A Unifying Bioinspired Synthesis of (−)-Asperaculin A and (−)-Penifulvin D

Abstract

The first syntheses of the isomeric dioxafenestrene natural products (−)-asperaculin A and (−)-penifulvin D are reported. Each target is formed selectively by choice of oxidant in a final divergent bioinspired Baeyer–Villiger (BV) reaction. Density functional theory calculations reveal that electrostatic interactions between the oxidant leaving group and the lactone motif accounts for a reversal of selectivity with H₂O₂/H₃O⁺ compared to peracids. Synthetic features include forging the polycyclic carbon framework with a diastereoselective meta-photocycloaddition biased by an ether substituent at the aryl α-position. The encumbered tertiary alcohol was installed by cyanation of a ketone intermediate followed by nonaqueous hydrolysis of the resulting delicate cyanohydrin.

Biosynthesis remains a rich source of inspiration for discovering new strategies and tactics in chemical synthesis.^1,2 Guided by this view, we were drawn to the isomeric triquinanes asperaculin A (3)³ and penifulvin D (2)⁴ (Scheme 1). These daunting structures present three consecutive quaternary stereogenic centers flanked by two tertiary centers across the heart of a tetracyclic dioxa[5.5.5.6]fenestrene motif.^5,6 The penifulvins exhibit potent insecticidal properties, but mechanistic details are still unknown. An elegant study by Zou revealed that the penifulvins originate biosynthetically from a silphinene core via an enzyme-mediated Baeyer–Villiger (BV) pathway.⁷ During the preparation of this manuscript, the same lab extended this mechanism to aspergilanes.⁸ We similarly considered the possibility that asperaculin A and penifulvin D share a biosynthetic origin via a BV reaction, but also, that chemical synthesis of both structures through a divergent biomimetic BV reaction was a feasible prospect. Here, we report the first chemical synthesis of a member of the aspergilanes, asperaculin A, through the postulated BV approach. We also show that a simple change of oxidant reverses the selectivity in favor of the isomeric penifulvin D, a first chemical synthesis of a ring-hydroxylated penifulvin. Reagent and substrate influence on regioselectivity in BV reactions has been studied,⁹⁻¹⁴ but to our knowledge, such levels of reversal of selectivity by choice of simple reagents have not been described. The study aligns with the revised biosynthetic relationship between aspergilanes and penifulvins^3,8 and offers new mechanistic insight into regioselective BV reactions. The concise approach to both targets, and by extension to their designed congeners, enables access to new substrates for enzymatic-modification and mechanistic studies.

Scheme 1. Biosynthesis and Chemical Syntheses of Penifulvins and Aspergilanes from a Silphinene Core (Top); Retrosynthetic Analysis (Bottom).

A synthetic connection between the silphinenes and the penifulvins was realized in pioneering studies by Mülzer and Gaich. Concise asymmetric syntheses of penifulvins A–C¹⁵⁻¹⁷ were accomplished through meta-photocycloadditions to assemble the silphinene cores. Wender and Sirois independently applied the same approach to penifulvin A,¹⁸ and we reported an efficient formal catalytic asymmetric synthesis.¹⁹ The aspergilanes, on the other hand, have withstood attempts of total synthesis. Mehta explored a Pauson–Khand approach,²⁰ and more recently, Chakraborty synthesized des-hydroxy asperaculin A.²¹

At the outset, we identified ketone 1b as the critical gateway to asperaculin A and penifulvin D. Installation of the tertiary alcohol in this structure is challenging; a 3D-projection reveals the alcohol to form part of a triad of 1,3-related pseudo-axial substituents. The proximal methyl group thus shields the α-carboxylate position for late-stage manipulation.²¹ We therefore devised a diastereoselective addition of a carboxyl anion equivalent to ketone 4 to create this motif. The tricyclic 4 was further deconvoluted into a simple meta-photocycloaddition precursor 6. The key meta-photocycloaddition has an almost unrivaled history of enabling concise syntheses.²²⁻²⁵ Using high-energy light as the benign sole reagent, it introduces immense complexity that map onto biological targets from simple substrates. It is, however, also a capricious reaction, whose generality and applicability remain underexplored.

