Traversing Biosynthetic Carbocation Landscapes in the Total Synthesis of Andrastin and Terretonin Meroterpenes

Abstract

Meroterpenes derived from dimethylorsellinic acid (DMOA) and farnesyl pyrophosphate have attracted much biosynthetic attention, yet only recently have synthetic solutions to any family members appeared. A key point of divergence in DMOA-derived meroterpene biosynthesis is the protoaustinoid A carbocation which can be diverted to either the berkeleyone, andrastin, or terretonin structural classes via cyclase-controlled rearrangement pathways. Herein we show that the protoaustinoid bicyclo[3.3.1]nonane nucleus can be reverted to either andrastin or terretonin ring systems under abiotic reaction conditions. The first total syntheses of members of these natural product families are reported as their racemates.

Graphical abstract

Traversing Biosynthetic Carbocation Landscapes in the Total Synthesis of Andrastin and Terretonin Meroterpenes

Over 100 structurally fascinating, fungal-derived natural products are assembled from 3,5-dimethylorsellinic acid (DMOA) and farnesyl pyrophosphate (FPP).^[1] These meroterpenes, while historically the subject of intense biosynthetic interest,^[1,2] have only very recently seen members succumb to total synthesis efforts.^[3] The andrastins (see 1–6),^[2l,4] terretonins (see 10–15),^[2i,5,6] and their downstream congeners in particular, represent a large subset of DMOA-derived members with interesting biological profiles and unique carbocyclic frameworks (Figure 1).^[1] To date these compounds have evaded successful chemical synthesis attempts.^[7] Herein we report the first total synthesis of members of these families using a strategy that is inspired by nature’s pathway, but uses completely abiotic chemistry. Density functional theory (DFT) calculations have also shed light on the specific molecular interactions responsible for shaping the energetic landscape in the early stages of DMOA-derived meroterpene biosynthesis.^[8]

Figure 1.

Representative andrastin and terretonin-type meroterpenes.^[6]

A key point of divergence in DMOA-derived meroterpene synthesis is the protoaustinoid carbocation (17), which is derived from FPP/DMOA-conjugate 16 via a cyclase-mediated polyene cyclization (Figure 2).^[1b] We estimate that over 80% of all DMOA-derived meroterpenes traverse 17 or a related intermediate. Proton loss of H_a affords protoaustinoid A (18) which serves as a precursor to the berkeleyone, austin, and related families.^[1b] In Aspergillus nidulans, this process is mediated by the cyclase AusL.^[2e] The biosynthesis of the terretonin and andrastin families is generally believed to involve a 1,2-shift pathway prior to proton loss (see 17→19) thus converting the bicyclo[3.3.1]nonane skeleton into a 5,6-fused ring system. In Aspergillus terreus, the Trt1 cyclase mediates selective proton abstraction of H_b forming a trisubstituted alkene,^[2i,2j,2m] while in Penicillium chrysogenum, a methyl proton (H_c) is selectively abstracted by AdrI.^[2l] Two of our groups (T.R.N and T.J.M.) recently prepared protoaustinoid A (18) en route to distinct total syntheses of berkeleyone A.^[3] While these pathways did not directly proceed through carbocation 17, we wondered if 18 (or a suitable synthetic surrogate) could be “reverted” to the andrastin and terretonin ring systems directly in the absence of enzymes, possibly driven by a thermodynamic preference for 5 and/or 8. If feasible, this could potentially open up synthetic access to nearly half of the remaining DMOA-derived members believed to proceed through carbocation 17.

Figure 2.

