Synthesis of vitisin A & D enabled by a persistent radical equilibrium

Abstract

The first total synthesis of the resveratrol tetramers vitisin A and vitisin D is reported. Electrochemical generation and selective dimerization of persistent radicals is followed by thermal isomerization of the symmetric C8b–C8c dimer to the C3c–C8b isomer, providing rapid entry into the vitisin core. Computational results suggest that this synthetic approach mimics Nature’s strategy for constructing these complex molecules. Sequential acid-mediated rearrangements consistent with the proposed biogenesis of these compounds afford vitisin A and vitisin D. The rapid synthesis of these complex molecules will allow for further study of their pharmacological potential.

Graphical Abstract

Resveratrol and its oligomers comprise a natural product class of hundreds of structurally distinct compounds; however, challenges associated with isolating individual molecules in adequate supply and purity have precluded rigorous evaluation of their pharmacological potential.¹ Numerous research groups have devised innovative synthetic approaches to access these complex molecules, including cationic cyclizations,² transition metal catalysis,³ and reagent-controlled bromination.⁴ It has been proposed that in nature, oligomerization proceeds via phenoxyl radical intermediates.¹ Indeed, this biosynthetic hypothesis has inspired our own efforts in this area. We recently reported the use of persistent phenoxyl radicals (e.g. 1) for the synthesis of C8–C8′ resveratrol dimers pallidol and quadrangularin A,⁵ and tetramers nepalensinol B (3) and vateriaphenol C (4, Figure 1).^6,7 It has further been proposed that interconversion of the C8–C8′ and C3–C8′ constitutional isomers is the key step in the divergent biogenesis of related natural products possessing C3–C8′ connectivity.¹ Herein, we leverage the facile equilibrium between persistent phenoxyl radicals and their corresponding quinone methide dimers (i.e. 1 and 2, respectively, Figure 1) to achieve the first total synthesis of vitisin A and vitisin D (6 and 7, Figure 1).

Figure 1.

Divergent reactivity of persistent radicals in the synthesis of resveratrol tetramers.

A new opportunity for C3–C8′ resveratrol oligomer synthesis was discovered while investigating the 8A–9A equilibrium (Figure 2A). Discontinuity in the Van’t Hoff analysis above 50 °C suggested an alternative and irreversible reaction path was accessible to 9A (see Figure S3 of the Supporting Information).⁶ Importantly, no such discontinuity was observed in corresponding experiments with dimers featuring tert-butyl groups (8B/9B) in lieu of the TMS groups in 8A/9A. Subsequent NMR analysis revealed that upon heating the C8–C8′ dimer 8A had rearranged to the δ-viniferin core 11. We speculated that this rearrangement proceeded via the intermediacy of the C3–C8′ dimer 10A, which was assumed to be less energetically favorable than the C8–C8′ dimer 8A but primed to lose the trimethylsilyl group and aromatize the phenolic moiety. Indeed, computations suggest that the C3–C8′ bond in 10A is 7.4 kcal/mol weaker than the C8–C8′ bond in 8A.⁸ However, the simple fact that 9A is in equilibrium with 8A (for which the C8–C8′ bond dissociation enthalpy was previously determined to be 16.4 kcal/mol⁶), implies that it is also in equilbrium with the C3–C8′ dimer 10A. Presumably, 10A is not observed due to its rapid decomposition to the δ-viniferin core (11). Computations predict that unimolecular expulsion of the TMS cation proceeds with a significant barrier,⁹ suggesting that adventitious water in the acetone promotes desilylation and concomitant aromatization. Remarkably, the C8–C8′ to C3–C8′ isomerization and subsequent cyclization to 11 occurs in nearly quantitative yield (Figure 2A).

Figure 2.

A) Discovery of C8b–C8c to C3c–C8b isomerization from thermodynamic study of persistent radical equilibria. B) Retrosynthetic analysis for the synthesis of vitisin tetramers based on this late stage homolytic isomerization. *C8-C8′ BDEs determined experimentally;⁶ BDEs for the corresponding C3-C8′ dimers are given relative to these experimental values.⁸

