Enantioselective Total Synthesis of (–)-Psiguadial A

Abstract

The first enantioselective total synthesis of the antiproliferative natural product (–)-psiguadial A is reported. This approach features the enantioselective synthesis of a complex tricyclic terpenoid precursor, the union of that precursor with a polyketide component by an enolate-ortho-quinone methide coupling reaction to form a highly congested carbon–carbon bond, and an acid mediated intramolecular hydration ring-closure leveraging a fully substituted alkene to generate the unique oxepane core structure of the natural product.

Graphical Abstract

A biomimetic approach harnessing reactive species enabled the enantioselective synthesis of the complex meroterpenoid psiguadial A through an enolate–ortho-quinone methide (oQM) umpolung strategy and late-stage cationic ring closure. The final fully decorated aromatic core was successfully synthesized by adaptation of modern aryl methyl ether deprotection and formylation methodologies, establishing efficient access to this class of natural products.

The extracts of the leaves and bark of the common guava (Psidium guavajava [Myrtaceae]) have yielded over 50 distinct natural products to date with myriad biological properties including anti-inflammatory, anti-bacterial, and anticancer activities.^1–4 Psiguadial A (1) is a polycyclic meroterpenoid that was isolated along with its congener psiguadial B in 2010 by Ye and coworkers and is a potent antiproliferative compound (HepG2 IC₅₀ = 61 nM).⁵ Meroterpenoids are fascinating compounds that represent the merger of multiple metabolic pathways that give rise to structures that offer opportunity for innovative chemistries.^6–7 The cellular mechanisms that underlie the biological activity of the psiguadials and related compounds have not been fully characterized, and together with their fascinating structures, these natural products present great opportunity for study toward new therapeutics.

Reisman and coworkers have recently accomplished an elegant total synthesis of psiguadial B,^8–9 however psiguadial A (1) presents a starkly different synthetic challenge, featuring a cis fused cyclopropane and is unique among the many meroterpenoids isolated to date in that the terpenoid moiety (Figure 1, A/B/C ring system) is joined to the phloroglucinol unit (Figure 1, E ring) via a highly functionalized oxepane ring (Figure 1, D ring). This unique core presents a dense array of contiguous stereogenic centers with high hydrocarbon character that lack convenient functional groups that might serve as synthetic handles. In addition, the C1’–C4 and C1–O bond connections feature fully substituted carbons that are particularly challenging to generate synthetically. Several biomimetic approaches to meroterpenoids that feature phloroglucinol moieties have been reported, often leveraging β-caryophyllene which contains a trans fused cyclobutene.^10–14 In a broad biomimetic study of meroterpenoids, psiguadial A was described by Cramer and coworkers as a minor constituent (~1%) in mixtures resulting from cationic cascade and rearrangement sequences¹⁵ that are reminiscent of the proposed biosynthetic origins of several meroterpenoids.¹⁶ To date, no total synthesis of psiguadial A has been accomplished.

Figure 1.

Retrosynthetic analysis of psiguadial A (1).

While extremely challenging, we viewed both the C1’–C4 and C1–O bonds as strategic disconnections. We envisaged that these might be formed by an enolate-ortho-quinone methide (oQM) coupling reaction and an acid-mediated cyclization, respectively (Figure 1). The late-stage excision of the carbonyl function (4 → 3, Figure 1) would set the stage for the cationic ring closure (3 → 2, Figure 1), though we recognized the potential for unproductive rearrangement and elimination chemistries that might intervene in that process. We recently developed an enolate-ortho-quinone methide coupling methodology^17,18 with an eye toward the psiguadials. We report here the first total synthesis of psiguadial A leveraging this powerful strategy for the formation of highly congested carbon–carbon bonds.

With those thoughts in mind, we targeted the scaffold 2 for our approach with the installation of the final oxygen atom and the formyl groups as the endgame in our sequence. The critical C1–O bond in intermediate 2 would be formed in a biomimetic acid-mediated etherification via the alkene 3. The alkene was envisioned to arise from the redox adjustment of the enone 4, which we would generate directly from the doubly benzylic halide 5 and the silyl enolether 6 utilizing our enolate-ortho-quinone methide coupling reaction.¹⁷

