Enantioselective Total Synthesis of Clavirolide C. Applications of Cu-Catalyzed Asymmetric Conjugate Additions and Ru-Catalyzed Ring-Closing Metathesis

Abstract

The first enantioselective total synthesis of clavirolide C, a member of the dolabellane family of diterpenes isolated from Pacific soft coral Clavularia viridis, is disclosed. The total synthesis features the application of chiral amino acid-based ligands in Cu-catalyzed asymmetric conjugate addition (ACA) reactions and a relatively rare application of catalytic ring-closing metathesis to access an eleven-membered ring structure. The total synthesis effort has spawned the development of a new protocol for NHC∙Cu-catalyzed ACA of alkylaluminum reagents to β-substituted cycloalkenones. The enantioselective clavirolide C synthesis requires seventeen steps (longest linear sequence), affords the target molecule in 3.5% overall yield, and confirms the stereochemical assignment for the natural product.

Clavirolide C, a member of the dolabellane family of diterpenes isolated from Pacific soft coral Clavularia viridis, contains a characteristic trans-bicyclo[9.3.0]tetradecane core structure (Scheme 1).^1,2 Our interest in this natural product stems from the viewpoint that an efficient and enantioselective total synthesis of clavirolide C would demonstrate, and, perhaps more importantly, challenge the utility of the state-of-the-art in catalyst and synthesis methodology.³ From the perspective of total synthesis,⁴ two of the more demanding structural attributes of this class of natural products are the strained eleven-membered carbocycle and the all-carbon quaternary stereogenic center⁵ that resides at the fusion of the medium ring and the cyclopentane moiety. Herein, we disclose the first enantioselective total synthesis of clavirolide C, featuring the application of chiral amino acid-based ligands in Cucatalyzed asymmetric conjugate addition (ACA) reactions⁶ and a Ru-catalyzed ring-closing metathesis (RCM).⁷ More noteworthy is that the present initiative has spawned the development of a protocol for NHC∙Cu-catalyzed ACA of alkylaluminum reagents to β-substituted cyclopentenones, one of the more difficult substrate sets for this class of transformations.

Scheme 1.

A Retrosynthesis Analysis for Clavirolide C

The retrosynthesis analysis adopted in our approach to clavirolide C is outlined in Scheme 1. We envisioned a route that would contain three stereoselective conjugate additions, leading to control of stereochemistry at C8, C4 and the critical all-carbon quaternary stereogenic center, C1. The stereochemical outcome of the conjugate addition to cyclo-undecenone I (stereochemistry at C4) might be addressed by utilization of the conformational preferences of the medium ring (substrate control)⁸ or through the use of an appropriate chiral complex (catalyst control).³ Catalytic RCM, albeit aimed to access a challenging eleven-membered ring, would convert diene III to II, which would in turn serve as the precursor to unsaturated tricycle I. The most direct strategy to access cyclopentanone III would be through aldol addition of cyclic silyl enol ether V and aldehyde IV, both of which might be derived from a catalytic ACA.

We began by searching for an efficient and selective protocol to reach the intermediate represented by V (Scheme 1). We were cognizant from the outset that identifying an effective catalytic ACA to access V would be challenging. Only recently, a small number of disclosures regarding catalytic ACA of β-alkyl-substituted cyclic enones has appeared,⁹ but such protocols almost exclusively deal with reactions of six- and seven-membered ring cyclic enones.¹⁰ Our earlier studies involving chiral bidentate NHC∙Cu complexes and dialkylzinc reagents had proven entirely ineffective with β-substituted cyclopentenones (<2% conv).^10b Thus, we turned to the chiral Cu complex of NHC-sulfonate (S, S)-1 (Table 1),¹¹ a recent and particularly active catalyst, and the more Lewis acidic Al-based alkylating agents (vs dialkylzinc reagents).¹² As depicted in entry 1 of Table 1, treatment of β-substituted cyclopentenone 4 and commercially available Me₃Al, in the presence of 2.5 mol % (S, S)-1 and 5 mol % Cu(OTf)₂, gives rise to 75% conversion of 4 to S-5 (36 h, -78 °C) in 73.5:26.5 er. We rationalized that increasing the freedom of rotation of the chiral NHC's N-aryl unit might enhance its effective size. Such considerations led us to prepare and examine complexes (S)-2 and (S)-3 which, as shown in entries 2-3 of Table 1, furnish the desired β,β-disubstituted cyclopentanone S-5 with improved selectivity (86:14 and 93:7 er, respectively). When the ACA, promoted by 3.75 mol % 3, is quenched with Et₃SiOTf (-78 °C), enolsilane S-6 is obtained in 92:8 er (84% ee) and 72% yield (entry 4, Table 1). To reach complete conversion, slightly higher catalyst loading is therefore required (3.75 mol % in entry 4 vs 2.5 mol % in entries 1-3).

