Structural Revision and Total Synthesis of Caraphenol B and C

Abstract

Chemical syntheses of two stereochemically unique resveratrol dimers, caraphenols B and C, have shown that their structures are misassigned. Thoughts on their potential chemical etiology led to an alternate structural proposal that has been confirmed through synthesis, one indicating that the substituents on their respective indane systems exist in a relative trans,trans orientation rather than the originally postulated all-cis arrangement.

Throughout the world, plants utilize resveratrol and the hundreds of molecules derived from its self-merger as a front line chemical defense against fungal infection.¹ Such ubiquity may be a consequence of the family’s tremendous architectural diversity, with the structures of the natural products drawn within Figure 1 (1–8) reflecting a small subset of the dimeric materials within the collection, cores whose further decoration with additional resveratrol units at multiple positions affords even greater complexity.² Interest in these materials during the past decade has been high,³ be it from efforts to understand their biogenesis,⁴ test the power of specific synthetic strategies and methods against specific frameworks,⁵ or study their chemical biology more fully (particularly for properties outside fungal infection).⁶

Figure 1.

Structures of selected dimeric resveratrol-based natural products (not drawn with absolute stereochemistry).

Our endeavors have been focused largely on the synthetic front, with our key goal being the identification of a unified and global synthetic approach capable of delivering any member within the family in a controlled fashion.⁷ To date, we have synthesized over 20 different natural products at the dimeric, trimeric, and tetrameric levels encompassing a range of different types of ring systems, including indanes, 7-membered rings, dihydrofurans, and bicyclic frameworks. It was in this vein that the structures of caraphenols B and C (1 and 2),⁸ two isolates from the dried roots of Caragana sinica (which has been used as a folk medicine in China to treat hypertension and contusions), caught our attention.^9,10

Although many indane-based natural products are known within the family, these would appear to be the only two materials that orient their pendant substituents in an all-cis orientation, with specific novelty at the two starred positions.¹¹ Indeed, other members, such as ampelopsin D (3)¹² and quadrangularin A (4)¹³ place their labelled B- and C-aryl rings trans. As we have recently shown,^7ab indane-based materials such as 3 and 4 can subsequently be converted into other dimeric cores like ampelopsin F (5)¹² and pallidol (6)¹⁴ through electrophilic activation in a potentially biomimetic process. To the best of our knowledge, however, no material yet isolated can similarly be traced to the unique stereochemistry of caraphenol B and C; other diastereomers of 5 and 6 exist, such as 7,^7d,8 but they reflect alteration in double bond geometry in what would be similar electrophile-induced cyclizations of indane starting materials, not variations in core ring stereochemistry. Intriguingly, reduced variants of 1 and 2 obtained from alternate plant sources (such as 8)¹⁵ also match the trans-stereochemical arrangement typical of the family. In this communication, we show through chemical synthesis that the structures of caraphenol B and C (1 and 2) are misassigned, and, in line with the rest of the family, possess an all-trans configuration of substituents on its indane core.

Our efforts to forge the all-cis stereochemical arrangement of the proposed structure of caraphenol B (1) began from permethylated ampelopsin D (9)^7ab as shown in Scheme 1. Following a standard reaction sequence involving dihydroxylation of its lone olefin, oxidation of the resultant secondary alcohol under Swern conditions, and elimination using p-TsOH in benzene at 80 °C, we accessed α,β-unsaturated ketone 10 in 45% overall yield. Our hope was that hydrogenation of its lone double bond, as controlled by the single chiral center within the molecule, would generate the needed all cis-stereochemical arrangement of the target. Pleasingly, this expectation was met as compound 11 was produced as a single diastereoisomer in 73% yield using catalytic Pd/C and H₂ in a 3:1 mixture of MeOH and EtOAc containing Et₃N.¹⁶ Subsequent removal of the methyl ethers with BBr₃ then afforded 1. However, as noted in the inset table within Scheme 1, our synthetic material possessed signals for its three aliphatic protons in its ¹H NMR spectrum with multiplicities and coupling constants similar to that reported for the natural isolate, but with very different chemical shifts.

Scheme 1.

Synthetic efforts to prepare the proposed structures of caraphenols B and C indicating the need for a revision.

