Enantioselective Total Syntheses of Akuammiline Alkaloids (+)-Strictamine, (−)-2(S)-Cathafoline, and (−)-Aspidophylline A

Abstract

The akuammiline alkaloids are a family of natural products that have been widely studied for decades. Although notable synthetic achievements have been made recently, akuammilines that possess a methanoquinolizidine core have evaded synthetic efforts. We report an asymmetric approach to these alkaloids, which has culminated in the first total syntheses of (−)-2(S)-cathafoline and the long-standing target (+)-strictamine. Moreover, the first enantioselective total synthesis of aspidophylline A is described.

The akuammiline alkaloids are a family of bioactive natural products that have been studied for over a century.¹ To date, over 30 akuammilines have been isolated, examples of which are shown in Figure 1 (1–6). These natural products can be divided into four structural subclasses, with completed total syntheses recently reported in three of these categories. Aspidophylline A (3), an example of the furoindoline-containing subclass, has been accessed synthetically by our group² and the laboratories of Zhu³ and Ma.⁴ In addition, our laboratory has completed the total synthesis of picrinine (6), a C5-oxidized akuammiline.⁵ With regard to the skeletally rearranged akuammilines, breakthroughs include total syntheses of vincorine (4) by Qin,⁶ Ma,⁷ and MacMillan,⁸ and total syntheses of scholarisine A (5) by Smith⁹ and Snyder.¹⁰

Figure 1.

Representative akuammiline alkaloids 1–6.

Despite these synthetic triumphs, one structural subclass of the akuammilines has remained inaccessible. Namely, compounds that possess a methanoquinolizidine core have not yet succumbed to total synthesis. We have targeted the total synthesis of two such alkaloids: strictamine (1) and 2(S)-cathafoline (2), isolated in 1966¹¹ and 2014,¹² respectively. As a salient feature of their caged methanoquinolizidine cores, each of these natural products also contains a densely functionalized cyclohexane ring that is part of two conjoined [3.3.1]-azabicycles. Strictamine (1) possesses an indolenine motif and four stereocenters, whereas 2(S)-cathafoline (2) bears an indoline unit and five contiguous stereocenters. The rich history, structural complexity, and bioactivity¹³ of 1 have prompted many synthetic efforts beginning with seminal studies by Dolby and Bosch and Bennasar.^14a–d More recent approaches by the laboratories of Sakai,^14e Cook,^14f Tokuyama,^14g Matsuo,^14h and Zhu¹⁴ⁱ have also been put forward. Herein, we describe an enantioselective approach to several akuammilines, including family members that contain the elusive methanoquinolizidine core. These efforts result in the first enantioselective total syntheses of aspidophylline A (3), strictamine (1), and 2(S)-cathafoline (2).

Our retrosynthetic analysis for the enantioselective total syntheses of strictamine (1) and 2(S)-cathafoline (2) is shown in Scheme 1. It was envisioned that both natural products could be derived from tetracyclic chloride 8. In the forward sense, late-stage deprotection and cyclization would forge the methanoquinolizidine core (see transition structure 7).¹⁵ In turn, chloride 8 would arise from methanolysis and subsequent chlorination of lactone 9. This key late-stage compound would be accessed from phenylhydrazine (11) and ketolactone 12 using an uncommon reductive variant of the interrupted Fischer indolization reaction.^16–18 The success of this step would lead to the introduction of two new rings and two stereogenic centers, including the C2 stereocenter seen in 2(S)-cathafoline (2) (via transition structure 10) and the challenging C7 quaternary center common to all akuammilines. Ketolactone 12, which would also be a viable intermediate toward aspidophylline A (3), would be prepared from enone 13, the product of a gold-mediated cyclization¹⁹ of enantioenriched silyl enol ether 14.

Scheme 1.

Our synthetic studies commenced with the asymmetric construction of the akuammiline [3.3.1]-azabicyclic core and the formation of enal 20 (Scheme 2). Beginning with dibenzoate 15, a Trost desymmetrization²⁰ was performed. In the event, dibenzoate 15 was treated with sulfonamide 16 in the presence of a suitable Pd precatalyst and (R, R)-DACH-phenyl Trost ligand. Direct saponification of the product furnished alcohol 17 in 89% yield. Subsequent oxidation delivered enone 18, a crystalline compound that was deemed suitable for X-ray analysis.²¹ By virtue of the heavy atom, it was found that the desired C3 stereocenter had been introduced in the Trost desymmetrization step. To attempt the key gold-mediated cyclization, ketone 18 was advanced to silyl enol ether 14, which in turn was subjected to a modification of Li’s cyclization conditions.¹⁹ This transformation smoothly delivered a 10:1 mixture of isomeric bicyclic adducts 13 (96% ee) and 19, even when performed on multigram scale. Although the bicycles were inseparable, exposure of the mixture to a previously established two step epoxidation/Wittig olefination sequence⁵ allowed enantioenriched enal 20 to be isolated in 49% yield from silyl enol ether 14.

Scheme 2.