The difference in the C-C BDEs of the corresponding tert-butylated dimers (8B/10B) is nearly twice as large (13.9 kcal/mol), reflecting the greater steric repulsion in the C3–C8′ dimer resulting from the shorter C–C bonds relative to C–Si bonds. Indeed, the C3–C8′ and C8–C8′ BDEs in the dimers which lack any ortho substitution (8C/10C) are computed to be only 3.6 kcal/mol apart, funneling both of these intermediates toward the δ-viniferin core. These computations are consistent with the hypothesis that resveratrol oligomerization relies upon equilibration of the C8–C8′ and C3–C8′ constitutional isomers.¹ In fact, numerous research groups have realized the direct conversion of resveratrol (S1) to δ-viniferin (11A, Figure 2A) through single electron oxidation strategies (see Table S1 of the Supporting Information for a summary of these efforts).¹⁰ While excellent yields for conversion to 11A have been realized, extension of this strategy to higher-order oligomers has not achieved the same degree of success. The most noteworthy example was reported by Sako and co-workers in their semi-synthesis of vitisin B (5A).^10g By treating isolated (+)-ε-viniferin (S4) with silver acetate in methanol at elevated temperatures (50 °C), they observed conversion to vitisin B (5A) in 40% yield on 20 milligram scale (see Figure S2 of the Supporting Information). The C3c–C8b-fused resveratrol tetramers exhibit some of the most compelling biological activities among members of the resveratrol oligomer class reported to date; therefore, an approach to their synthesis reliant on readily available materials is desirable. For example, vitisin B (5A) was found by Lee and co-workers to be a potent inhibitor of the NS3 helicase of hepatitis C (IC₅₀ = 3 nM), while vitisin A (6) was also active against the same target (IC₅₀ = 35 nM).¹¹ To the best of our knowledge, similarly rigorous biological analysis of vitisin D (7) has not yet been reported.

Quinone methide dimer 2 serves as the linchpin for the proposed isomerization approach to the C3c–C8b resveratrol tetramers (Figure 2B). Importantly, this scaffold forms exclusively as a single diastereomer upon oxidation of racemic starting material 12, meaning dimerization occurs selectively between the same enantiomeric precursors in the same fashion required to access vitisin B (5A). Preservation of this stereochemical integrity during C8b–C8c to C3c–C8b isomerization would directly convert 2 to the core of vitisin B (5). As 5A is proposed to be the biogenic precursor to both 6 and 7,¹² this novel formal [1,5]-shift could in principle provide access to multiple biologically active members of the C3c–C8b tetramers. Hypothesizing that the late-stage C8b–C8c to C3c–C8b isomerization would be preceded by our recently reported method for the electrochemical dimerization of phenylpropenoid scaffolds,¹³ attention turned to the synthesis of the dimerization starting material – protected ε-viniferin analog 12. This scaffold was previously prepared using an approach developed by the Snyder group in one of their seminal contributions to this field;^4c however, it was envisioned that a benzofuran precursor (i.e. 13) might offer a more direct route to 9. Kim and Choi recently reported an approach to 13A to access permethylated resveratrol dimers,^3b providing an excellent starting point for our synthesis.

Execution of this plan commenced with the preparation of the requisite benzofuran precursor while adopting a new protecting group strategy, and gratifyingly, cyclodehydration of 14A afforded benzofuran 15A in 85% yield (Figure 3). Unfortunately, C–H arylation of 13A did not occur – instead these efforts were plagued by silyl deprotection. Benzyl ethers could be replaced for silyl ethers at this stage to allow for elaboration to 16 (see Figure S5 of the Supporting Information for details). Alternatively, isopropyl ethers have been demonstrated to be similarly robust to methyl, yet more readily cleaved, so this protection strategy was employed to access 16 in a shorter sequence with fewer protecting group manipulations (Figure 3). After C–H arylation,¹⁴ 13B was converted to 16 by Lewis acid-mediated deprotection,¹⁵ Kishi reduction,¹⁶ and silyl protection in a three-step sequence that did not require intermediate purification. Reduction of the C8b-ester and a Parikh-Doering oxidation¹⁷ delivered aldehyde 17, and a Wittig olefination¹⁸ smoothly afforded the silyl protected ε-viniferin analog 12A. At this point, our dimerization method utilizing anodic oxidation was employed to access 2A.¹³ Deviation from the published conditions was required to ensure full solubility, and, after adding dichloromethane as a co-solvent and decreasing the electrolyte concentration by half, 12A was converted to the desired quinone methide tetramer 2A in 63% yield (Figure 3).

Figure 3.

Synthesis of silyl-protected bis-quinone methide 2A from the electrochemical dimerization of 12A.