Table 1.

Activity of NHC∙Ag(l) Complexes for ACA of 4 with Me₃Al^a

entry	NHC∙Ag complex	product	conv (%);b yield (%)c	erd	ee (%)d
1	1	S-5	75; nd	73.5:26.5	47
2	2	S-5	97; nd	86:14	72
3	3	S-5	86; 80	93:7	86
4	3e	S-6	>98; 72	92:8	84

^aReactions performed under N₂ atm.

^bDetermined by analysis of ¹H NMR spectra of unpurified mixtures.

^cYields of purified products.

^dDetermined by chiral GLC analysis; see the Supporting Information for details.

^eReaction was performed with 3.75 mol % 3 and 7.5 mol % Cu(OTf)₂; 4.0 equiv Et₃SiOTf was added after 36 h (-78 °C, 4 h); see the Supporting Information for details, nd = not determined.

Enantioselective synthesis of aldehyde 11 (Scheme 2) involves another class of chiral ACA catalysts developed in these laboratories.¹³ The sequence illustrated in Scheme 2 begins with a two-step procedure involving reaction of unsaturated lactone 7 and Me₂Zn with (CuOTf)₂∙toluene, all of which are commercially available, in the presence of chiral amino acid-based ligand 8 and benzaldehyde (Zn-enolate must be trapped in situ for high yield).¹³ The resulting Cu-catalyzed ACA/aldol product is converted to 9 in 86% overall yield and >99:<1 er. For the Cucatalyzed transformation to proceed efficiently, ligand modification was required, since use of the previously reported chiral Schiff base, bearing an -N(H)n-Bu terminus (vs. -NEt₂ in 8), gives rise to an inefficient process (10 mol % catalyst, 48 h, 60% conv, 40% yield, >99:<1 er). Conversion to amide 10 (81% yield for two steps) is followed by a three-step sequence that delivers fragment 11 as an equal mixture of diastereomers (inconsequential to the total synthesis) in 72% overall yield.

Scheme 2.

Enantioselective Synthesis of Aldehyde Fragment 11

With enantiomerically enriched cyclic enol silane R-6 (92:8 er, obtained from ACA with R-3) and aldehyde 11 (>99:<1 er) in hand, we searched for an effective procedure for the union of the two fragments through a diastereoselective aldol addition (Scheme 3). After exploring a range of conditions, we established that treatment of the lithium enolate derived from 6 (n-BuLi, 22 °C) with BEt₃ at -78 °C, followed by the addition of aldehyde 11, leads to the formation of aldol adduct 12 (75% total yield). The desired β-hydroxyketone is generated in >95% anti-aldol selectivity (C11-C10), a stereochemical preference that is due to the introduction of BEt₃ to the mixture.^14a Approach of the aldehyde anti to the slightly larger butenyl unit is, however, not significantly favored (C1-C11, 1.5:1 dr).¹⁴

Scheme 3.