Key steps in the early stages of DMOA-derived meroterpene biosynthesis showing the divergence of cationic vinylogous acid tautomers (17-enol1 and 17-enol2) to form protoaustinoid A (18), preterretonin A (8), and andrastin E (5). Energy values shown are in kcal/mol and were obtained using Gaussian ‘09 at the mPW1PW91/6-31+G(d,p)//B3LYP/6-31+G(d,p) level of theory.^[9,10]

Although the synthetic strategy described above was computationally predicted to be favorable for the conversion of protoaustinoid A (18) to andrastin E (5) (ΔG = -2.0 kcal/mol, see SI for details),^[9,10] DFT calculations indicate that formation of the less stable 1,1-disubstituted alkene-containing preterretonin A (8) is unfavorable by 5.9 kcal/mol. Analysis of the carbocation intermediates believed to be involved in these conversions (i.e. 17 and 19), revealed that the tautomeric form of the vinylogous acid functionality determines whether the protoaustinoid or andrastin framework is energetically favored. For one pair of regioisomers (17-enol1 and 19-enol1), the two different ring systems are comparable in energy (ΔG = +0.4 kcal/mol) and kinetically accessible (ΔG^‡ = +9.1 kcal/mol). However, the regioisomeric vinylogous acid 17-enol2 presents an exergonic pathway to the andrastin 5,6-fused bicyclic framework, and 19-enol2 (ΔG = -8.4 kcal/mol), by a nearly barrierless pathway (ΔG^‡ = +0.8 kcal/mol for TS-enol2). The basis for the disparate stability of these isomers is derived from greater stabilization of the carbocation by the neighboring carbonyl oxygen lone-pair in 19-enol2 than the more distal carbonyl in 19-enol1 (see SI for details). Intriguingly, these theoretical findings suggest that controlling tautomeric state may be an effective molecular approach by which biosynthetic machinery can influence the chemical fate of the proposed cationic intermediates.

While these calculations indicate the facility of the rearrangement, surprisingly we have had difficulty in forming either preterretonin (8) or andrastin E (5) via treatment of protoaustinoid A (18) (or its corresponding A-ring ketone) with acid, wherein tautomerization between enol isomers is expected to occur. Under a variety of acidic conditions, only decomposition or complex, unidentified product mixtures have resulted. We initially attributed these failures to the difficulty in protonating the exocyclic olefin over several more basic functional groups combined with the sensitivity of the 1,3-diketone. Seeking alternative chemoselective methods for carbocation generation, we were drawn to recent work by Shigehisa and co-workers who reported an exceedingly mild, Co-catalyzed hydroalkoxylation reaction via an oxidative variant of the classic Mukaiyama hydration process.^[11,12] Substrate 21 was prepared in one-step from tetracyclic building block 20, an intermediate we have prepared in gram quantities en route to the synthesis of berkeleyone A (Scheme 1).^[3a] Subjecting 21 to modified Shigehisa conditions (cat. Co(II), PhSiH₃, N-fluoropyridinium salt (F⁺) oxidant) yet omitting the alcohol trapping agent cleanly promoted the desired rearrangement.^[13] Notably, we have been unable to elicit this bond rearrangement on 21 using a variety of Brønsted acids, although the corresponding regioisomeric vinylogous ester of 21 gives detectable quantities of rearranged product.^[14] When one equivalent of phenylsilane and N-fluoropyridinium salt were employed, the andrastin framework was favored over the exocyclic, terretonin-type alkene system (~3:1 ratio of 22:23 along with recovered 21). Increasing the equivalents of F⁺ oxidant and phenylsilane to 2.5 allowed for very clean formation of 22 as essentially a single olefin isomer in 90% isolated yield. The structure of the newly minted 5,6-fused ring system was confirmed by single crystal X-ray diffraction and following Krapcho-type demethylation (LiCl, DMSO, Δ), (±)-andrastin D (4) was furnished in 77% yield thus completing the first total synthesis of a member of this meroterpene class.

Scheme 1.

