-2, despite their differences in specific rotations. These results confirmed previous studies that demonstrated both (−)-1 and (+)-2 could be converted to (+)-3, whose absolute configuration had been determined by X-ray crystallography.²⁹ Similarly, the extensive synthetic work by the Borschberg group has provided additional opportunities to confirm the absolute configuration of many of the Aristotelia alkaloids.

While the stereoselective synthesis and X-ray crystal structure of (−)-5 established the relative C9 configuration, the specific rotation of the natural product is not known. As such, it is impossible to definitely determine whether (−)-5 or (+)-5 is the naturally occurring enantiomer through simple comparisons of the optical activity. However, in lieu of optical activity, the bioactivity of 5 was reported. Both 3 and 5 isolated from A. chilensis were shown to be potent α3β4 nicotinic acetylcholine receptor (nAChR) antagonists (IC₅₀ = 0.40 μM and 0.96 μM, respectively). By comparison, (+)-5 possessed similar levels of activity (IC₅₀ = 0.89 μM) and was significantly more potent than (−)-5 (IC₅₀ = 3.4 μM). And while it is possible that the non-naturally occurring enantiomer of 5 is the more active enantiomer, the similarity in activity between the synthetic and isolated samples highly suggests that (+)-5 is the naturally occurring configuration.

If (+)-5 is the naturally occurring enantiomer, this raises an interesting question: why would a single alkaloid be produced that is of the opposite enantiomeric series as the other Aristotelia alkaloids? We propose the answer may lie in the biosynthetic pathway proposed in Scheme 5, which converts (−)-1 into (+)-5. Oxidation of the C3 position of the indole of (−)-1 reveals the imine 29 that can undergo intramolecular attack by the secondary amine to produce iminium 30. A [3,3]-sigmatropic rearrangement of 30 with the monoterpene core provides iminium 31. The reformation of the aryl imine and subsequent elimination of water furnishes the 4-quinoline of (+)-5. Remarkably, the sigmatropic rearrangement accomplishes the stereochemical inversion of the azabicyclic core without inverting the configuration of the two bridgehead carbons. Although efforts to validate this conversion of (−)-1 to (+)-5 experimentally are currently underway, it is worth noting that the transformation of an indole to quinoline alkaloid is reminiscent of the biosynthesis of the cinchona alkaloids.³⁰

Scheme 5.

Plausible mechanism for the biosynthetic conversion of (−)-1 to (+)-5.

7. Conclusions

In this Synpacts, we have highlighted several approaches that have been adopted to synthesize the Aristotelia alkaloids, namely the use of intramolecular Ritter-like and aza-Prins-like reactions. In each of these approaches, chiral pool monoterpenes have been employed in attempts to access enantioenriched products; however, in some cases these efforts were thwarted by the pseudosymmetry of the monoterpene building blocks. Fortunately, by bypassing these mechanisms for racemization, the enantioselective synthesis of numerous Aristotelia alkaloid has been reported, providing definitive proof of their relative and absolute configurations. Remarkably, despite being first identified >40 years ago, these intriguing alkaloids are likely to gather a renewed interest due to their recent identification as potent antagonists of the nAChRs, that have been implicated in numerous diseases and disorders of the central nervous system.³¹ The synthetic strategies discussed above will undoubtedly facilitate studies into this important bioactivity.

Biosketches

Andrew Riley (right) was born in Lawrence, KS, home of the University of Kansas. After earning his B.A. in chemistry from Ohio Wesleyan University in 2010, he returned to KU for his graduate studies in the laboratory of Prof. Thomas Prisinzano investigating the synthesis of natural product-derived probes of the central nervous. He then went on to conduct postdoctoral research with Prof. Paul Hergenrother at the University of Illinois at Urbana-Champaign as an NIH Postdoctoral Fellow. In 2018, he made the journey north to begin his independent career at the University of Illinois at Chicago. Working alongside talented scientists, including postdoctoral researcher Malaika Argade (left), Dr. Riley and his research group focus on the synthesis and medicinal chemistry of natural products with potential to treat pain and addiction.