Although we were unable to obtain a crystalline variant of synthetic 1 to verify its connectivity unambiguously, we note the following in support of our structural assignment: 1) a crystal structure was obtained of the hydrogenation product of 12 (i.e. 13)¹⁷ and following methylation of its phenol, its NMR spectra were identical to those of an over-reduced variant of 11 generated through hydrogenation in the absence of Et₃N, 2) the positioning and coupling constants of the indane protons in 11 are nearly the same as those in synthetic 1, suggesting that no rearrangement or stereochemical change¹⁸ occurred during the BBr₃ deprotection, and 3) extensive HSQC, HMBC, and NOESY experiments confirm the proposed linear structure of 1, including the observation of expected nOe enhancements between the indane protons.¹⁹ Finally, the synthesis of the permethylated variant of caraphenol C (16) through a similar route from permethylated quadrangularin A (14) afforded material with a ¹H NMR spectrum that was similarly distinct from natural caraphenol C in terms of peak placements; deprotection of this material did not prove possible as side-reactions predominated, likely due to Friedel–Crafts chemistry as noted by the conversion of intermediate 15 into 17 (structure confirmed by X-ray analysis). As such, assuming accuracy in our syntheses, alternate structures for caraphenols B and C were needed.

Our hypothesis, as hinted during the opening discussion, was a trans,trans arrangement of substituents which would alter the stereochemistry in 1 and 2 at the starred carbons indicated in Figure 1. Despite this seemingly profound change in terms of how aryl rings would be situated, we believed that these materials would have spectral properties consistent with the data sets of the natural isolates. For example, the original isolation team based their stereochemical assignment on three separate nOe enhancements between the indane ring protons,⁸ the same three we also observed with our synthetic 1; our own molecular mechanics optimizations, however, indicated that these nOe enhancements could also occur in trans-oriented materials. Moreover, our global experience with rings of this type in the resveratrol class has revealed that coupling constants of indane protons cannot be correlated directly to cis- or trans-arrangements as is possible in many cyclohexane systems.

As such, we set out to prepare these materials as well as other diastereomeric variants to provide full confirmation for this alternate structural hypothesis. Our route to our revised structure for caraphenol B (i.e. 22) is shown in Scheme 2, wherein initial operations afforded compound 20 in 5 standard chemical steps²⁰ from protected paucifloral F (18)^7ab by way of aldehyde 19. The stereochemistry of this material (20) could be isomerized to the all-trans arrangement of 21 through the action of NaOMe in MeOH at 65 °C, providing a compound with coupling constants and proton positionings similar to that reported for natural caraphenol B. Upon deprotection of 21 with BBr₃, however, we obtained a 1:4.5 ratio of 22 and 23; a similar outcome occurred following deprotection of 21, with a 1:2.8 ratio of 22 and 23 obtained in higher yield (76% combined, separable by HPLC or careful preparative thin-layer chromatography). These outcomes may result from reprotonation of the same intermediate boron-based enolate. Thus far, we have been unable to convert 23 into 22 under any basic conditions, but all spectroscopic data for 22 match that of the natural isolate. As such, though the final yield of 22 is non-optimal,²¹ we identified the structure of caraphenol B.

Scheme 2.

Total synthesis and structural confirmation of caraphenol B as 22.

Finally, application of an analogous route with permethylated isopaucifloral F (24, Scheme 3) provided both 28 and 29 in a ~1:1.8 ratio and more modest final yield,²² with the only difference here being that compound 26 (structure confirmed by X-ray) could not be deprotected successfully²³ and at least one additional uncharacterized side-product was formed. As expected, however, 28 matched all data for natural caraphenol C.

Scheme 3.

Total synthesis and structural confirmation of caraphenol C as 28.

In conclusion, this work has demonstrated that the structures originally reported for two unique natural products within the resveratrol class are in error, and should be revised to the trans-based arrangement²⁴ typical of the entire family (i.e. that of 22 and 28). As such, it remains an open question whether or not Nature really samples any stereochemical diversity within its resveratrol-derived indane cores. Future work is directed towards improving the stereocontrol needed to access the revised structures of caraphenols B and C exclusively and evaluating the biochemical potential of all materials.