Total syntheses of andrastin D (4), preterrenoid (7), terrenoid (9), and terretonin L (10). Reagents and conditions: a) PCC (4 equiv), DCM, 25 °C, 2 h, 91%; b) Co(II)-cat. (5 mol %), PhSiH₃ (2.5 equiv), F⁺ (2.5 equiv), PhH, rt, 2 h, 90%; c) LiCl (40 equiv), DMSO, 120 °C, 2 h, 77%; d) Co(II)-cat. (10 mol %), PhSiH₃ (2.5 equiv), TsCl (1.5 equiv), MeOH, rt, 3 h, 43% (85% BRSM); e) LiCl (40 equiv), DMSO, 120 °C, 1.5 h, 68%; f) MMPP (2 equiv), MeOH, 0 °C, 2 h, 70%; g) NaOMe (0.1 equiv), MeOH, 5 °C, 24 h, 46% + 32% of 24; PCC = pyridinum chlorochromate, MMPP = magnesium monoperoxyphthalate.

While this strategy offers high yielding, selective entry into the andrastins, it presents a challenge for the synthesis of terretonins as the thermodynamic product is obtained. Cognizant of the disparate alkene patterns produced by titanocene-mediated epoxide opening cascades, as compared to their acid-mediated counterparts,^[15] we wondered if a purely radical-based homoallyl-type rearrangement/HAT process could be used to forge 23 preferentially (Figure 3).^[16] After significant exploration, we discovered that when 21 was subjected to conditions developed by Carreira for radical olefin hydrochlorination (cat. Co(II), PhSiH₃, TsCl, MeOH),^[17] we observed formation of 23 in 43% isolated yield with only trace amounts of 22 (<5%). The remaining mass balance was largely recovered starting material and we did not observe the expected tertiary chloride product.^[18] We suspect that the steric encumbrance surrounding radical intermediates 24–26 precludes efficient trapping with TsCl and a final hydrogen atom transfer (HAT) from the less-hindered methyl position ensues preferentially forming 23.^[15,19,20] We also assume that under the more powerfully oxidizing Co(II)/F+ conditions, radical 26 is converted into carbocation 27 which is ultimately funneled into 22 via loss of a proton.^[11]

Figure 3.

Abiotic radical-based HAT isomerization process

Demethylation of 23 (LiCl, Δ) afforded preterrenoid (7), thus allowing us to explore late-stage oxidative chemistry potentially leading to the hallmark terretonin lactone ring systems which are found in several permutations (see colored rings, Figure 1).^[2i] Stereoselective oxidation of 7 with magnesium monoperoxyphthalate (MMPP) afforded terrenoid (9) as a single diastereomer whose structure was confirmed by single crystal X-ray diffraction. Treating this key biosynthetic precursor with catalytic quantities of NaOMe in MeOH afforded (±)-terretonin L (10) as the major product (46%), thus formally accomplishing the same retro-Claisen/esterification cascade as the putative hydrolase Trt14 in Aspergillus oryzae.^[2i] Notably a single diastereomer was obtained, likely due to a preference for the newly-formed methyl stereocenter to reside in the pseudoequatorial position away from the top face of the congested polycyclic ring system. Interestingly, a ring-opened ester assigned as 28 was also isolated in 32% yield; this structure would correspond to 6,7-deoxyterretonin G (see 15, Figure 1). Abe and co-workers did not observe a product of this type when reacting terrenoid (9) with Trt14, indicating the chemoselectivity of this enzyme.

In summary, we have examined early, fundamental steps in DMOA-derived meroterpene synthesis both experimentally and computationally. By initially analyzing the energetic landscape of the system, we realized that a biomimetic 1,2-shift should be feasible in the absence biosynthetic machinery and by investigating abiotic reaction conditions, we demonstrate that selective formation of either andrastin or terretonin scaffolds are possible from protoaustinoid bicyclo[3.3.1]nonane skeletons. With a better understanding of the key skeletal forming chemistry in DMOA-derived meroterpene biosynthesis, efforts can now focus on exploring the synthesis of more highly oxidized congeners. Finally, in-depth studies on how the aforementioned cyclase enzymes manipulate these cationic intermediates, particularly how they prevent energetically favorable rearrangements, may lead to new insights in meroterpene enzymology.