Aza-[3 + 3] Annulations: A New Unified Strategy in Alkaloid Synthesis

1. Introduction

Cycloaddition and annulation reactions are among the most powerful methods in organic synthesis because of their ability to afford multiple bond formations with regio- and stereochemical control.^1,2 Accordingly, these reactions constitute highly convergent strategies for constructing complex carbocycles and heterocycles whether through a concerted, step-wise, or sequential process. Over the last twelve years, we have been developing a formal [3 + 3] cycloaddition³ or aza-annulation^4–9 that involves condensing vinylogous amides¹⁰ with α,β-unsaturated iminium salts [Scheme 1]. This annulation provides a unique approach to bond formation in the synthesis of piperidinyl heterocycles that represent a prevalent structural motif among a diverse array of biologically relevant natural products.^4–8,11

Scheme 1.

An Aza-[3 + 3] Annulation.

Mechanistically, as illustrated in Scheme 2, intermolecular reactions of vinylogous amides 1 with α,β-unsaturated iminium ions 2 involve a tandem Knoevenagel condensation/6π-electron electrocyclic ring-closure of 1-azatrienes 3^12,13 to give heterocycles 4.^14,15 The challenge in this and other related annulations is the regiochemical control,⁵ head-to-head versus head-to-tail [i.e., the Hickmott-Stille’s aza-annulation],¹⁶ which can be unpredictable and lead to complex mixtures.^4–6,17 However, the use of α,β-unsaturated iminium salts in our annulations has led to head-to-head regioselectivity in almost all reactions.⁵ For the intramolecular variant of this reaction,¹⁸ there would not be such regioselectivity issues. We were the first to report an intramolecular version of this aza-annulation reaction,⁵ that entails a different mechanistic sequence. Specifically, with vinylogous amides 5, our intramolecular aza-[3 + 3] annulation involves a tandem sequence of N-1,4-addition followed by C-1,2-addition/β-elimination to give quinolizidines 7. Both the inter- and intramolecular annulation pathways represent an attractive sequential or domino transformation¹⁹ and can be considered as a stepwise formal [3 + 3] cycloaddition in which two σ bonds are formed in addition to a new stereogenic center adjacent to the heteroatom.

Scheme 2.

A Mechanistic Overview.

After establishing the feasibility of the aza-[3 + 3] annulation, we developed a highly stereoselective variant of the intermolecular formal [3 + 3] cycloaddition using chiral vinylogous amides 8, leading to dihydropyridines 10 in high diastereoselectivity [Scheme 3].²⁰ A subsequent, detailed mechanistic study of this stereoselective annulation provided support to it being the first example^12,13 of stereoselective 6π-electron electrocyclic ring-closure of 1-azatrienes [see 9A].¹⁴ In an alternative approach toward a stereoselective aza-[3 + 3] annulation, we were able to achieve a torquoselective ring-closure of 1-azatrienes containing acyclic chirality at the C-terminus [see 13A] derived from α,β-unsaturated iminium ions 12 [Scheme 4], thereby representing an unexplored venue for controlling this pericyclic process through 1-azatrienes 13A that contain an acyclic stereochemical manifold [red arrow].²¹

Scheme 3.

Aza-Annulations with Chiral Vinylogous Amides.

Scheme 4.

Aza-Annulations with Chiral Iminium Ions.

Additionally, we have focused on identifying an asymmetric version of our aza-[3 + 3] annulation using chiral amine salts.²² After screening over 30 chiral amine salts, chiral amine 17 at 25 mol% could provide ~ 70% ee for the intramolecular aza-[3 + 3] annulation of vinylogous amide 16. Calculations indicated that Pro-R TS-18a is more stable relative to the corresponding diastereomeric Pro-S TS-18b by 1.41 kcal mol⁻¹. This difference can be attributed to the steric repulsion between the diphenylhydroxymethyl substituent on the pyrrolidine ring and the cyclohexenone moiety in Pro-S TS-18b. It was postulated that the asymmetric intramolecular formal aza-[3 + 3] cycloadditions promoted with C1-symmetrical amine catalysts, such as 17, proceed predominantly through Pro-R TS-18a giving rise to (R)-18a as the major enantiomer.

Another major methodological development in this area was the unveiling of an interesting variant of the carbo-[3 + 3] annulation²³ that competes with the aza-[3 + 3] annulation pathway. We observed that annulations of the piperidine-based exocyclic vinylogous urethanes 19 proceeded through an unexpected carbo-[3 + 3] annulation pathway,^3,24 leading to hexahydroquinolines 22 [Scheme 6]. Mechanistically, it likely involved a tandem C-1,4-addition/Mannich-type cyclization through respective intermediates 20 and 21. This finding invoked historical perspectives from Prelog’s work in 1949²⁵ and Stork’s enamine work in the 1950’s.²⁶ With the latter case, a distinct contrast could be generalized whereby annulations of α,β-unsaturated carbonyl systems with enamines mostly led to carbo-[3 + 3] annulation,²⁴ those with enaminones, or vinylogous amides, almost exclusively proceeded down the pathway of an aza-[3 + 3] annulation.

Scheme 6.

An Unexpected Carbo-[3 + 3] Annulation.

Given the distinct six-membered nitrogen heterocyclic motif of many structurally diverse and exciting alkaloids, we recognized that our aza-[3 + 3] annulation can be a powerful synthetic strategy in alkaloid synthesis, and that it represents both an attractive and complimentary approach to aza-[4 + 2] cycloadditions in constructing piperidines [Figure 1]. While aza-[4 + 2] cycloadditions remain a versatile synthetic method, aza-dienes and imines employed for aza-[4 + 2] cycloadditions are not always the most accessible or the easiest substrates to handle given the problems of isomerization and hydrolysis.^1,27 Conversely, our aza-annulation utilizes more readily accessible and easily handled vinylogous amides and enals. In addition, our aza-[3 + 3] annulation can be biosynthetic in nature because it invokes Robinson’s double Mannich-type process,^28–30 which can be categorized as Type-I aza-[3 + 3] annulation with ours being Type-II, according to Harrity’s classifications.⁷

Figure 1.

Aza-[3 + 3] Annulation: A Complementary to Aza-[4 + 2].

Despite such potential significance, the synthetic scope of this type of aza-[3 + 3] annulation has been insufficiently explored until the last 15 years. An impressive array of aza-[3 + 3] annulations, including earlier work, had been accounted for in a collection of reviews in 2005.^4–8 Summarized in Figure 2 are selected examples of total syntheses and methodological developments reported recently,³¹ and notably, these examples also include elegant aza-[3 + 3] concepts that are not directly related to our aza-annulation. In our own efforts, we have been focusing on total syntheses of alkaloids with the most significant development in the last six years being the evolution of the aza-[3 + 3] annulation into an attractive, unified strategy for alkaloid synthesis. Our completed total syntheses reflect both the competitive nature of this annulation strategy in the overall synthetic length and its flexibility in accessing structural and stereochemical diversity, and that it has positioned us with a unique perspective in the alkaloid community. Without detracting from known elegant approaches in constructing alkaloids, whether or not documented in this review, we highlight here our complete total synthesis efforts to unequivocally demonstrate that this aza-[3 + 3] annulation can serve as a highly useful and powerful strategy in organic synthesis [Figure 3].

