A Concise Synthesis of Berkelic Acid Inspired by Combining the Natural Products Spicifernin and Pulvilloric Acid

Abstract

We describe a concise synthesis of the structurally novel fungal extremophile metabolite berkelic acid – an effort leading to an unambiguous assignment of C22 stereochemistry. Our synthetic approach was inspired by the recognition that berkelic acid displays structural characteristics reminiscent of two other fungal metabolites, spicifernin and pulvilloric acid. Based on this notion, we executed a synthesis that features a Ag-catalyzed cascade dearomatization-cycloisomerization-cycloaddition sequence to couple two natural product inspired fragments. Notably, a spicifernin-like synthon was prepared with defined C22 stereochemistry in seven steps and three purifications (24–28% overall yield). A potentially useful anti-selective conjugate propargylation reaction was developed to introduce the vicinal stereodiad. An enantioconvergent synthesis of the other coupling partner, the aromatic precursor to pulvilloric acid methyl ester, was achieved in eight steps and 48% overall yield. The total synthesis of berkelic acid and its C22 epimer was thus completed in 10 steps longest linear sequence and 11–27% overall yield.

Berkeley Pit Lake in Montana, a 30 billion gallon flooded copper mine and largest superfund clean-up site in the United States, is an unlikely source of structurally novel natural products. Yet, this highly acidic, heavy-metal contaminated poisonous broth harbours microbial life, including an extremophilic Penicilium fungus that produces the unique tetracyclic chroman/isochroman spiroketal berkelic acid (Fig. 1). This compound, isolated by Stierle et al in 2006, was found to possess selective activity against the human ovarian cancer cell line OVCAR-3 (GI₅₀ 91 nM) and moderate inhibitory activity against the matrix metalloproteinase MMP-3 (1.87 µM) and the cysteine protease caspase-1 (98 µM).¹

Figure 1.

Original and revised structures of berkelic acid.

The stereochemistry of berkelic acid was originally assigned as shown in structure 1 on the basis of NMR experiments, although the configuration at the quaternary stereocenter C22 and absolute configuration were left undetermined. Recently, the Fürstner group reported studies leading to a revision of the relative stereochemistry of berkelic acid as shown in structure 2 through total synthesis of the corresponding methyl ester.² Through elegant synthetic, NMR, and crystallographic studies, they further revealed that the originally proposed relative stereochemistry does not represent a thermodynamic minimum because of a key syn-periplanar interaction between the C25 methyl substituent and C16 methylene group.³ Subsequently, Snider and coworkers reported their total synthesis of berkelic acid, which established its absolute configuration as shown in 2, and putatively assigned the stereochemistry at the quaternary center as C22-S.^{4, 5} Herein, we wish to report a concise synthesis of the two C22 epimers of berkelic acid (2) that fully corroborates the revised stereochemistry and unambiguously resolves the remaining issue of C22 stereochemistry.

Our approach was inspired by the recognition that the original assigned berkelic acid structure 1 represents a formal combination of the natural products spicifernin⁶ (3) and pulvilloric acid⁷ (4, Scheme 1).⁸ Based on this notion, we developed a strategy that would emulate this hypothetical combination and designed a suitable spicifernin-like synthon such as enolether 7,⁹ available via metal-catalyzed cycloisomerization.¹⁰ Participation of this material in a [4+2] cycloaddition with the ortho-quinone methide tautomer 5 of pulvilloric acid (4) would deliver spiroketal 1. It did not escape our attention that this chemistry could potentially be implemented with minimal oxidation state adjustments.¹¹

Scheme 1.

