Total Syntheses of the C19 Diterpenoid Alkaloids (-)-Talatisamine, (-)-Liljestrandisine, and (-)-Liljestrandinine by a Fragment Coupling Approach.

The C19 diterpenoid alkaloids (C19 DTAs) are a large family of natural products, many of which modulate the activity of ion channels in vivo and are therefore of interest for the study of neurological and cardiovascular diseases. The complex architectures of these molecules continue to challenge the state-of-the art in chemical synthesis, particularly with respect to efficient assembly of their polcyclic ring systems. Here, we report the total syntheses of (-)-talatisamine, (-)-liljestrandisine, and (-)-liljestrandinine, three aconitine-type C19 DTAs, using a fragment coupling strategy. Key to this approach is a 1,2-addition/semipinacol rearrangement sequence which efficiently joins two complex fragments and sets an all-carbon quaternary center.

The extracts of the Aconitum and Delphinium genera of plants have long been used in traditional medicine and as poisons for hunting and battle.[1] The diterpenoid alkaloids are a family of natural products associated with the toxicity of these flowering plants, and many of these compounds exhibit analgesic,[2,3] anti-inflammatory, antihypertensive, and antiarrhythmic[4] properties. (−)-Talatisamine (1, Figure a) is a representative aconitine-type C19 diterpenoid alkaloid (C19 DTA) that selectively blocks inwardly rectifying K+ ion channels over Na+ and Ca2+ ion channels in rat neurons,[5] and was found to attenuate neurocytotoxicity induced by β-amyloid oligomers.[6] Minor congeners, such as (−)-liljestrandisine (2) and (−)-liljestrandinine (3), which vary in the methylation and oxidation pattern, have also been isolated.[7,8]

Figure 1

Representative members of the aconitine-type C19 diterpenoid alkaloids and synthetic strategy considerations.

Biosynthetically, the C19 DTAs arise via the ent-atisane cyclase pathway, in which the originally formed denudatine framework (4), containing a bicyclo[2.2.2]octane CD ring system, undergoes skeletal rearrangement to the corresponding [3.2.1] CD ring system (5) (Figure b).[9] The cationic rearrangements involved in the biosynthesis of these natural products have also played an important role in many of the total syntheses of the C19 DTAs, including the pioneering efforts in the 1970s by Wiesner and co-workers,[10−13] and more recent contributions by the Sarpong,[14] Fukuyama,[15] and Inoue laboratories.[16] In each of these approaches, the denudatine-type bicyclo[2.2.2]octane is first assembled by a Diels–Alder reaction, and then rearranged at a late stage to the aconitine-type skeleton. Beyond these bioinspired strategies, a number of other approaches to the C19 DTAs have been disclosed,[17−28] as well as an elegant synthesis by Gin and co-workers of the C18 norditerpenoid alkaloid neofinaconitine (not shown).[29]

As part of a larger program aimed at developing fragment coupling strategies to prepare structurally complex diterpenes,[30] we became interested in the C19 DTAs as synthetic targets. We envisioned that scission of 1 through the C10–C11 and C7–C8 bonds of the central B ring would disconnect the highly bridged hexacyclic framework into two fragments of similar size and complexity (Figure b, inset). In the forward direction, the key steps of synthesis would focus on joining the AF ring system with an intact CD bicyclo[3.2.1]octane; we recognized that this approach would hinge on the development of an efficient method to forge the central C11 quaternary center. In this communication, we report the total syntheses of (−)-talatisamine (1), (−)-liljestrandisine (2), and (−)-liljestrandinine (3), which feature a 1,2-addition/semipinacol rearrangement as the key fragment coupling tactic. These efforts have also resulted in a correction to the original structure assignment of (−)-liljestrandisine (2).[8]

Given the conceptual strategy outlined in Figure , we developed the retrosynthesis of 1 shown in Figure . We sought to simplify 1 to amine 6 by disconnection of the C17–N and C7–C8 bonds; in the forward sense, we envisioned use of an N-centered radical cascade to close the E and B rings in a single transformation. Amine 6 was expected to be accessible from the corresponding ketone, 7, via a series of reductions and amination at C19. To join the AF and CD rings, we designed a two-step sequence involving (1) 1,2-addition of an organometallic reagent derived from alkenyl bromide 10 to epoxy ketone 9, and (2) semipinacol rearrangement of 8, in which the strain-release of the epoxide opening would serve as a thermodynamic driving force for formation of the hindered C11 quaternary center. Although the semipinacol rearrangement has been used in many total syntheses,[31] it is not typically employed as part of a fragment coupling strategy; we anticipated that this approach could highlight the utility of this overall two-step tactic for building polycyclic systems. Finally, epoxyketone 9 and alkenyl bromide 10 would be prepared from the simple starting materials cyclopent-2-en-1-one (11) and phenol (12), respectively. It was expected that this approach to (−)-1 would also provide access to (−)-(2) and (−)-(3) through adjustments to the final sequence of reduction steps.