Figure 2.

Recent Chemistry Related to Aza-[3 + 3] Annulations.

Figure 3.

A Unified Strategy for Alkaloid Synthesis: Completed and Almost Completed Alkaloids to Date.

2. An Intermolecular Aza-[3 + 3] Annulation

2.1. Aza-spirocycles: Synthesis of 2-epi-(±)-perhydrohistrionicotoxin

Aza-spirocycle (−)-histrionicotoxin was isolated from the skin extracts of neotropical tree frog Dendrobates histrionicus.³² In addition to its unique aza-spirocenter, (−)-histrionicotoxin and its saturated derivatives have shown potency as noncompetitive blockers of nicotinic receptor-gated channels,³³ thereby making them attractive and popular targets for synthetic chemists.^34,35 As illustrated in Scheme 7, our synthetic efforts featured a highly stereoselective aza-[3 + 3] annulation of α,β-unsaturated iminium salt 26 and N-benzyl aminopyrone 25 to generate the requisite aza-spirocenter 24. An unprecedented decarboxylation of the 2-pyrone ring in 24 followed by hydrogenation afforded the n-amyl side chain of 2-epi-(±)-perhydrohistrionicotoxin.³⁶

Scheme 7.

Retrosynthetic Analysis of 23.

We began our total synthesis of the non-naturally occurring 2-epi-(±)-perhydrohistrionicotoxin by constructing cyclohexylidene α,β-unsaturated iminium salt 26 in seven steps from 2-cyclohexenone [Scheme 8]. To this end, an epoxidation of 2-cyclohexenone 27 followed by alkylation with n-butyllithium and subsequent TBS protection afforded ketone 28. At this stage a Horner-Wadsworth-Emmons reaction proved to be unsuccessful, therefore we utilized a Peterson olefination protocol of 28, thereby affording the desired homologation. Next, a DIBAL-H reduction followed by Dess-Martin periodinane [DMP] oxidation afforded enal 29, which upon treatment with equal amounts of piperidine and acetic anhydride heating at 85 °C for 3 hours gave cyclohexylidene α,β-unsaturated iminium salt 26. At this point, the vinyl iminium salt 26 was poised to undergo an intermolecular aza-[3 + 3] annulation with N-benzyl aminopyrone 25. Thus, 26 and vinylogous amide 25 were heated at 150 °C to afford aza-spirocycle 24 in 64% yield as a single diastereomer unambiguously assigned using nOe experiments. Next, a quantitative hydrogenation of 24 afforded 2-pyrone 30 [Scheme 9]. By this approach, the three contiguous stereocenters of perhydrohistrionicotoxin were established in a very short sequence, thereby highlighting this novel approach to aza-spiroundecane ring systems with high diasteromeric control at the aza-spirocenter.

Scheme 8.

Aza-[3 + 3] of 25 and 26.

Scheme 9.

Synthesis of 2-epi-(±)-perhydrohistrionicotoxin.

At this stage, decarboxylations of 30 proved to be a formidable challenge, as a variety of methods were inadequate. We found a solution by employing a unique LAH-mediated decarboxylation protocol, followed by treatment of the crude reaction mixture to 60 psi of H₂ in the presence of Pd-C to afford 32 with the desired n-amyl side chain in 60% overall yield. Acid-mediated desilylation and subsequent debenzylation using Pearlman’s catalyst generated 2-epi-(±)-perhydrohistrionicotoxin 23 in 90% yield overall. This completed an eleven-step total synthesis of 2-epi-(±)-perhydrohistrionicotoxin in 21% overall yield and revealed the pivotal aza-[3 + 3] annulation to be a unique strategy towards aza-spirocycles.

2.2. Cis-decahydroquinoline Alkaloids

2.2.a. Synthesis of (−)-4a,8a-diepi-pumiliotoxin C

In the late 1960s, pumiliotoxin C was isolated from the skin secretions of the Central American frog species Dendrobates pumilio.³⁷ Possessing an interesting biological profile, pumiliotoxin C has been proven to act as a noncompetitive blocker of acetylcholine receptor channels. Additionally, it inhibits the specific binding of [³H]BTX-B to brain membranes with an IC₅₀ value at 45 µM.³⁸ Due to its synthetic challenges, notably the cis-fused aza-decalin core with three contiguous stereocenters, pumiliotoxin C has been an attractive synthetic target.^39,40 Retrosynthetically, we envisioned (−)-pumiliotoxin C to arise from an olefination/hydrogenation sequence of cis-decahydroquinoline 34, which would be obtained via hydrogenation of the C4a-C8a ring fusion in 35 [Scheme 10].¹⁴ Reductive removal of the chiral auxiliary in cycloadduct 36 would afford 35. Lastly, dihydropyridine 36 would be obtained via a stereoselective aza-[3 + 3] annulation of chiral vinylogous amide 38 and α,β-unsaturated iminium salt 37.

Scheme 10.

Retrosynthetic Analysis of (−)-Pumiliotoxin C.

Our efforts toward (−)-pumiliotoxin C commenced with reacting chiral vinylogous amide 38 with a pre-formed α,β-unsaturated piperidinium acetate salt 37 to afford aza-[3 + 3] annulation product 36 in 77% yield with high diastereoselectivity [Scheme 11]. Subsequent hydrogenation of the endocyclic olefin provided chiral vinylogous amide 39 in 94% yield. At this stage, olefination of the carbonyl was then attempted; however, Tebbe’s olefination of 39 failed to give any desired product. Instead, cleavage of the TBS group in 39, followed by hydrogenation using Pd(OH)₂ and NH₄OCHO provided the free amide 40.

Scheme 11.

Stereoselective Aza-[3 + 3] of 37 and 38.

With vinylogous amide 40 in hand, direct hydrogenation of the olefin employing a variety of conditions failed. We speculated that the nitrogen atom required a protecting group to facilitate the hydrogenation. Therefore, protection of the free vinylogous amide in 40 using TFAA, followed by high pressure hydrogenation, afforded a mixture of alcohol 41 and ketone 42 [Scheme 12]. A Dess-Martin periodinane oxidation of the mixture then afforded ketone 42 in 68% yield over two operations as a single diastereomer. The relative stereochemistry based on nOe experiments revealed an anti relationship between the C2 and C8a hydrogens. We postulated that the conformation of 40-N-TFA is favored (Spartan AM1 calculations: a minimum of 2.01 kcal mol⁻¹ over other conformers) as it relieves much of the pseudo A^1,2 strain between the N-trifluoroacyl group and the n-propyl group at C2, thereby resulting in delivery of hydrogen from the bottom face. While this hydrogenation did not afford the desired syn C2-C8a relationship required for (−)-pumiliotoxin C, this finding represented a rare and highly stereoselective entry to the anti relative stereochemistry at C2 and C8a of cis-1-azadecalins.

Scheme 12.

Hydrogenation of 40-N-TFA.