Synthetic Strategy

Given the ambiguity related to the absolute stereochemistry at C22, we opted for a synthesis that would enable access to the two C22 epimers of fragment 6 (Scheme 2). Starting with commercially available methyl 2-ethyl-3-oxobutanoate (8), the corresponding (L)-^tBu valinate-derived enamine 9 was prepared (82–88% yield) and alkylated with methyl iodide to afford the α-quaternary substituted imine derivative 10 with high stereoselectivity (> 15:1 dr).¹² The absolute stereochemistry at C22 was determined by a single crystal X-ray diffraction analysis of the cyclic 4-bromo-2-nitrophenylhydrazone derivative 11.¹³ Continuing with the synthesis, hydrolysis of crude imine 10 was followed by a titanium tetrachloride-mediated dehydrative aldol reaction with (4-methoxybenzyloxy)ethanal yielding enone 12 in 42–45% yield (3 steps) from enamine 9. We explored various options to introduce the α-methyl-substituted propargyl unit and settled on an approach that entails a conjugate addition of a metalated propargyl/allenyl species to enone 12. Although the stereoselective propargylation of aldehydes is well precedented, we could find only one example of the corresponding conjugate addition in the literature.¹⁵ After substantial experimentation, we found that addition of enone 12 to a cold (−78 °C) dark red solution of a cuprate derived from adding (4-(trimethylsilyl)but-3-yn-2-yl)lithium to a suspension of CuBr·SMe₂ in TFEF (−78 °C) efficiently effected the desired conjugate propargylation.¹⁶ Although the anti-selectivity was acceptable, the stereogenic quaternary center did not impart any facial selectivity, leading to an inseparable equimolar mixture of R,S- and S,R-diastereomers 13a and 13b.¹⁷ As such, this crude mixture was carried forward by treatment with methanolic potassium carbonate, followed by oxidative deprotection to yield compounds 14a,b. Proton NMR analysis of chromatographically homogeneous material (with correct elemental analysis),¹³ isolated in 70% yield from enone 12, indicated a complex mixture of equilibrating lactols and open-chain isomers. The corresponding mixture of enantiomers ent-14a,b was prepared from the (D)-^tBu valinate-derived enamine ent-9, or cheaper, by switching the additive from THF to HMPA during the alkylation of (L)-^tBu valinate-derived enamine 9.¹²

Scheme 2.

Synthesis of alkyne fragmenta

^aReagents and conditions: (a) L-^tBu-Val-NH₂, BF₃·Et₂O, PhH, reflux (82–88%); (b) LDA, PhMe, −78 °C, 1 h, THF (2.5 equiv.), −78 °C, 3 h, MeI, −78 °C, 17 h; (c) 1 M aq. HCl/THF (1:1), rt, 1 h; (d) TiCl₄, THF, 4 ÅMS, 0 °C, 30 min, NEt₃, −78 °C, 1 h, PMBOCH₂CHO, −78 °C, 1.5 h, rt 1.5 h (42–45%, 3 steps); (e) 1-trimethylsilyl-1-butyne, ^sBuLi, THF, −78 °C, 2 h, CuBr·SMe₂, −78 °C, 1 h, then add 12, −78 °C, 24 h; (f) K₂CO₃, MeOH, rt, 2 h; (g) DDQ, CH₂Cl₂/H₂O (7:1), rt (70%, 3 steps); (h) 4-Br-2-NO₂PhNHNH₂·HCl, EtOH, reflux, 2 d (40%, 2 steps).

A concise enantioconvergent synthesis of the precursor to pulvilloric acid 4 begins with a cross-coupling of triflate 16 – obtained from commercially available methyl 2,4,6-trihydroxybenzoate 15 in 91% yield – with 1-heptenylboronic acid to afford styrene derivative 17 (91% yield, Scheme 3). Installation of the homobenzylic alcohol was best achieved via oxidation with mCPBA of the MOM protected derivative of 17, followed by benzylic epoxide reduction. Racemic alcohol 18 was thus obtained in 76% yield for the three-step sequence. Screening of a set of enzymes to mediate a kinetic resolution identified a lyophilized formulation of a lipase from Alcaligenes sp. to effect the transesterification (vinyl acetate) with high enantioselectivity at ∼50% conversion.¹⁸ Alcohol 19 and acetate 20 were isolated in 51% and 46% isolated yield and 93% and 95% ee respectively. Alcohol 19 was easily recycled to the desired acetate 20 via Mitsunobu esterification (78% yield). Simultaneous removal of the protecting groups was achieved via stirring in acidic methanol (quant.). Condensation of the resulting triol 21 with triethyl orthoformate according to an adapted procedure described for the synthesis of pulvilloric acid (4) yielded isochroman acetal 22 (99% yield), the precursor to pulvilloric acid methyl ester. Although it has been reported that the carboxylic acid corresponding to 22 will yield pulvilloric acid (4) upon removal of ethanol under ultra high vacuum, ^7d we opted to explore Lewis acid-promoted in situ dearomatization of 22 as described below.