Figure 2

Retrosynthetic analysis of (−)-talatisamine (1).

The synthesis of epoxy-ketone 9 began with the asymmetric Michael addition of dimethyl malonate (13) to 11 using the chiral gallium–sodium–BINOL catalyst ((S)-12) developed by Shibasaki and co-workers,[32−34] which furnished cyclopentanone 14 in 88% yield and 91% enantiomeric excess (ee) (Scheme a). The ketone of 14 was protected as the dioxolane to give 15, which was alkylated with 16 and subjected to HCl in refluxing acetone to yield hydrindenone 17 in 67% yield over two steps. The required epoxide was accessed by first converting 17 to the bromohydrin, which was isolated as a single diastereomer after trituration.[35] Treatment with triethylamine followed by recrystallization furnished epoxyketone 9 in 62% yield (two steps) and >99% ee.

Scheme 1

Enantioselective Synthesis of Epoxyketone 9 and Alkenyl Bromide 10

Reagent abbreviations: m-CPBA, meta-chloroperbenzoic acid; NMI, N-methylimidazole; ABNO, 9-azabicyclo[3.3.1]nonane N-oxyl radical; MeObpy, 4,4′-dimethoxy-2-2′-bipyridine.

The synthesis of the bicyclo[3.2.1]octadiene 10 commenced with a diastereoselective intramolecular meta-photocycloaddition of aryl ether 18 following the protocol reported by Sugimura (Scheme b).[36,37] Epoxidation of 19 and treatment with HCl resulted in Grob-type fragmentation to afford 20. Luche reduction of 20 afforded the diaxial 1,3-diol, which was protected as a siliconide (21). The chiral auxiliary was cleaved by oxidation of the secondary alcohol under conditions reported by Stahl and co-workers[38] followed by addition of K2CO3 and methanol to liberate alcohol 22. Upon oxidation to the ketone and conversion to enol triflate 23, alkenyl bromide 10 was generated using a Ni-catalyzed enol-triflate halogenation developed in our laboratory.[39]

Having prepared 9 and 10, we investigated the key fragment coupling step (Scheme ). To this end, alkenyl bromide 10 was submitted to lithium-halogen exchange and the corresponding alkenyllithium was added to epoxyketone 9 at −94 °C to give the 1,2-addition product which, after quenching with TMSCl, was isolated as silyl ether 8 in 77% yield as a single diastereomer. To our delight, treatment of 8 with catalytic TMSNTf2 (10 mol %) at −78 °C smoothly effected the desired semipinacol rearrangement. Under these conditions, ketone 7 is isolated in 97% yield on a 4 g scale. This remarkable two-step process forges the key C10–C11 bond and highlights the power of the 1,2-addition/semipinacol rearrangement as a fragment coupling tactic for complex polycycles. Deprotection of the TMS-ether enabled cyclization to form the strained lactone 24, thereby differentiating the two esters at C4. Conversion of the ketone to the enol triflate and Pd-catalyzed reduction afforded olefin 25 in excellent yield.

Scheme 2

Synthesis of Lactone 25 Using a 1,2-Addition/Semipinacol Rearrangement as the Key Fragment Coupling Tactic

Reagent abbreviations: TMSCl, trimethylsilyl chloride; TMSNTf2, N-(trimethylsilyl)bis(trifluoromethanesulfonyl)imide; DTBMP, 2,6-di-tert-butyl-4-methylpyridine; TCA, trichloroacetic acid; KHMDS, potassium bis(trimethylsilyl)amide.

At this stage, we turned our attention to formation of the E and B rings by the proposed N-centered radical cascade (Scheme a). Access to the aminyl radical precursor required selective aminolysis of lactone 25 with ethylamine to give amide 26. The secondary alcohol of 26 was methylated with trimethyloxonium tetrafluoroborate (Me3OBF4). Selective reduction of the C19-amide to the imine was achieved using the Ir-catalyzed hydrosilylation developed by Brookhart, and further reduction with NaBH(OAc)3 afforded amine 27 in excellent yield.[40] A three step sequence involving desilylation, chemoselective oxidation of the allylic alcohol, and protection of the remaining secondary alcohol as the triethylsilyl ether gave CD enone 28. Finally, treatment of 28 with N-chlorosuccinimide furnished the corresponding N-chloroamine (structure not shown). Unfortunately, efforts to effect the N-centered radical cascade under a number of conditions, including Bu3SnH/AIBN,[41−46] failed to give the desired product. Instead, under these conditions, the major product was 29, presumably resulting from a 1,5-hydrogen atom transfer at C6 followed by 1,4-addition to the enone.