Ketone 42 was then subjected to a Peterson olefination protocol, since various Wittig-type olefination conditions proved to be inadequate [Scheme 13]. An ensuing acid-mediated elimination and subsequent hydrogenation gave 43 in 43% yield over three operations. We found that the stereoselectivity of the hydrogenation of the exocyclic olefin at C5 ranged from 8:2 to 9:1 in favor of the reduction taking place at the convex face of the cis-1-azadecalin. Notably, the major isomer was later assigned based on nOe experiments of 45. Reductive removal of the trifluoracetyl group in 43 using NaBH₄ resulted in a mixture of the non-naturally occurring alkaloids 4a,8a-diepi-(−)-pumiliotoxin C 45 and 2-epi-(+)-pumiliotoxin C 44 as a 8:2 to 9:1 ratio in 25% overall yield. The low isolated yields of 44 and 45 were attributed to the slow rate of removing the TFA group and the reaction not being driven to completion.

Scheme 13.

Synthesis of 4a,8a-diepi-(−)-pumiliotoxin C.

2.2.b. Synthesis of (+)-Lepadin F and (+)-Lepadin G

The lepadin family, comprised of eight cis-decahydroquinoline alkaloids, was isolated from various sources such as Clavelina lepadinformis,^41a flatworm Prostheceraeus villatus,^41b tropical marine tunicate Didemnum sp.,^41c and Australian great barrier reef ascidian Aplidium tabascum [Scheme 14].^41d Their biological activity profiles include tyrosine kinase inhibition, cytotoxicity, antiplasmodial and antitrypanosomal properties as well as antimalarial properties,^41a-d thereby attracting attention from the synthetic community.^42–49

Scheme 14.

The Lepadin Family of Alkaloids.

The stereochemical relationships of the lepadins can be categorized into three subsets, as illustrated in Scheme 15. The most challenging aspect would be the 1,3-stereochemical relationship at C2 and C8a, which can be syn as in lepadins A-E and H, or anti as in F and G. We envisioned that each subset could arise in a stereodivergent manner from a common intermediate 46, which can be accessed via an intermolecular aza-[3 + 3] annulation of chiral vinylogous amide 38 and enal 49. While iminium ion chemistry has served us well, one drawback here was the solubility of iminium salt. Alternatively, we later developed a more operationally simple means of accessing quinolines by a TiCl₄-initiated aza-annulation. We found that while the initial product afforded by a Lewis acid-mediated aza-annulation as a 51:49 mixture of 46 and 2-epi-46, it could be thermally equilibrated to pure 46 through a sequence of pericyclic aza-ring opening and ring closure.⁵⁰

Scheme 15.

A Stereodivergent Approach to the Lepadins.

Retrosynthetically, (+)-lepadin F 48 was envisioned to materialize from decahydroquinoline 49 through a side chain installation via Julia-Kocienski olefination and an esterification utilizing Yamaguchi’s protocol [Scheme 16].⁵¹ Homologation of vinylogous amide 46 would afford an α,β-unsaturated ester, which in turn would be hydrogenated to give cis-decahydroquinoline 49. Lastly, dihydropyridine 46 would be derived from a highly diastereoselective aza-[3 + 3] of chiral vinylogous amide 38 with α,β-unsaturated iminium species 50.

Scheme 16.

Retrosynthetic Analysis of (+)-Lepadin F.

We commenced our synthesis of (+)-lepadin F by submitting vinylogous amide 38 and an α,β-unsaturated iminium salt 50 to aza-[3 + 3] annulation conditions affording core dihydropyridine 46 in 70% yield with 96:4 diastereoselectivity [Scheme 17]. Osmium tetroxide dihydroxylation of the C3-4 olefin afforded diol 51 in 60–70% yield, and subsequent reductive removal of the C4-OH group in the presence of excess TFA and triethylsilane was performed. The free alcohol was then acylated to give vinylogous amide 52 in 90% yield over two operations.

Scheme 17.

Aza-[3 + 3] Annulation of 28 and 50.

Our next synthetic challenge was seen during hydrogenation of the endocyclic olefin, for over-reduction of the C5-carbonyl group in 52 was observed in a variety of protocols. We found a solution by homologating acetate 52 via a three-step sequence featuring Eschenmoser’s episulfide contraction,affording α,β-unsaturated ester 53 exclusively as the E-isomer [Scheme 18]. A double hydrogenation was then achieved to afford 49 in 91% yield as a 5:1 mixture of separable diastereomers with respect to the C5 stereochemistry. It is noteworthy that the chiral auxiliary plays two roles: (1) in the establishment of the C2 stereochemistry during the aza-[3 + 3] annulation step, and (2) as a steric blocking group setting up three stereocenters in the double hydrogenation of 53. The chiral auxiliary in 49 was reductively removed, affording a free vinylogous amide which was immediately Boc-protected in situ. Subsequent deacylation afforded free alcohol 54 in 77% yield over two operations.

Scheme 18.

Double Hydrogenation of Ester 53.

At this stage, we were ready to complete the total synthesis of (+)-lepadin F by installing the side chains. To this end, a two-step sequence of Dess-Martin periodinane oxidation and NaBH₄ reduction was performed to invert the C3-alcohol to its desired stereochemistry [Scheme 19]. This resultant alcohol was then silylated to give silyl ether 55 in 91% yield over three operations. A DIBAL-H reduction of the ester in 55, and a subsequent DMP oxidation revealed an aldehyde 56 which was poised to undergo a Julia-Kocienski olefination. Thus, utilizing sulfone (S)-57, installation of the alkyl side chain to core 56 afforded alkene 58 in 90% yield. Next, the alkene was hydrogenated over Pd/C and TBAF-mediated desilylation gave alcohol 59 in nearly quantitative yield over two steps. We then performed an esterification employing Yamaguchi’s protocol and (E)-oct-2-enoic acid, and subsequent global deprotection afforded (+)-lepadin F in twenty steps with 15.2% overall yield from chiral vinylogous amide 38.

Scheme 19.

Total Synthesis of (+)-Lepadin F.

Our spectroscopic data for (+)-lepadin F matched those reported by Carroll and coworkers⁴¹ for the natural (+)-lepadin F and Blechert’s synthetic sample,⁴⁸ thereby allowing us to claim a completed total synthesis. Yet we believed there to be a high margin of error in determining the C5’ stereochemistry, since not only was the C5’ stereocenter never defined in the isolation report, the C5’ stereocenter is acyclic and highly insulated on the side chain. Consequently, we synthesized the C5’ epimer of (+)-lepadin F commencing with advanced intermediate aldehyde 56 and sulfone (R)-57 in the requisite Kocienski-modified Julia olefination [Scheme 20].⁵² Spectroscopic comparisons of both ¹H and ¹³C NMR datasets of our synthetic (+)-lepadin F and (+)-5’-epi-lepadin F with Carroll’s natural (+)-lepadin F enabled us to confirm the correct relative stereochemistry at C5’ in (+)-lepadin F as S.

Scheme 20.

Total Synthesis of (+)-5’-epi-lepadin F.