With enal 20 in hand, our next task was to install a C7 alkyl substituent and access ketolactone 12 (Scheme 3). Toward this endeavor, enal 20 was first oxidized by NIS in the presence of potassium carbonate in methanol²² to provide enoate 21. Next, treatment of enoate 21 with vinyl ether 22 and NIS furnished an intermediate mixed acetal bearing an iodide,²³ thus setting the stage for a Ueno–Stork cyclization.²⁴ Exposure of this compound to radical cyclization conditions efficiently delivered C7-alkylated product 23 with diastereocontrol about the two newly formed stereocenters. From 23, the desired ketolactone 12 could be readily accessed via a sequence involving hydrolysis and reduction (23 → 24), followed by lactonization and alcohol oxidation (24 → 12).²

Scheme 3.

Having accessed ketolactone 12 in an enantioenriched form, we opted to pursue the asymmetric total synthesis of aspidophylline A (3) (Scheme 3). As suggested earlier, three prior total syntheses of 3 have been reported, all of which deliver racemic material.^2–4 Analogous to our earlier efforts,² albeit with a nosyl protecting group instead of a tosyl group, ketolactone 12 was exposed to phenylhydrazine (11) and TFA to provide Fischer indolization adduct 25. In the same pot, treatment of 25 with K₂CO₃ in methanol furnished furoindoline 27, presumably via transition structure 26. This transformation serves to introduce two new rings, four new bond linkages, and the challenging C7 quaternary stereocenter, all with complete diastereoselectivity. For reasons that shall soon become apparent, it should be noted that no evidence of epimerization at C16 is seen in this complexity-generating transformation. To complete the synthesis of (−)-aspidophylline A (3), interrupted Fischer indolization²⁵ adduct 27 underwent nosyl cleavage using a solid-supported thiol resin, followed by N-formylation.

We next directed our attention to the methanoquinolizidine alkaloids, which have evaded prior synthetic efforts.¹⁴ As shown in Scheme 4, Fischer indolization product 25 was accessed as demonstrated previously. However, this intermediate was directly intercepted by a hydride nucleophile. Interestingly, this process occurred in high yield and with complete control of stereochemistry to give the depicted 2S diastereomer of product 9. The overall conversion of 12 to 9 represents an uncommon reductive variant¹⁸ of the interrupted Fischer indolization reaction, and is also one of the most complex examples²⁶ of this venerable synthetic method. It proceeds by introducing two stereocenters (at C2 and C7) and establishes an intricate pentacyclic framework bearing five contiguous stereocenters.

Scheme 4.

From product 9, we pursued the seemingly simple conversion to hydroxyester 29 (Scheme 4). Although intermediate 25 readily underwent smooth ring-opening (see Scheme 3), we were disappointed to find that exposure of its reduced counterpart (9) to methoxide or two-step hydrolysis/methylation protocols failed to furnish 29. Instead, C16 epimer 28 was formed. Exhaustive efforts to overcome this roadblock were undertaken. For example, epimerization under a variety of acidic or basic conditions, attempted lactone opening via amidation,²⁷ Pd-catalyzed ring-opening,²⁸ or S_N2 displacements,²⁹ all proved fruitless. With our best efforts thwarted, we pursued a more stepwise approach. Reduction of 9, followed by selective silylation of the less hindered primary alcohol, proceeded without stereochemical erosion at C16 and gave alcohol 30. Lastly, an oxidation/esterification sequence (30 → 31), with in situ desilylation, provided a reliable means to synthesize the coveted late-stage intermediate 29. Of note, the Dess–Martin oxidation was achieved in the presence of the indoline, a conversion that was not possible when other oxidants were tried.

Having accessed alcohol 29, we set our sights on accessing the elusive methanoquinolizidine alkaloids (Scheme 5). Initial efforts to elaborate aminoalcohol 29 without using protecting groups were deemed challenging. However, a mesylation/chlorination sequence³⁰ was found to be viable and delivered aminochloride 8 in 77% yield. To our delight, chloride 8 could be elaborated to strictamine (1) and 2(S)-cathafoline (2). The former of the natural products was accessed via oxidation of the indoline with PCC, followed by denosylation. To access 2(S)-cathafoline (2), an N-methylation/deprotection sequence was performed. In both cases, the denosylation step proceeded with the desired in situ chloride displacement by the liberated amine to establish the hexacyclic methanoquinolizidine framework. NMR spectra of synthetic (+)-1 and (−)-2 were found to be in accord with spectra for the natural samples.

Scheme 5.

In summary, we have completed the first total syntheses of two akuammiline natural products that possess a methanoquinolizidine core. Our asymmetric approach to 1 and 2 features a gold-mediated cyclization to assemble the [3.3.1]-azabicyclic core of the natural products, a reductive interrupted Fischer indolization reaction to introduce the key C7 quaternary stereocenter and access late-stage compounds, and a series of carefully executed late-stage transformations to complete the total syntheses. Moreover, we have also completed the first enantioselective total synthesis of aspidophylline A (3). These studies constitute new achievements in the popular area of akuammiline alkaloid synthesis, and provide many lessons that should impact future endeavors in the synthesis of complex molecules.