We have long been interested in carbonyl-enabled reactivity and the chemistries unlocked by latent aromatic compounds via umpolung strategies. Our group previously reported an enolate-ortho-quinone methide coupling reaction in which silyl enol ethers and silyl ketene acetals are joined smoothly and diastereoselectively with silyl protected phenolic benzyl chlorides under the action of anhydrous fluoride to give a variety of functionalized β–(2-hydroxyphenyl)-carbonyl compounds that are not accessible by alkylative methods with conventional metal enolates and alkyl halide electrophiles (Scheme 1a).¹⁷ In these operationally simple reactions, starting materials are mixed in cold dichloromethane (–78 °C) whereupon a solution of anhydrous tetramethylammonium fluoride (TMAF) in dichloromethane is added via cannula. Under the action of fluoride, both the nucleophilic and electrophilic reaction partners are revealed in situ, and join smoothly to afford the alkylated products. The stereochemical outcome of these reactions was consistent with an open transition structure in which the C–O bond dipoles are opposed and the oxygen atoms are not bridged by Lewis acidic species.

Scheme 1.

a. Initial studies from our laboratory on enolate-ortho-quinone methide coupling reactions. b. Model scaffolds relevant to the psiguadials. [a] Product was obtained as a mixture of diastereomers with respect to the benzylic stereogenic center (marked with an asterisk). TMS = trimethylsilyl, TBS = tert-butyldimethylsilyl, TMAF = tetramethylammonium fluoride, MOM = methoxymethyl.

In pursuit of psiguadial A, we applied this approach in the context of model systems that more closely mirrored the structure of the natural product and other related meroterpenoids (Scheme 1b, 13-15). The structures of 14 and 15 were confirmed unambiguously by X-ray crystal diffraction, as was a p-nitrobenzoate derivative of 14 (see Supporting Information).¹⁸ In these more complicated contexts, we observed stereoselective alkylation reactions that established the critical and highly congested C1’–C4 bond and the relative stereochemistry at C5, C4, and C1’ that would be required in the natural product. We also demonstrated that suitably substituted diphenylmethane units would be amenable to the synthetic approach as designed. With these results in hand, we set about generating the complex terpenoid fragment required to generate the fully decorated core of psiguadial A.

The diphenylmethane unit 5 is accessible in a handful of steps from commercial 3,5-dimethoxybromobenzene (16), and was employed in our preliminary work affording the model substrate 15 (Scheme 2a). Formylation of 16 via the Villsmeier protocol (POCl₃ and N,N-dimethylformamide [DMF]) and protecting group exchange efficiently provided the TBS-protected substituted benzaldehyde 17.^20–21 Nucleophilic addition of phenylmagnesium bromide afforded the secondary alcohol 18 which was converted to the key electrophile 5 in nearly quantitative yield under the action of thionyl chloride. We elected to pursue the terpenoid fragment of psiguadial A starting from (+)-3-carene (9), adapting a sequence toward enone 8 described by Carreira and coworkers thus integrating the dimethylcyclopropane moiety at the outset (Scheme 2b).^22–23 Ozonolysis of (+)-3-carene (9) and capture of the resultant aldehyde in situ as the corresponding dimethyl acetal afforded the ketone 19 in 96% yield. Conversion of 19 to corresponding silylenolether 20 was achieved in 99% yield by deprotonation of 19 with lithium N,N-diisopropylamide (LDA) and capture of the resultant enolate with chlorotrimethylsilane (TMSCl). Under the action of tin(IV) chloride, the silylenolether 20 underwent an intramolecular Mukaiyama-type aldol ring closure, and subsequent acid-mediated elimination of the resultant methyl ether afforded the cycloheptenone 8 in 47% yield over the two steps.^24–26 Allylation of the lithium enolate derived from 8 on the convex face with allyl bromide afforded the alkylated ketone 21 in 55% yield as a single diastereomer.²⁷ Conjugate reduction of 21 with L-selectride and capture of the resultant enolate with iodomethane afforded the complex ketone 7 in 66% yield, accompanied by a small amount of the secondary alcohol resulting from direct reduction of the ketone (not shown) which could be recycled back to 21 after reoxidation. Adapting Tanaka’s work,²⁷ a Wacker oxidation followed by an intramolecular aldol condensation under the action of sodium tert-amyloxide in benzene at 23 °C afforded the bicyclic α,β-unsaturated ketone 22 in 56% yield over two steps. Regioselective installation of the methyl group was achieved by treatment of the ketone 22 with LDA followed by iodomethane in cold tetrahydrofuran (THF) to give an alkylated ketone (not shown); the alkylated ketone was selectively enolized again using LDA and treated with chlorotrimethylsilane to give the corresponding silylenol ether 6 as a single isomer in quantitative yield. The silylenol ether is sensitive to hydrolysis but bench stable for hours and does not undergo unproductive [1,5]-sigmatropic rearrangements provided the material is kept at or below 23 °C (room temperature).²⁸