With the silyl-protected tetrameric material in hand, the key C8b–C8c to C3c–C8b isomerization step was next investigated. In order to determine if the persistent radicals would escape the solvent cage prior to the formal [1,5]-shift, possibly leading to mismatched C3c–C8b oligomers, a crossover experiment was performed with 8A and a differentially protected analog (see Figure S4 of Suppporting Information). Indeed, the crossover products were observed, suggesting C–C fragmentation and diffusion is competitive with isomerization. However, the dimer model system was not sufficient to determine if the stereochemical integrity of 2A would be completely eroded. Subjecting 2A to the thermal isomerization conditions readily accessed the vitisin B core (5B), but as a mixture of four C3c–C8b dihydrobenzofuran (DHB) isomers (Sequence 1, Figure 4). Increasing the temperature improved the trans/cis ratio of the DHB rings, presumably due to thermal epimerization; however, the facial selectivity of C3c–C8b recombination remained unchanged. Gratifyingly, the TBS ethers were readily cleaved with HF-triethylamine to afford two compounds – 19 and 20. These trans-DHB isomers arise from each possible facial addition of C8b to C3c during the formation of 18 (Sequence 1, Figure 4) suggesting that the stereochemical integrity is preserved during the thermal isomerization. The O-silyl deprotection conditions also resulted in epimerization of cis-DHB isomers to the corresponding trans-DHBs, which is well precedented in the literature,^3f,g thereby delivering only the two observed products. To support the hypothesis that the formal [1,5]-shift only occurs through the relative configuration depicted by 2A, a second crossover experiment between 2A and the corresponding TIPS-protected analog 2B was performed (Figure 4). After thermal isomerization, the crossover product was observed, further supporting that C–C fragmentation and diffusion is competitive with in-cage recombination to yield the isomer. However, upon TBS deprotection with HF-triethylamine, 19 and 20 were the only observed products, suggesting that thermal isomerization proceeds without loss of stereochemical integrity afforded by C8b–C8c dimerization.

Figure 4.

C8b–C8c to C3c–C8b isomerization, Friedel-Crafts cyclization, and deprotections deliver vitisin A (3).

Inspired by these results, we set out to develop conditions for O-silyl deprotection and protodesilylation to complete the synthesis. Global desilylation was achieved by addition of a methanolic solution of hydrochloric acid upon completion of the isomerization of 2A yielding vitisin A (6) and its C7b, C8b-isomer (23). Vitisin A (6) is proposed to arise from vitisin B (5A) in the biosynthesis of these compounds.¹² Acid-mediated cleavage of the C7b–O bond upon protonation of the DHB (21) affords a quinone methide (22) to which C10a of the adjacent resorcinol ring adds in a 7-exo-trig cyclization (Sequence 2, Figure 4). Of note, the C10a–C7b bond formation exclusively delivers the relative configuration depicted in Figure 4, with the C8b configuration dictating the facial selectivity of the cyclization.

In an attempt to prevent Friedel-Crafts rearrangement and directly reveal vitisin B (5A), intermediates 19/20 were exposed to milder protodesilylation conditions.¹⁹ Remarkably, vitisin D (7) and the isomer 26 were isolated after treating 19/20 with SiMe₃Cl, KI, and H₂O in MeCN for one hour (Figure 5A). These compounds are the result of two acid-promoted cyclizations; after cyclization to vitisin A (6) in situ (Sequence 2, Figure 4), protonation of the stilbene at C8c results in formation of the quinone methide tautomer 24/25, which is trapped by 7-exo-trig cyclization by C10d of the adjacent resorcinol ring (Sequence 3, Figure 5A). Acid-mediated conversion of vitisin A (6) to vitisin D (7) has been proposed in the biosynthesis of these compounds,¹² thus it is perhaps unsurprising that the in-situ generation of hydroiodic acid would give rise to 7 and 26.

Figure 5.

A) C8b–C8c to C3c–C8b isomerization, Friedel-Crafts cyclizations, and deprotections deliver vitisin D (4). B) C5c-TMS group survives fluoride-mediated desilylation attempts.

Alkaline conditions were next attempted in an effort to target vitisin B (5A).²⁰ It was quickly determined that the C3/5b aryl silyl groups are cleaved via fluoride-mediated-desilylation; however, the C5c-TMS group remains intact (27/28, Figure 5B). Our investigations have demonstrated that the C3c–C8b DHB is quite labile, and in fact, conversion to vitisin A (6) occurs under a range of conditions (see Table S6 of the Supporting Information). Despite an extensive evaluation of desilylation conditions, access to vitisin B (5A) from this approach remains an outstanding challenge, the solution to which will be reported in due course.

Herein we have reported the first total synthesis of the resveratrol tetramers vitisin A (3, 3.3% overall yield) and vitisin D (4, 3.7% overall yield) in 10 and 11 steps from 14A, respectively. Our approach utilizes persistent radicals to enable a unique, late-stage, formal [1,5]-shift that is consistent with the biosynthetic hypothesis of these complex molecules. The persistent radicals arise from mild anodic oxidative conditions, and the subsequent biomimetic transformations rapidly convert dimeric intermediates to tetrameric structures, which will enable the investigation of the pharmacological potential of these complex molecules.