Enantioselective Total Synthesis of Clavirolide C

Next, we addressed the problem of establishing conditions for efficient eleven-membered ring closure through a catalytic RCM process. After extensive experimentation, we found that subjection of allylic ether 12 (4:1 dr)¹⁵ to 10 mol % of Ru carbene 13¹⁶ under high dilution conditions (ClCH₂CH₂Cl, 10^-3 M) with slow addition of the substrate to a solution of catalyst (83 °C, 6 h) results in clean formation of the desired ring structure;¹⁷ cycloalkene 14 is obtained in 70% yield after purification and exclusively as the trans isomer (>95%). The present transformation represents a rare example of an efficient catalytic RCM that leads to the formation of an eleven-membered ring.¹⁸ Two additional points regarding catalytic RCM of 12 merit mention: 1) Use of the related second-generation Cy₃P-containing complex¹⁹ under identical conditions leads to <10% conversion to the desired macrocycle 14 (>90% 12 recovered).²⁰ 2) Attempted RCM with the parent allylic alcohol results in the formation of a complex mixture of products; when the derived α,β-unsaturated ketone is used, the product mixture is contaminated with 30-35% of the homodimeric 22-membered ring.

Esterification of the secondary alcohol with 2-bromopropionyl bromide, followed by a SmI₂-mediated Reformatsky reaction provides lactone 15.²¹ Removal of the silyl protecting group under acidic conditions (30 mol % CSA, 0 °C) and oxidation of the resulting secondary alcohol (MnO₂), furnishes enone 16, the identity of which has been confirmed through X-ray crystal structure analysis. The four-step sequence, beginning with 14, proceeds in 35% overall yield and 3:1 dr (R stereochemistry at C18 is major; inconsequential to the total synthesis).

Conjugate addition of Me₂CuLi∙LiI to enone 16 in the presence of TMSCl,²² and desilylation of the resulting enol silane with HF∙pyr, leads to eleven-membered ring ketone 17 in 9:1 diastereoselectivity (at C4) in 71% yield (>98% R at C18).²³ As illustrated in Scheme 3, the identity of 17 has been established through X-ray crystal structure analysis. It is important to note that approach of the Cu-based reagent from the periphery⁸ of the medium ring structure is expected to deliver the undesired stereochemistry at C4. Spectroscopic analysis (nOe measurement) indicates that, in equilibrium with the s-trans isomer, there exists a significant amount of the enone s-cis conformer in solution.²⁴ It is plausible that the eleven-membered enone's s-cis conformer reacts more readily to furnish the requisite stereoisomer as the major product.

Conversion of the tertiary alcohol in 17 to the derived mesylate is directly followed by subsequent elimination to afford the corresponding β,γ-unsaturated lactone (tetrasubstituted alkene). Direct treatment (without purification) of the unsaturated tricycle with DBU (in MeOH) delivers (-)-clavirolide C in 82% overall yield (two steps).²⁵

The enantioselective synthesis of clavirolide C requires seventeen steps (longest linear sequence) and affords the target molecule in 3.5% overall yield. These investigations confirm the stereochemical assignment for clavirolide C, which was originally based on indirect correlation with products derived from catalytic hydrogenation of other members of this class of natural products (bearing olefins at C3-C4).^1a

The total synthesis demonstrates the utility of Cu-catalyzed ACA reactions of unsaturated heterocycles with dialkylzinc reagents and provides another notable illustration of the special attributes of the thermally robust phosphine-free Ru-based olefin metathesis catalyst 13. Completion of the enantioselective synthesis required the development of an effective Cu-catalyzed protocol for generation of all-carbon quaternary stereogenic centers⁵ by reactions of alkylaluminum reagents with one of the more difficult substrate classes in catalytic ACA.²⁶ Nonetheless, these investigations point to the yet unsolved problem of identifying a stereoselective method for aldol additions involving cyclic enol ethers adjacent to quaternary stereogenic centers, especially in cases where such neighboring substituents are similar in size (e.g., 6).²⁷ Control of selectivity in these challenging transformations through an effective chiral catalyst is perhaps the most efficient and attractive option.²⁸ Studies along these lines are in progress.