Considering that two complex structures differing only at a remote and highly insulated stereocenter could still be differentiated spectroscopically, we were prompted to synthesize (+)-lepadin G and its C5’ epimer in an attempt to concisely determine its correct relative stereochemistry. Therefore, total syntheses of both (+)-lepadin G and (+)-5’-epi-lepadin G were performed commencing with ent-38.⁵² Ent-55 was quickly accessed by our intermolecular aza-[3 + 3] annulation of ent-38 with the appropriate vinyl iminium salt and further synthetic manipulations in 40% overall yield [Scheme 21].⁵² Ent-55 was then transformed to ent-59 (or its C5’ epimer in the case of (S)-57) in 43% yield over thirteen steps.

Scheme 21.

Synthesis of (+)-Lepadin G and C5’ Epimer.

The synthetic sequence differed from that of (+)-lepadin F in the second-to-last step, whereby installation of a different ester side chain, namely (2E,4E)-octadienoic acid, under Yamaguchi conditions afforded 60. An ensuing global deprotection afforded both (+)-lepadin G 61 and (+)-5’-epi-lepadin G 62. As before, spectroscopic comparisons of the NMR datasets of our synthetic (+)-lepadin G and (+)-5’-epi-lepadin G with Carroll’s natural (+)-lepadin G sample suggested that the correct relative stereochemistry at C5’ should be R for (+)-lepadin G.

3. An Intramolecular Aza-[3 + 3] Annulation

Over the last decade, we have been developing an aza-[3 + 3] annulation reaction as a general strategy in alkaloid synthesis.^{4,14,15,18,20–23,52–58} The intramolecular variant of this reaction has proven to be valuable in natural product synthesis. Specifically, the intramolecular aza-[3 + 3] annulation of vinylogous amides tethered to a vinyl iminium motif 63 proceed through a tandem sequence of N-1,4-addition and C-1,2-addition/β-elimination that can lead to a variety of nitrogen heterocycles [Scheme 22]. As will be discussed, our endeavors into the methodological development of this annulation and concurrent total syntheses have confirmed the pivotal intramolecular aza-[3 + 3] annulation to be advantageous in constructing a diverse array of alkaloid natural products 64a-64f.^6a,55–58

Scheme 22.

An Intramolecular Aza-[3 + 3] Annulation.

3.1. Piperidinyl Heterocycles: Synthesis of (+)-Gephyrotoxin

We recognized an opportunity to showcase our intramolecular annulation of a vinylogous amide tethered to an α,β-unsaturated iminium salt in the formal total synthesis of (+)-gephyrotoxin 65.^18b As illustrated in Scheme 23, we noticed that of the three known total syntheses^59,60 of (±)-gephyrotoxin,⁶¹ Kishi’s intermediate 66, a tricyclic heterocycle, could be easily accessed by our intramolecular aza-[3 + 3] annulation and would provide a successful employment of this diastereoselective reaction. Our plan was to intercept aza-annulation substrate 67 in eight steps from ethyl acetoacetate via condensation of 68 with 1,3-cyclohexanedione.

Scheme 23.

Retrosynthetic Analysis of (+)-Gephyrotoxin.

To complete the formal total synthesis, chiral vinylogous amide 69 (or 70) was submitted to aza-[3 + 3] annulation conditions by heating the substrate with piperidinium acetate in an ethyl acetate/ethanol solvent mixture. After two hours, we found that the major isomer 73-β of our annulation, after hydrogenation, was not Kishi’s intermediate, but was the opposite stereoisomer [Scheme 24]. We suspected that the unprotected primary alcohol was distorting the conformation of the transition state, and therefore we explored various silyl protecting groups to provide a substrate that would undergo the annulation with the desired stereochemical outcome.

Scheme 24.

Formal Synthesis of (+)-Gephyrotoxin.

To that end we found that the reaction of tert-butyldiphenylsilyl (TBDPS)-protected alcohol 70 was the most amenable to the desired outcome, providing the diastereomeric cycloadducts after hydrogenation in a ratio of 60:40 for 73-α/73-β. Subsequent desilylation and chromatographic separation of the two diastereomers led to the isolation of tricyclic compound 74-α, which matched Kishi’s intermediate by NMR and optical rotation. Of importance, we observed that the tricyclic cycloadducts from our aza-annulations were less stable than anticipated, and to improve their stability, the crude reaction mixture containing the cycloadduct would be directly submitted to hydrogenation conditions to reduce the endocyclic double bond. This prudent, one-pot protocol provided 73, and would prove to be an important modus operandi for our subsequent endeavors in synthesizing other aza-heterocycles. The overall sequence to Kishi’s intermediate proved to be concise and demonstrated the synthetic feasibility of this intramolecular formal [3 + 3] cycloaddition reaction in the synthesis of natural products.

3.2. Quinolizidine Heterocycles: Synthesis of (±)-2-Deoxylasubine II

Despite success over the last decade, when using a vinylogous urethane tethered to a vinyl iminium salt 77 [Scheme 25], our annulation strategy had not been advantageous in constructing quinolizidine structural motifs, only providing low yields. This problem was rectified once we found that the role of the counteranion [X⁻] in these iminium salts provided a vital function in the annulation.²² While iminium salts with acetate as the counteranion were wholly inadequate, employing the more reactive trifluoroacetate salt proved to be the solution, thereby rendering a vinylogous urethane deployable in the aza-[3 + 3] annulation Furthermore, we were cognizant of the great potential in exploiting vinylogous urethanes in our formal cycloaddition since the resultant methoxy carbonyl group provides an excellent functional handle, amenable to further chemical transformations or removal.

Scheme 25.

Retrosynthetic Analysis of 2-Deoxylasubine II.

(±)-2-Deoxylasubine II 75^62–67 became our first successful entry in constructing a member of the quinolizidine family of alkaloids by our intramolecular annulation.⁵⁸ Retrosynthetically, the requisite quinolizidine nucleus 76 would be generated by our pivotal aza-annulation of substrate 77 [Scheme 25]. We elected to access vinylogous urethane 77 by amine 78, which could be constructed from propargyl alcohol 79. A concise end-game strategy would feature a Barton decarboxylation at C3 of vinylogous urethane 76.

Our efforts commenced with amine 78 prepared in five steps from propargyl alcohol 79 [Scheme 26]. Reaction of amine 78 with alkynoate 80 led to vinylogous urethane 81. The stereochemistry of the vinylogous urethane 81 was determined to be exclusively Z, likely due to favorable internal hydrogen bonding.

Scheme 26.

Synthesis of Annulation Substrate 82.

At this point, a two-step sequence of TBAF-mediated desilylation and subsequent oxidation gave the aza-annulation substrate 82 in 86% yield with a Z/E ratio of 3.3:1 for the enal. With the vinylogous urethane tethered to the enal 82 in hand, the more reactive piperidinium trifluoracetate salt was exploited to trigger the annulation, and in situ hydrogenation allowed isolation of the annulation product 76 in 62% yield over two operations [Scheme 27]. To complete the total synthesis, hydrogenation of cycloadduct 76 using Adams’ catalyst gave ester 83 in 95% yield as a single diastereomer. Lithium iodide-promoted demethylation of ester 83, followed by Barton’s standard protocol for decarboxylation of the resultant acid, afford (±)-2-deoxylasubine II 75 in 33% overall yield for the sequence.