Scheme 3.

Synthesis of aromatic fragmenta

^aReagents and conditions: (a) Tf₂O, lutidine, CH₂Cl₂, 0 °C, 16 h (91%); (b) ⁿC₅H₁₁CHCHB(OH)₂, 5% Pd(dppf)Cl₂, K₂CO₃, THF/H₂O (10:1), Δ, 2.5 h (91%); (c) MOMCl, ⁱPr₂NEt, CH₂Cl₂, 0 °C → rt, 18 h; (d) mCPBA, CH₂Cl₂, rt, 5 h (84%, 2 steps); (e) Pd/CaCO₃, H₂, MeOH, rt, 20 h (90%); (f) Lipase (Alcaligenes sp., lyophilized), MTBE, 4 ÅMS, vinyl acetate, rt, 7 d; (g) 0.25 M HCl in MeOH, rt, 15 h (100%); (h) (EtO)₃CH, TFA, rt, 15 h (99%); (i) PPh₃, DEAD, HOAc, PhMe, rt, 7 h (78%).

As noted above, we were intrigued by the possibility to effect in situ dearomatization of lactol 22 to pulvilloric acid methyl ester under conditions that would allow tandem C–C bond formation with spicifernin-like fragment 14. We speculated that Ag⁺ would have a proper balance of hard Lewis acidic properties to induce removal of ethanol from 22, and sufficient alkynophilic character to induce cycloisomerization of alkynol 14 to enolether 23 (Scheme 4).¹⁹ Gratifyingly, stirring a solution of lactol 22 (1 equiv.) and AgSbF₆ (3.5 equiv.) in the presence of alkynols 14a,b (2.6 equiv.) resulted in the formation of methyl berkelate 26 (from 14a) and four additional diastereomeric berkelates 25 (from 14b)²⁰ in a ratio of ∼ 6:4, indicating a slight kinetic preference for the formation of 26. We hypothesize that AgSbF₆ instigated a reaction cascade involving: (1) in situ formation of ortho-quinone methide 24,²¹ (2) cycloisomerization of 14 to enolether 23, and (3) coupling via [4+2] cycloaddition.²²

Scheme 4.

Synthesis of berkelic acid (2) and C22-R diastereomer 27a

^aReagents and conditions: (a) 14a,b or ent-14a,b (2.6 equiv.), 22 (1 equiv.), Et₂O, rt, 2 h; (b) (Bu₃Sn)₂O (35 equiv.), PhMe, Δ, 8 h for 2, 14 h for 27.

Because the methyl berkelate diastereomers were not separable by chromatography, they were carried forward as a crude mixture. Although Fürstner and coworkers disclosed that they could not identify conditions for the selective deprotection of the methyl benzoate in the presence of the aliphatic methyl ester,² we found that (Bu₃Sn)₂O in toluene accomplished the task when the reaction was interrupted at partial conversion.²³ Berkelic acid 2 was thus isolated in 35% isolated yield (from lactol 22) at 70% conversion and 46% yield after one recycling (77% based on theoretical maximum yield). Prolonged reaction times resulted in the formation of decarboxylated product 28 (∼4:1 mixture of C22 diastereomers). The corresponding C22-R diastereomer 27 was prepared via an identical sequence from ent-14a,b and lactol 22 in 26% yield. Only C22-S diastereomer 2 displayed spectral data fully congruent with natural berkelic acid,¹ thus establishing the complete stereostructure of this unique natural product for the first time. The rotation of synthetic 2 ([α]_D = −76.7, c = 0.06 in MeOH) agreed with those for natural ([α]_D = −83.5, c = 0.0113 in MeOH)¹ and Snider’s synthetic berkelic acid ([α]_D = −115.5, c = 0.55 in MeOH).⁴

In conclusion, we have achieved a highly convergent and efficient synthesis of berkelic acid that fully establishes the stereochemistry at C22 in a longest linear sequence of 10 steps and 11–27% overall yield from commercially available starting materials. Notably, we identified a unique Ag-catalyzed cascade dearomatization-cycloisomerization-cycloaddition sequence to couple two natural product inspired fragments, and a potentially useful anti-selective conjugate propargylation reaction.