Scheme 2.

a. Preparation of the polyketide synthesis component. b. Preparation of the sesquiterpenoid synthesis component. Conditions: (a) POCl₃, N,N-dimethylformamide (DMF), 25 → 100 °C, 83%. (b) BBr₃, CH₂Cl₂, −78 → 25 °C, 90%. (c) tert-butyldimethylsilyl trifluoromethanesulfonate (TBSOTf), NEt₃, CH₂Cl₂, 0 → 22 °C, 95%. (d) PhMgBr, diethylether (Et₂O), 0 °C, 78%. (e) SOCl₂, pyridine (pyr), Et₂O, 0 °C, 98%. (f) O₃, (CH₃)₂S, 8:1 CH₃OH:CH₂Cl₂, −78 → 25 °C, then CeCl₃ • 7H₂O, HC(OCH₃)₃, 25 °C, 96%. (g) lithium N,N-diisopropylamide (LDA), tetrahydrofuran (THF), −78 °C; chlorotrimethylsilane (TMSCl), −78 → 25 °C, 99%. (h) SnCl₄, CH₃CN, −20 °C, 55%. (i) camphorsulfonic acid (CSA), 4 Å MS, THF, 60 °C, 86%. (j) LDA, THF, −78 °C; allyl bromide, hexamethylphosphoramide (HMPA), −78 → 25 °C, 55%. (k) L-selectride, THF, 25 °C, then CH₃I, 66%, 10:1 dr. (l) PdCl₂ (10 mol%), CuCl, O₂, 5:1 DMF:H₂O, 66%. (m) NaOt-Am, benzene (PhH), 25 °C, 85%, 7:1 dr. (n) LDA, THF, −78 °C; CH₃I −78 → 0 °C, 83%. (o) LDA, THF, −78 °C; TMSCl, −78 → 0 °C, 99%.

With efficient access to both coupling partners established, we set about forging the critical C1’–C4 bond. Treatment of a mixture of 5 and 6 with TMAF in cold dichloromethane (–78 °C) led to the formation of the desired adduct 23 in 66% isolated yield in 1:1 dr with respect to the C1’ stereogenic center (Scheme 3). The observed products are consistent with the expected π-facial selectivity with respect to the enolate derived from 6, and we attribute the lack of stereochemical control at C1’ to compromised geometrical control of the putative oQM intermediate derived from 5. We dedicated considerable effort toward influencing oQM geometry to improve upon the stereochemical outcome of these reactions with respect to C1’; this outcome was anticipated however based on our work with the model systems 13–15. We explored electrophiles analogous to 5 with alternative substitution patterns and protecting groups about the aromatic rings but the stereochemical outcomes of these alkylations were not improved further. Nevertheless, the enolate-oQM coupling reaction is powerfully enabling; attempts to achieve formation of the congested C1’-C4 bond by fragment coupling of conventional metal enolates derived from 6 (e.g., treatment of substrates like 22 with LDA, lithium (bistrimethylsilyl)amide [LHMDS], and other common bases) and addition of 5 returned starting materials unchanged or resulted in decomposition. In any case however, these results are perhaps fortuitous: we plan to complete and study the bioactivity of C1’ epimers of multiple meroterpenoid natural products in due course.

Scheme 3.

Completion of the core structure of (–)-psiguadial A. Conditions: (a) tetramethylammonium fluoride (TMAF), CH₂Cl₂,–78 °C, 66%, 1:1 dr @ C1’. (b) NaH, (CH₃)SO₄, THF, 25 °C, 74%. (c) di-iso-butylaluminum hydride (DIBAl-H), THF, −78 °C. (d) camphorsulfonic acid (CSA), CHCl₃, 25 °C, 75% (two steps). (d) 25, AcOH, CH₂Cl₂, 40°C, 58% yield @ 80% conv. (e) tris(pentafluorophenyl)borane (BCF), pentamethyl disiloxane (PMDS), hexanes 25 °C. (f) tetra-n-butylammonium fluoride (TBAF), THF, 25 °C; H₂SO₄, H₂O/THF, 60 °C, 95%. (g) tert-butyldimethylsilyl trifluoromethanesulfonate (TBSOTf), NEt₃, THF, 25 °C, 66%. (h) nBuLi, THF, −78 °C; B(OCH₃)₃, −78 °C; H₂O₂, −78 → 25 °C, 93%. (i) tetra-n-butylammonium fluoride (TBAF), THF, 25 °C. (j) formamidine acetate (FAA), Ac₂O, THF, 50 °C, 38%, two steps.