Scheme 27.

Synthesis of (±)-2-deoxylasubine II.

The key feature of this investigation was the successful construction of a member of the quinolizidine family of alkaloids by using the more reactive piperidinium trifluoroacetate salt to effectively allow a vinylogous urethane to be employed in the aza-[3 + 3] annulation.

3.3. A Unified Strategy towards Indoloquinolizidine Heterocycles

An application of the intramolecular aza-[3 + 3] annulation as a unified strategy was illustrated by our syntheses of tangutorine^55a and deplancheine^55b [Scheme 28]. Tangutorine⁶⁸ 84 possesses a novel benz[f]indolo[2,3-a]quinolizidine skeleton and is related to the well-known monoterpenoid (+)-indole alkaloids such as (+)-deplancheine 91,⁶⁹ (+)-geissoschizine,⁷⁰ yohimbine and reserpine.^71–73 Natural products possessing the indoloquinolizidine substructure are prevalent among alkaloids that are derived biosynthetically from tryptophan. Syntheses^74–78 of these monoterpenoid indole alkaloids have often featured the classic Pictet-Spengler cyclization.^{72,73,79–81}

Scheme 28.

Retrosynthetic Analysis of Tangutorine and Deplancheine.

As outlined in Scheme 28, both tangutorine and deplancheine can be accessed from a similar indoloquinolizidine core, in which the indoloquinolizidine CD-ring 85 (or 92) should be attainable via the pivotal intramolecular aza-[3 + 3] annulation, depending upon the vinylogous amide (86 or 93) requisite for the core structure. Condensation of the free amine 88 with 1,3-diketone 87 for tangutorine, or reaction with 4-methoxy-3-buten-2-one 94 for deplancheine would provide the requisite substrates for the intramolecular aza-[3 + 3] annulation. Amino alcohol 88 would be prepared via a Heck cross-coupling of a protected 2-bromotryptamine 90 with an appropriate 3-carbon synthon. Notably, implementing a Heck coupling to construct the C2-C3 bond was used in place of the frequently employed Pictet-Spengler cyclization. The preparation of 90 would commence with tryptamine 89.

3.3.a. Synthesis of (±)-Tangutorine

To prepare aza-annulation substrate 96, tryptamine 89 was di-protected and brominated so that an ensuing Heck coupling would provide ester 95 in 82% yield [Scheme 29]. Further synthetic manipulations, including condensation with 1,3-cyclohexanedione 87 gave a vinylogous amide and oxidation with MnO₂ revealed the enal, thus giving the requisite intramolecular substrate 96. Under standard intramolecular aza-[3 + 3] annulation conditions followed by hydrogenation of the endocyclic olefin, pentacycle 85 was isolated in 56% yield. With this finding, a practical synthetic approach toward tangutorine and other tryptophan-derived monoterpenoid indole alkaloids was established, and furthermore, the study provided the first example of an aza-[3 + 3] annulation where the indole motif was present.

Scheme 29.

Synthesis of Pentacycle 85 via Aza-[3 + 3].

To complete the total synthesis of (±)-tangutorine 84 [Scheme 30], pentacycle 85 was transformed in three operations to keto-pentacycle 97 as a single diastereomer in 61% yield. To append a carbonyl group at C18 of the E-ring, we found that the LHMDS/HMPA/Mander’s reagent protocol was inadequate. Instead, pentacycle 97 was refluxed in THF in the presence of NaH and diethyl carbonate to yield β-ketoester 98 regioselectively as a mixture of diastereomers. Further synthetic manipulations were performed to afford the allylic alcohol, thereby concluding the synthesis of (±)-tangutorine 84.

Scheme 30.

Synthesis of (±)-Tangutorine.

3.3.b. Synthesis of (±)-Deplancheine

As illustrated in Scheme 31, the synthesis of deplancheine 91 began with an identical approach as tangutorine, with the divergent point being reaction of the free amine with 4-methoxy-3-buten-2-one 94.

Scheme 31.

Synthesis of Vinylogous Amide 99.

Thus, phthalimido ester 95 was obtained from tryptamine 89 in four steps, including Fukuyama’s Heck coupling protocol [Scheme 31]. Of note, It was determined that we could also employ Fu and Littke’s^82,83 conditions of excess methyl acrylate and a 1 : 1 ratio of Pd catalyst and a bulky t-Bu₃P ligand for the Heck coupling. To deprotect ester 95, we exploited Ganem’s reductive sequence, whereby slightly more than three equivalents of DIBAL-H cleanly afforded a hydroxyaminal (not shown). Further reduction with NaBH₄ provided an amide intermediate, which upon adding HOAc to the reaction mixture led to free amino alcohol 88. The formation of vinylogous amide 99 was then effected, using 4-methoxy-3-buten-2-one 94, giving an overall yield of 26% for the three operations. Notably, vinylogous amide 99 is distinctly manifested as the Z-conformer based on ¹H NMR coupling constants in addition to a well-defined peak for the NH suggesting a hydrogen bond between the NH and the carbonyl. After MnO₂ oxidation of 99, the strategic aza-[3 + 3] annulation of vinylogous amide 100 was investigated [Scheme 32].

Scheme 32.

Synthesis of (±)-Deplancheine.

As in our synthesis of (±)-tangutorine (vide supra), we employed piperidinium acetate salt to afford indoloquinolizidine tetracycle 92 after hydrogenation in 35% yield over three steps. Deprotection of 92 with TFA and CH₂Cl₂ in a 1:1 mixture followed by reduction with NaBH₄ completed the total synthesis of (±)-deplancheine 91, and gave a minor isomer which was believed to be the Z-isomer of (±)-deplancheine.

3.3.c. Synthesis of S-(−)-Deplancheine and R-(+)-Deplancheine

To accomplish the total syntheses of S-(−)-91 and R-(+)-91, vinylogous amide 100 was subjected to an enantioselective aza-[3 + 3] annulation²² using chiral amine salts S-101 or R-101 [Scheme 33]. After hydrogenation, indoloquinolizidine S-92 was found in 46% yield over three steps with an er of 80:20 in favor of the S-enantiomer. When employing R-101 as the chiral amine salt, R-92 was isolated in 15% yield over three operations with an HPLC ratio of 69:31 in favor of R-92. Similar chemical modifications, as in the racemic synthesis, were then performed independently to provide S-(−)-deplancheine in 42% overall yield and R-(+)-deplancheine in 83% overall yield. The syntheses of tangutorine 84 and deplancheine 91 confirmed the aza-[3 + 3] annulation to be a practical synthetic approach in constructing tryptophan-derived monoterpenoid indole alkaloids.

Scheme 33.

Synthesis of S- and R-Deplancheine.