In the context of this synthesis, the utility of a chiral benzyl alcohol substrate like 6 was an outstanding question, though substrates with an oxygen tether²⁶ and a benzylic acetal²⁷ provided encouraging precedence. An attractive feature of the planned approach was that all stereochemical information would originate from 6. We thus commenced with an asymmetric synthesis of this compound (Scheme 2). Direct alkylation of o-methyl acetophenone (7) with prenyl bromide followed by reduction of the resulting ketone with (+)-β-chlorodiisopinocampheylborane (DIPCl) gave 6 in multigram quantities and with an excellent 97.5:2.5 enantiomeric ratio. Unfortunately, irradiation of 6 at 254 nm gave only small amounts of photocycloadducts and a 75:25 preference for the undesired linear isomer. To address selectivity and efficiency issues, we conducted a systematic screening of alcohol protecting groups. Compared to tert-butyldimethysilyl, triisopropylsilyl, triethylsilyl, and tetrohydropyranyl ethers, ethoxy-methoxy (EOM) ether 8 gave a uniquely clean reaction profile when irradiated at 254 nm. A change to 300 nm light produced less byproducts and reviled a 60:40 kinetic preference in favor of the sought angular isomer 9a at low conversions. Allylic strain between the aromatic methyl group and the EOM-ether moreover induces a sufficient conformational bias during exciplex formation to give 9a and 9b as single detected diastereomers. The angular-to-linear product ratio drops with conversion, in part, due to decomposition of the more sensitive angular isomer (see Supporting Information Figures S3 and S4). Optimization of solvent, reaction time, and light source clarified that irradiation in pentane with 300 nm light until 75% conversion (∼70 h) gave the highest yield, 24% of 9a (measured by ¹H NMR spectroscopy using an internal standard). A medium pressure mercury lamp or 254 nm light gave a much faster reaction (∼30 min) but a lower net efficiency (<15% of 9a).

Scheme 2. Synthesis of Asperaculin A and Penifulvin D.

Determined by chiral HPLC.

Determined by ¹H NMR spectroscopy of the crude reaction mixture.

Thermal ellipsoids are drawn at 30% probability. Non-acidic hydrogen atoms are omitted for clarity.

A homogeneous sample of linear 9b could be re-equilibrated to a 24:76 mixture of 9a and 9b by irradiation at 300 nm for 72 h. With ready access to 9a, we turned to a radical ring opening of the vinyl cyclopropane. Because 9a was obtained as an intractable mixture with 8, the reduction was performed directly on the crude photomixture. Benkeser conditions with EtNH₂ as solvent^28,16 gave a clean reduction but produced a 74:26 mixture of isomers 10a/b. We were not able to isomerize 10b to 10a with pTSA or I₂. A switch to MeNH₂ improved the selectivity of 10a (77:23) and an isolated yield of 18% was obtained over two steps from 8. Additives such as t-BuOH or ethylenediamine gave a lower selectivity for 10a. We also explored ring opening of an unprotected 10a. The free alcohol reacted very slowly with extensive side product formation under Benkeser conditions, which underlines a second purpose of the protective group in enabling efficient rupture of the cyclopropane ring. Deprotection of 10a by acid catalysis followed by oxidation of the resulting secondary alcohol with pyridinium dichromate (PDC) then gave ketone 11 in 95% yield over two steps. Isolation of 11 is challenging, even from pentane, due to its volatility. Related to this property, we could not escape noticing its delightful marine and mineral fragrance. With scalable access to ketone 11 in hand, we turned to install the tertiary hydroxyl group.

Cyanide addition using trimethylsilyl cyanide (TMSCN)/ZnI₂²⁹ proceeded cleanly and exclusively from the molecule’s convex side. The orientation of the hydroxyl group is thus inverted compared to 8, but prudent for the synthesis. An isolated yield of 59% for amide 12 reflects the cyanohydrin’s lability to chromatographic purification and protic workup conditions. Attempts to remove the TMS group gave an immediate release of cyanide to reform ketone 11. Steric congestion from the 1,3-related methyl group likely contributes to this instability. To convert the nitrile into a carboxylate, we first evaluated the Parkins catalyst in EtOH/H₂O.³⁰ Stoichiometric amounts of this expensive platinum salt were needed due to poisoning of the catalyst by cyanide released from substrate decomposition. We therefore turned to nonaqueous conditions. Significantly, a rhodium-catalyzed reaction using acetaldehyde oxime as the formal water source gave amide 13.³¹ Further optimization revealed that [Rh(COD)Cl]₂ was as efficient as the previously used Wilkinson’s catalyst but gave a simpler workup. Moreover, it showed that a slow addition of the oxime enabled a higher conversion. Under the optimized conditions, amide 13 was obtained in 42% isolated yield, along with 37% recovered starting material.

With a completed silphinene framework in place, only an increase in the oxidation level stood between amide 13 and the target natural products. First, a Prilezhaev oxidation of the alkene with mCPBA gave epoxide 14 as a single diastereomer in 75% yield. The addition occurs exclusively from the olefin’s open face and thus proceeds without coordination interference from the amide group. Treatment of this epoxide with an excess of pTSA-hydrate then triggered a cascade that removed the silyl group and formed the first oxa-fenestrene intermediate, lactone 15, in 72% overall yield. Oxidation of the secondary alcohol with PDC provided the gateway ketone 1b in quantitative yield and set the stage for exploring BV oxidations.