Scheme 3

(a) Attempted N-Centered Radical Cascade Approach to (−)-1; (b) Completion of the Synthesis of (−)-Talatisamine (1) and Structural Revision of (−)-Liljestrandisine (41)

Reagent abbreviations: Proton Sponge, 1-8-bis(dimethylamino)naphthalene; NMI, N-methylimidazole; ABNO, 9-azabicyclo[3.3.1]nonane N-oxyl radical; MeObpy, 4,4′-dimethoxy-2-2′-bipyridine; TESOTf, triethylsilyl trifluoromethanesulfonate; DTBMP, 2,6-di-tert-butyl-4-methylpyridine; NCS, N-chlorosuccinimide; AIBN, 2,2′-azobis(2-methylpropionitrile); 1,3-DMBA, 1,3-dimethylbarbituric acid; MOMCl, chloromethyl methyl ether; TBAI, tetrabutylammonium iodide; LiHMDS, lithium bis(trimethylsilyl)amide; pyr, pyridine; RedAl, sodium bis(2-methoxyethoxy)aluminum hydride.

Having encountered challenges in forming the B and E rings through a cascade reaction, we pursued formation of the C–N and C–C bonds in a stepwise approach (Scheme b). To this end, aminolysis of lactone 25 with N-allylamine furnished amide 30 in 54% yield (64% BRSM). Reduction of the methyl ester followed by bis-methylation with Me3OBF4 delivered 32. The C19 amide was reduced to the amine using conditions analogous to those employed en route to 27; in this system, the Ir conditions proved uniquely effective at reducing the amide without also reducing the N-allyl group. The N-allyl substituent was cleaved under Pd-catalyzed conditions to give 33.[47] Cyclization to form the E-ring piperidine was achieved by intramolecular aziridination,[28] providing 34 in 74% yield. Treatment of aziridine 34 with acetyl bromide delivered alkyl bromide 35 in 83% yield. Notably, this transformation installed the C7 radical precursor and also introduced the two carbons of the N-ethyl substituent in the form of an acetamide. In analogy to 27, siliconide 35 was elaborated to 36 by desilylation, oxidation, and MOM-protection of the C14 alcohol. We were pleased to find that heating 36 with Bu3SnH and AIBN resulted in the desired radical cyclization,[29] closing the B ring and delivering hexacyclic ketone 37 in 99% yield.

To complete the syntheses of (−)-1 and (−)-2, hydrogenation of the strained C10–C12 alkene was followed by Mukaiyama dehydrogenation of the C16 ketone. Presumably, the strained bridgehead enone that initially forms undergoes oxy-conjugate addition upon addition of water and pyridine.[29,48] Treatment of 39 with Red-Al reduced the acetamide to the N-ethylamine and the ketone to the alcohol, delivering 40 with good selectivity for the axial diastereomer. Selective methylation of the C16 alcohol of 40 followed by MOM deprotection afforded (−)-talatisamine ((−)-1) in 77% yield over the final two steps. Alternatively, treatment of 40 directly with aqueous H2SO4 gave triol 41, and the 1H and 13C NMR data for which were consistent with that reported for (−)-liljestrandisine. Indeed, when the equatorial C16 epimer of 40 (not shown) was elaborated to compound 2 (Figure )—the originally proposed structure for (−)-liljestrandisine[8]—the 1H and 13C NMR did not match the literature data. On the basis of this synthetic work, we propose that the structure of (−)-liljestrandisine should be revised to 41.

The related target (−)-liljestrandinine (3) could be prepared from 38 by a slightly modified sequence. Dehydrogenation of 38 followed by addition of methanol and pyridine delivered C8 methyl ether 42 in 66% yield (Scheme ). Conversion of ketone 42 to the enol triflate followed by Pd-catalyzed reduction gave alkene 43. Amide reduction with lithium aluminum hydride followed by MOM deprotection and SN1 hydrolysis of the C8 methoxy group delivered (−)-3.

Scheme 4

Completion of the Synthesis of (−)-Liljestrandinine (3)

Reagent abbreviations: LiHMDS, lithium bis(trimethylsilyl)amide.

In conclusion, the C19 DTAs (−)-talatisamine, (−)-liljestrandisine, and (−)-liljestrandinine have been prepared in 31, 30, and 33 steps in the longest linear sequence, respectively, from phenol (12) (37, 36, and 39 steps total). Our synthetic approach leverages a 1,2-addition/semipinacol rearrangement sequence as a powerful tactic for the coupling of complex ring-containing fragments. Although efforts to use an N-centered radical cascade to simultaneously form the E and B rings were unsuccessful, the general bond constructions could be executed in a stepwise fashion by way of an intramolecular aziridination and subsequent radical cyclization. These studies highlight the 6-exo-trig cyclization of N-centered radicals as a prime area for future reaction development. Efforts to apply similar fragment coupling strategies in combination with radical cascade reactions to other DTAs are ongoing in our laboratory.