3.4. Tricyclic Marine Alkaloids: Cylindricines and Lepadiformines

In the early 1990s, Blackman reported the isolation of (−)-cylindricines A-K^84c-e from the marine ascidian Clavelina cylindrica collected in Tasmania. Additionally, two structurally related alkaloids, lepadiformine and fasicularin, were isolated from the marine ascidian Clavelina lepadiformis^84f and Nephtesis fasicularis,^84g respectively. Given the unique tricyclic structural motif, low natural abundance, and biological activity, an impressive collection of synthetic efforts towards these alkaloids have been published.^84–91 We envisioned an application of the intramolecular aza-[3 + 3] annulation as a unified strategy in the total synthesis of certain tricyclic marine alkaloids, specifically the cylindricine and lepadiformine alkaloid families [Figure 4].84a,b,h,i

Figure 4.

Cylindricines and Lepadiformines.

As will be discussed, our pivotal intramolecular aza-[3 + 3] annulation was remarkably efficient in providing the tricyclic scaffold of (−)-cylindricine C 101 and that of putative (−)-lepadiformine 106 (or: (−)-4-deoxo-2-epi-cylindricine C),^56b,84 yet this route could not be implemented in our synthesis of (−)-lepadiformine 102. To reconcile this, we devised a unified strategy to synthesize (−)-lepadiformine and (+)-cylindricines C-E (103–105) that would utilize an N-acyliminium ion/diene cyclization and a Wharton’s rearrangement to access a common intermediate.

3.4.a. Synthesis of (−)-Cylindricine C and 2-epi-(−)-Cylindricine C

In 2006, we reported the total synthesis of (−)-cylindricine C.⁵⁶ As aforementioned, we intended to utilize an intramolecular aza-[3 +3] annulation strategy that would be amenable for the nitrogen aza-heterocyclic motif found in lepadiformine and cylindricines. Retrosynthetically, the key intermediate tricycle 107 leading to the (−)-cylindricines would be accessed via a diastereoselective aza-[3 + 3] annulation of a chiral vinylogous amide tethered to a vinyl iminium ion 108 [Scheme 34]. Vinylogous amide 108 would be prepared from chiral amine 109, which can be constructed from vinyl oxazoline 110. The vinyl oxazoline would be prepared in five steps from L-serine,⁹² thereby providing the source of chirality in this work. The end-game strategy in our total synthesis would utilize a sequence featuring an interesting halohydrin formation of the C4-C5 olefin en route to the C4-carbonyl.

Scheme 34.

Retrosynthetic Analysis of 101.

Our total synthesis commenced with hydroboration of chiral synthon 110 with 9-BBN followed by a Suzuki-Miyaura coupling of the ensuing borane with vinyl triflate 111 to give ester 112 [Scheme 35]. This unsaturated ester was converted to an allyl alcohol in three simple operations, providing 113 in 61% yield for the four-step sequence. The free hydroxyl group of 113 was capped as the corresponding acetate followed by removal of the Boc group using TFA. At this stage, we had considerable flexibility in the choice of an appropriate vinylogous amide [see R¹, R² of 108 in Scheme 34]. Considering our development of a protocol for reductive ring-opening of α-pyrones,³⁶ we elected to employ bromopyrone 114 for vinylogous amide formation. Therefore, reaction of the free amine with 6-n-butyl-4-bromo-2-pyrone 114 led to amino pyrone 115 in 61% yield over three steps. Compound 115 was then converted to enal 116 in 85% yield over two operations.

Scheme 35.

Synthesis of Tetracycle 117.

With the annulation substrate 116 in hand, the vital intramolecular aza-[3 + 3] annulation proceeded smoothly by heating with 0.5 equiv of piperidinium acetate as the catalyst for vinyliminium formation. After 12 h, tetracyclic annulation product 117 materialized in 68% yield as a seperable 9:1 mixture of diastereomers favoring the desired isomer as shown. Importantly, this isomer possessed the necessary configuration at C10 of the alkaloid.

The remaining goal was to install the ketone at C4, yet this proved to be a formidable challenge. We ultimately found the solution to be an intriguing three-step sequence involving chlorohydrin formation [Scheme 36]. Specifically, reaction of tetracycle 117 with 3.0 equiv NCS in aqueous t-BuOH generated chlorohydrin 118 as a single diastereomer in 76% yield (configuration of alcohol not ascertained). TPAP-oxidation and subsequent reductive dechlorination of the tertiary Cl group using Zn and HOAc provided ketone 119 in 65% yield. Significantly, ketone 119 had the correct stereochemistry at C5 of (−)-cylindricine. α-Pyrone 119 was then converted to 120 in an interesting stepwise sequence: (1) hydrogenation provided a partially reduced dihydropyrone; (2) an ensuing sodium cyanoborohydride reduction in the presence of HCl (or HOAc) then gave a reductive decarboxylation product; (3) and subsequent desilylation afforded alcohol 120 in 86% yield over three steps.

Scheme 36.

Synthesis of (−)-Cylindricine C 101.

At this point, a Stork-Crabtree directed hydrogenation was pursued, yet we obtained the non-naturally occurring (−)-2-epi-cylindricine C 121 in 54% yield and 120 was recovered in 35% yield [entry i in Scheme 36]. Alternatively, we found that a remote hydroxyl-directed reduction of the vinylogous amide with Na(OAc)BH₃ gave (−)-cylindricine 101 in 83% yield (entry ii).

3.4.b. Synthesis of (−)-4-deoxo-2-epi-cylindricine C

(−)-4-deoxo-2-epi-cylindricine C (or, putative (−)-lepadiformine) was synthesized^56b in a three-step sequence commencing with tetracycle 117. Notably, 117 is the same tetracycle used in (−)-cylindricine C synthesis prior to chlorohydrin formation. Aza-tetracycle 117 was hydrogenated over Pd/C to afford a reduced intermediate (not shown) as a single diastereomer in 90% yield [Scheme 37]. This 2-pyrone intermediate was reductively ring-opened utilizing LAH followed by hydrogenation.³⁶ After desilylation with TBAF, the reaction sequence afforded putative (−)-lepadiformine 106 along with a minor isomer, (−)-4-deoxycylindricine C 119.

Scheme 37.

Synthesis of Putative (−)-Lepadiformine 106.

While the details are unknown, a possible mechanism for this transformation is illustrated in Scheme 37. After hydrogenation, pyrone 118 is obtained after a 1,6-hydride reduction followed by deconjugative protonation. This intermediate could then lose CO₂ through a retro-Diels-Alder cycloaddition. Subsequent hydrogenation should then be favored from the bottom face of 118, providing a rationale for the stereochemical outcome at C2.

Encouraged by this facile reductive ring-opening of the pyrone motif, we attempted the synthesis of (−)-lepadiformine. Unfortunately, our efforts in applying the aza-[3 + 3] annulation towards (−)-lepadiformine were thwarted. The reason for abandoning this strategy towards (−)-lepadiformine was that after hydrogenating the minor isomer 120 of our aza-[3 + 3] annulation, reductive ring opening of 121 gave an inseperable mixture of epi-lepadiformines (122 and 123) in low yields [Scheme 38]. The 1:1 ratio of diastereomers implied that following extrusion of CO₂ through a retro-Diels-Alder step, both faces of the resultant amino diene intermediate 124 were equally susceptible to hydrogenation.

Scheme 38.

Attempt at (−)-Lepadiformine 102.