In principle, the C1–O bond could arise from a straightforward intramolecular oxa-Michael reaction^29–30 joining the phenol with the proximal cyclopentenone (Scheme 4). We directed considerable effort toward this strategy with 23 as well as model systems 13–15 under acidic and basic reaction conditions, however we observed no evidence of conjugate addition; starting materials were isolated unchanged or accompanied by mixtures of ketals resulting from condensation of the phenol at the carbonyl function directly. The C10 methyl group likely presented an environment too crowded for the desired engagement at C1 and thus the ultimate solution to this key bond construction was more involved. After protection of the phenol 23 as a methyl ether, the cyclopentenone was reduced to the corresponding allylic alcohol with DIBAl-H, and treatment of the crude alcohol with camphorsulfonic acid in chloroform resulted in extended elimination of the alcohol to afford the diene 24 in 56% yield over three steps. The diene 24 was selectively reduced to the fully substituted alkene 3 under the action of diimide^31–36; slow addition of glacial acetic acid in dichloromethane to a warm (40 °C) solution of 24 and potassium azodicarboxylate (25) in dichloromethane afforded the alkene 3 in 58% yield (82% conversion) after three reaction cycles (see Supporting Information for experimental details).

Scheme 4.

An oxa-Michael approach to the oxepane core.

The alkene 3 features arylmethyl ether protecting groups which proved quite robust but removal without unproductive decomposition was achieved under mild conditions via a modified Piers-Rubinsztajn reaction;³⁷ treatment of 3 with tris(pentafluorophenyl)borane (BCF) and pentamethyl disiloxane (PMDS) followed by tetra-n-butylammonium fluoride (TBAF) afforded the free bisphenol, setting the stage for the key C1–O bond formation. Exposure of the bisphenol to aqueous sulfuric acid and THF induced an acid-mediated hydration reaction^38–40 of the alkene forging the C1–O bond and affording pure oxepane 26 in 95% yield over two steps (based on the estimated conversion of 24 to 3) as a single stereoisomer; protonation of the alkene may occur preferentially on the top face opposite the dimethylcyclopropane unit establishing the requisite configuration at C10, followed by capture of the resultant cation by the proximate phenol by virtue of the C4 stereochemistry. It is also possible that the protonation event is a reversible process and the thermodynamic cation stereoisomer leads to 26 as observed. In any case, exchange of the bromine atom for a hydroxyl group was achieved via standard protection and borylation chemistry^41–42 to afford the phenol 27 in 93% yield. The balance of the material in this transformation was simple protodehalogenated material (not shown). Comprehensive two-dimensional NMR experiments 26 and protodehalogenated materials (including nOe and correlation spectra) are consistent with data obtained from psiguadial A and confirm the stereochemistry at C10 (see Supporting Information). We installed the two aryl aldehydes in one operation under the action of formamidine acetate (FAA);⁴³ deprotection of 27 with TBAF and immediate treatment of the resultant bisphenol with FAA and acetic anhydride (Ac₂O) in warm THF afforded psiguadial A (1) in 38% over two steps. Synthetic 1 was in spectroscopic agreement with the natural material described by Shao.⁵ The formylation reaction with FAA and Ac₂O was critical to our success; this procedure avoids harsh reagents and reaction conditions and smoothly afforded the natural product. The Rieche protocol,^44–45 while quite successful in the Reisman synthesis of psiguadial B,^7–8 caused only extensive degradation of synthetic intermediates in our hands, affording inseparable mixtures that we have not been able to conclusively identify; these degradation materials also appear to be present in the cation cascade study described by Cramer.¹⁵

In summary, we completed a convergent, enantioselective total synthesis of psiguadial A that leverages an efficient enolate-ortho-quinone methide coupling reaction. This powerful methodology forges the key C1’–C4 bond, which could not be achieved by alkylative methods with conventional metal enolates and alkyl halide electrophiles. The unique oxepane core 26 of the natural product was completed in just 17 steps from (+)-3-carene (9), showcasing an elegant and biomimetic acid-mediated ring closure event in a highly complex setting. Finally, the FAA/Ac₂O fomylation protocol proved uniquely effective in the end-game of our synthesis and is a powerful method for the preparation of complex formylated phenols. Our current efforts are directed toward the application of enolate-ortho-quinone methide coupling strategies to the psidials⁴ and other related meroterpenoid natural products to support biological studies with these fascinating scaffolds.

Supplementary Material

Detailed experimental procedures, spectra and X-ray data (PDF).