We first turned to peracid oxidants.³² Pleasingly, mCPBA activated by either acid (TFA, pTSA) or base (NaHCO₃) all gave an ∼95:5 preference for the formation of penifulvin D. The base mediated conditions were higher-yielding, and penifulvin D was obtained in 46% isolated yield. Peracetic acid gave a 90:10 selectivity. The successful formation of penifulvin D is significant, as it represents the first chemical synthesis of a ring-hydroxylated penifulvin, and it provides a simple alternative to enzyme-mediated BV oxidations in this context. We next sought conditions that would enable selective formation also of asperaculin A. Much to our delight, oxidation with H₂O₂ (aq.)³³ activated by pTSA gave a 50:50 mixture of asperaculin A and penifulvin D. The selectivity for asperaculin A could be further increased using the noncoordinating triflic acid. By monitoring the reaction progress, we also found that the ratio increases over time due to the decomposition of the more fragile penifulvin D. Quenching the reaction at full conversion (2 min) gave a 77:23 selectivity in favor of asperaculin A and an isolated yield of 52%. Spectroscopic data for both natural products agreed with the literature values. The structures were also corroborated by single-crystal X-ray diffraction (scXRD) analysis.

To understand the intriguing reversal of selectivity in the BV oxidation, we modeled the reaction pathways^34,35 by density functional theory (DFT) calculations using the m06-2x-d3 functional and the 6-31G** basis set (Scheme 3). As expected, only the Si-face of ketone 1b is accessible for nucleophilic addition. Two Criegee intermediates, CI_asp and CI_pen, related by rotation around the C–O bond were found for each oxidant. For the peracid reaction,³⁴ the lowest paths found to each product respectively proceeded via the closed transition states TS_asp and TS_pen. The calculated ΔΔG^‡₂₉₈ was 1.9 kcal mol^–1, which corresponds to a 96:4 ratio in favor of penifulvin D. The model thus aligns well with experimental data. A longer C···C bond to the migrating carbon in TS_pen than in TS_asp (1.82 Å vs 1.75 Å) reveals a later transition state leading to penifulvin D. Donation of electron density from the oxygen on the migrating carbon into σ*_C–C in TS_pen facilitates migration and contributes to the preference for penifulvin D. The acetal arrangement of penifulvin D is also more favored than the bis-lactone of asperaculin A reflecting in a 5.2 kcal mol^–1 lower free energy (see Supporting Information Table S4 for a brief model study supporting the analysis). For the acid-catalyzed H₂O₂ oxidation, an explicit oxonium ion was needed to find computational reaction pathways with reasonable energy levels. This model replicates the experimentally observed preference for asperaculin A (ΔΔG^‡₂₉₈ = 2.7 mol kcal^–1) and the faster reaction rate with H₂O₂ compared to peracids. The critical factor behind the calculated selectivity is a stabilization by a strong electrostatic interaction between a proton on the oxonium ion and the lactone motif in TS_asp. Such interactions are not available in TS_pen. In absolute terms, the selectivity is overestimated by ∼1.8 kcal mol^–1, likely connected to the difficulty of accurately accounting for the energy of solvation for protons.

Scheme 3. DFT Investigation of BV Regioselectivity.

The peracid pathway (top) is modeled with AcOOH as the oxidant. The hydrogen peroxide pathway (bottom) is modeled with H₂O₂/H₃O⁺ as the oxidant. TS = transition state. CI = Criegee intermediate. Energies are calculated at 298 K.

In summary, a unifying bioinspired synthesis of (−)-asperaculin A and (−)-penifulvin D via divergent BV oxidations was completed in 13 steps from abundant o-methyl acetophenone. Difficult isolation of volatile intermediates and arduous purification contribute to reduced yields for certain steps. On the other hand, the approach balances such drawbacks with simple starting material, a dramatic increase in molecular complexity in the photochemical step, and a versatility to reach several targets. DFT calculations on the final BV oxidation with peracids reveal that donation of electron density from the oxygen on the migrating carbon facilitates cleavage of the C–C bond and contributes to the preference for penifulvin D. Electrostatic interactions between the oxidant leaving group and the lactone motif in turn explain the reversal of selectivity with H₂O₂/H₃O⁺. It seems plausible that similar effects are at play also in enzymatic systems. Additional synthetic highlights include a diastereoselective meta-photocycloaddition biased by an ether substituent at the aryl α-position that broadens the scope of this remarkable reaction. The successful hydrolysis of a sensitive cyanohydrin by rhodium catalysis supports a wider utility with complex substrates. From a strategic perspective, we note that the developed pathway maps onto cyclase and oxidase phases resonant with Nature’s approach to assembling complex molecules.³⁶ Efforts to further generalize divergent BV reactions and mechanistic studies on the biological activity of aspergilanes and penifulvins are ongoing in our laboratory and will be reported in due course.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.orglett.1c00955.

• Synthetic procedures, characterization data, and copies of ¹H and ¹³C NMR spectra for all isolated compounds; Optimization and kinetic data for photochemistry; Crystallographic details; DFT procedures and energies; Comparison of spectroscopic data between natural and synthetic material (PDF)

• DFT geometries (XYZ)

Accession Codes

CCDC 2042599–2042600 contain the supplementary crystallographic data for this paper. These data can be obtained free of charge via www.ccdc.cam.ac.uk/data_request/cif, or by emailing data_request@ccdc.cam.ac.uk, or by contacting The Cambridge Crystallographic Data Centre, 12 Union Road, Cambridge CB2 1EZ, UK; fax: +44 1223 336033.

The authors declare no competing financial interest.