3.4.c. Syntheses of (+)-Cylindricines C-E and (−)-Lepadiformine

While developing our intramolecular aza-[3 + 3] formal annulation strategy, we became cognizant of a deficiency in all synthetic strategies towards these families of alkaloids: an N-acyliminium cyclization approach has been exclusively employed in total syntheses of (−)-lepadiformine 102^93,94 and never for any of the cylindricines [Figure 4].^87,95–98 Specifically, Weinreb^88a and Kibayashi^88b employed this N-acyliminium cyclization to construct the C5-C10 bond in the aza-spirocyclic AC-ring of (−)-lepadiformine 102.⁹⁹ This approach suggested that the desired cis-fused 1-aza-decalinic AB-ring at C5-C10 in cylindricines 103–105 may not be a product of the N-acyliminium cyclization approach, since the reaction provided the desired trans relative stereochemistry at C5-C10 of (−)-lepadiformine 102. We decided to test the validity of this presupposition by pursuing the total syntheses of (+)-cylindricines C-E and (−)-lepadiformine via a common intermediate derived from an N-acyliminium, or aza-Prins, cyclization and a seldom used Wharton’s rearrangement.^84h,i This common intermediate 125, we believed, would relate structurally through an epimerization at C5 [Scheme 39].

Scheme 39.

Retrosynthetic Analysis of (+)-Cylindricines C-E and (−)-Lepadiformine.

As illustrated in Scheme 39, aza-tricycle 125 would link the synthetic pathways towards (+)-cylindricines C-E 103–105 and (−)-lepadiformine 102, by either a C5-epimerization en route to (+)-cylindricines, or a C4-deoxygenation leading to (−)-lepadiformine. Since our original plan of a tandem Mannich strategy to the aza-tricycle was thwarted by many difficulties,^84h,i we elected to prepare the vital tricycle 125 from 128 by utilizing an aza-Prins type N-acyliminium addition followed by Wharton’s rearrangement.

Our synthesis commenced with metalating iodide 130, and addition of the ensuing alkyllithium to the enantiomerically pure (S)-lactam 129 provided the ring-opened, Boc-protected amino ketone 128 in 61% yield [Scheme 40]. A formic acid-induced aza-Prins cyclization afforded an inseperable diastereomeric mixture of aza-spirocycles 131 in which the allyl cation intermediate had been trapped by the formate anion.

Scheme 40.

Synthesis of Spirocycle 134.

Next, a three-step sequence was employed to convert diastereomeric spirocycles 131 to epoxy ketone 132. This intermediate was subjected to a Wharton rearrangement using 5.0 equiv hydrazine and 0.5 equiv HOAc to yield the desired transposed allyl alcohol 133-trans in 66% yield as a mixture of inseperable diastereomers. While we observed this trans-allyl alcohol in most trials, we found the cis-allyl alcohol 133-cis in 42% yield when using 10 equiv HOAc. This interesting phenomenon observed in Wharton’s rearrangement may be associated with the stereochemistry of the epoxide or the amount of HOAc used. Subsequent MnO₂ oxidation of 133-trans provided spirocycle 134 in 90% yield.

Removal of the Boc group in enone 134 with TFA resulted in a free amine that underwent an in situ Michael cyclization to provide the desired aza-tricycle 125, the common intermediate, in 72% yield [Scheme 41]. Additionally, we isolated a second product, which after careful scrutiny, learned that the aza-tricycle underwent rapid epimerization at C5 when exposed to silica gel, thereby giving the TBDPS-protected form of (+)-cylindricine C 103 as well as 125 in a 1:1 mixture. It turned out that TBAF-mediated deprotection of the silyl group in trans-azadecalin 125 concomitantly epimerized C5 of the aza tricycle, providing (+)-cylindricine C 103 and importantly verified the link between cylindricines and lepadiformine. Subsequent chemical manipulations then afforded (+)-cylindricines D and E [104 and 105 in Scheme 42].

Scheme 41.

Synthesis of (+)-Cylindricine C.

Scheme 42.

Synthesis of 102, 104 and 105.

To complete the synthesis of (−)-lepadiformine 102, as illustrated in Scheme 42, a four-step sequence was pursued. The C4-carbonyl of the common tricyclic intermediate 125 was reduced with NaBH₄ and the resultant β-alcohol was converted to the corresponding xanthate. Barton-McCombie deoxygenation was then performed, and subsequent desilylation afforded (−)-lepadiformine 102 in 57% yield over four operations. Ultimately, we established a unique and potentially biosynthetic link between the cylindricines and lepadiformine, and that the C5-H can be epimerized but possibly only with an aza-tricyclic intermediate.

3.5. A Unified Strategy Towards Azaphenalene Alkaloids

As a defensive mechanism, ladybird beetles (Coccinellidae) release an orange fluid that contains a mixture of defensive alkaloids, serving to protect them from predators.^100,101 In the early 1970s, Tursch and coworkers reported the isolation of a family of structurally related azaphenalene alkaloids—namely, myrrhine 136, hippodamine 142, convergine 143, precoccinelline 144 and coccinnelline 145—from the aforementioned fluid secreted by Coccinellidae.¹⁰² In 1976, Ayer reported the first total syntheses¹⁰³ of these five azaphenalene alkaloids, and subsequently, other syntheses have been published.^104,105 With one exception,^103a installation of the equatorial methyl group was performed in the latter stages of the synthesis.

3.5.a. Stereoselective Synthesis of Myrrhine, Hippodamine, Convergine, Precoccinelline and Coccinelline

To advance our quest in demonstrating the utility of our pivotal intramolecular aza-[3 + 3] annulation as a unified strategy to access different N-heterocyclic manifolds known in naturally occurring alkaloids, we synthesized the family of Coccinellidae defensive alkaloids 136 and 142–145 [Figure 5].⁵⁷ Our strategy utilizes a common intermediate, the aza-tricyclic core 141, which is attained via an intramolecular aza-[3 + 3] annulation.

Figure 5.

Potential Routes toward the Azaphenalene Alkaloids.

We envisioned that the symmetric nature of myrrhine 136, hippodamine 142 and precoccinelline 144 could be exploited and retrosynthetically could be derived via four distinct pathways depending upon the stereochemical outcome of the intramolecular aza-[3 + 3] annulation and the location of the methyl substituent [Figure 5].

We considered many approaches, pathways A-D, towards these azaphenalene alkaloids. If vinylogous amide 140 undergoes an aza-[3 + 3] annulation selectively to favor the anti (defined by the angular hydrogen atoms in blue) annulation product 141, hippodamine 142 and precoccinelline 144 could be rapidly constructed, as 141 would match three of the four stereocenters in 142 and 144 (pathway A). However, access to myrrhine 136 would necessitate an epimerization of 141. A more expeditious synthesis of 136, as illustrated by pathway B, would be to utilize syn (defined by the angular hydrogens in red) aza-tricycle 135 via a syn-selective annulation of 140. In pathway C, the carbon tether of vinylogous amide 146 would contain the methyl group of these natural products, and an enantioselective total synthesis of 142 (136 and 144 are achiral or meso) could be observed if 146 can be made in an optically enriched manner. Lastly, an enone intramolecular aza-[3 + 3] annulation of vinylogous amide 139 could be employed to obtain the azaphenalene alkaloids via 137 and/or 138.

With these approaches, we elected to pursue pathways A and B because it is not trivial to build 146 in an optically enriched manner, and a useful enone version of our pivotal aza-[3 + 3] annulation had yet to be developed. Thus, the key common intermediate 141 would be prepared by an aza-[3 + 3] annulation of vinylogous urethane 140, which would be accessed in several steps from cis-1,3-disubstituted lactam 149 with glutarimide 147 and bromide 148 as the essential starting points [Scheme 43].

Scheme 43.

Retrosynthetic Analysis of 141.

Our synthesis commenced with reductive alkylation of glutarimide 150 with the Grignard reagent generated from bromide 148 [Scheme 44]. The reduction proceeded stereoselectively, affording lactam 149 in 75% yield exclusively as the 1,3-syn isomer. To access vinylogous urethane 152, we elected to employ an Eschenmoser’s episulfide contraction. Thus, 149 was converted to thiol imidate 151 in 90% yield over two steps, and was subsequently treated with PPh₃ in the presence of DIPEA in CH₃CN. After 48 h, vinylogous urethane 152 was isolated in 90% yield.

Scheme 44.

Synthesis of Annulation Precursor 152.

The substrate 140 for the aza-[3 + 3] annulaion was then prepared in 82% yield over two operations from 152 [Scheme 45]. Treatment of vinylogous urethane 140 with the more reactive piperidinium trifluoroacetate salt provided aza-tricycle 141 stereoselectively as a single isomer with anti relative stereochemistry in 51% yield. Implementing our one-pot protocol involving in situ hydrogenation of the cycloadduct 141, after the formal cycloaddition, gave 153 in 43% yield over two steps. In this case, Pd(OH)₂/C was chosen as the catalyst due to its lower tendency, relative to Pd/C, to be poisoned by the amine. At this stage, tricycle 153 possesses three of the four stereocenters in hippodamine 142 and precoccinelline 144, and therefore represents the common intermediate upon which stereodivergent conversion of 153 to precoccinelline, coccinelline, hippodamine and convergine were accomplished [vide infra, Schemes 48 and 49].

Scheme 45.

Synthesis of Key Aza-Tricycle 153.

Scheme 48.

Stereodivergent Hydrogenation of 153.

Scheme 49.

Total Syntheses of 142, 143, 144 and 145.

To complete the total synthesis of myrrhine 136, we needed an adequate epimerization of the stereocenter (shown in red) of the common intermediate 141. As illustrated in Scheme 46, we elected to pursue an aromatization-reduction sequence by treating aza-tricycle 141 with DDQ, and the ensuing pyridinium salt 154 was hydrogenated over Adams’ catalyst in AcOH to give the all-syn aza-ticycle 155 in 54% yield for the two-step protocol as a 2:1 mixture of diastereomers with respect to the ester group.

Scheme 46.

Hydrogenation of Pyridinium Salt 154.

Submitting this 2:1 diastereomeric mixture of 155-ax and 155-eq to saponification conditions afforded the equatorial acid 156-eq as the predominant diastereomer [Scheme 47]. This indicated that the axial isomer underwent significant epimerization to the more stable equatorial acid. Unfortunately, a subsequent Barton’s decarboxylation protocol of the diastereomeric mixture provided myrrhine 136 in very low yield with almost complete recovery of the equatorial acid starting material. This outcome implied that the equatorial acid 156-eq is much less reactive to decarboxylation than the axial acid. Therefore a different protocol was employed.

Scheme 47.

Completing the Synthesis of Myrrhine.

Revisiting pyridinium salt 154 in Scheme 46, instead of using Adams’ catalyst in AcOH we had found conditions to hydrogenate the salt using 14 psi H₂ over Pd(OH)₂ in MeOH, that would provide 155 in 44% yield over two steps as a 5:1 mixture of diastereomers. We opted to saponify this diastereomeric mixture of 155-ax and 155-eq (5:1 isomeric ratio in favor of 155-ax) using LiI in EtOAc, providing lithium carboxylate 157-ax in 73% yield with no epimerization observed [Scheme 47]. Lastly the mixture was acidified and the ensuing acid was submitted to Barton’s decarboxylation conditions to give myrrhine 136 in 31% yield over a four-step sequence.

To complete the stereodivergent syntheses of the other members in the Coccinellidae family— precoccinelline, coccinelline, hippodamine and convergine—we returned to aza-tricycle 153, the common intermediate, as the stereodivergent starting point. To this end, hydrogenation of aza-tricycle 153 over Adams’ catalyst proceeded in a stereodivergent manner, giving a 2:1 isomeric ratio of 158a and 158b [Scheme 48]. Of importance, precoccinelline 144 and its N-oxide, coccinelline 145, possess the framework observed in major isomer 158a, while 158b resembles hippodamine 142 and convergine 143. Rather than chromatographically separating the mixture of esters 158a and 158b, we elected to resolve the esters via alkaline hydrolysis of equatorial ester 158a. Thus, the crude mixture of esters 158a,b was treated with aqueous KOH and a simple extraction provided unreacted axial ester 158b in 36% overall yield.

To isolate acid 159a, the aqueous phase was acidified, thereby affording 159a in 49% yield over two operations from 153. The synthesis of precoccinelline 144 was finished by subjecting acid 159a to Barton’s decarboxylation protocol, providing the desired alkaloid in 43% overall yield [Scheme 49]. An ensuing oxidation with m-CPBA afforded coccinelline 145 in an excellent 96% yield.

Unfortunately, the synthesis of hippodamine 142 proved to be a more formidable challenge, whereby preliminary efforts to directly hydrolyze axial ester 158b as well as epimerization (K₂CO₃/MeOH and DBU/toluene protocols) of the axial ester group to the more stable equatorial ester 158a failed. We attributed the unsuccessful hydrolysis attempt to the axial ester 158b being hindered, thereby rendering it inaccessible. On the other hand, both of the epimerization protocols were inadequate with complete recovery of the starting material. Ultimately, we found conditions that were amenable to epimerize then hydrolyze the ester, thus transforming 158b to acid 159b.

Therefore, treatment of ester 158b with KHMDS at −78 °C followed by quenching the reaction mixture with MeOH at 0 °C gave the desired equatorial ester that was subsequently hydrolyzed with aqueous KOH to afford acid 159b [Scheme 49]. Hippodamine 142 was then obtained in 43% yield by decarboxylation of 159b. An ensuing oxidation with m-CPBA afforded convergine 143 in 88% yield. This work provided a novel approach to the Coccinellidae family and further established the pivotal aza-[3 + 3] annulation as a unified strategy to intercept alkaloid families.

4. Concluding Remarks

After accomplishing a diverse array of total syntheses, we believe that the aza-[3 + 3] annulation has provided us with a unique perspective in alkaloid synthesis. We recognize that the aza-[3 + 3] annulation represents a powerful strategy in organic synthesis that can be both flexible in accessing structural and stereochemical diversity and competitive in overall efficiency.

Scheme 5.

Asymmetric Intramolecular Aza-[3 + 3].