Naturally occurring polycyclic polyprenylated acylphloroglucinols (PPAPs: Scheme 1) commonly have a highly substituted bicyclo[3.3.1]nonanone core.1 Hyperforin (1), a representative of this family, was isolated from the herb St. John’s wort (Hypericum perforatum).2a Hyperforin exhibits various biological activities, including mild antidepressant activity,3 antimalarial activity,4 human histone deacetylase inhibitory activity,5 and CYP3 A4 induction activity.6 Enhancement of a specific biological activity through structural modification is an important direction in drug-discovery research, and thus establishing a flexible, asymmetric total synthetic route is a fundamental prerequisite.
Their structural complexity and potential utility as pharmaceutical leads make PPAPs very attractive synthetic targets. The total syntheses of garsubellin A (2),7 clusianone (3),8 and nemorosone (4)8d have been accomplished.9 The use of elegant biomimetic approaches to construct the bicyclic core has resulted in some of these racemic syntheses being short and applicable for the production of structurally diverse analogues.8b,e, 9e The catalytic asymmetric synthesis of PPAPs, however, remains a daunting challenge; there is only one asymmetric synthesis of PPAPs (that of 3), which involved a late-stage kinetic resolution using a stoichiometric amount of chiral lithium amide.8c Hyperforin (1) contains an additional chiral quaternary center at C8 compared to 2–4, thus providing a greater obstacle to its synthesis. We report herein the first catalytic asymmetric total synthesis of ent-hyperforin (5, the antipode of 1).10
We previously developed a catalytic asymmetric Diels–Alder reaction between diene 7 and dienophile 8, which was promoted by a cationic iron complex (10 mol %) derived from pybox ligand 6 (Scheme 2).11 Product 9, which contains contiguous tertiary and quaternary stereocenters (corresponding to C7 and C8 of 5), was obtained in 93 % yield and 96 % ee with complete exo selectivity (d.r.>33:1).12 This reaction is practical: reactions can be routinely conducted on up to 20 g scales (average 89 % ee).13 We thus planned our synthesis of 5 (Scheme 2) based on this powerful catalytic asymmetric reaction.
After conversion of 9 into allyl enol ether 10, the key bicyclic compound 12 would be constructed by a Claisen rearrangement (10→11) and intramolecular aldol cyclization, according to previous model studies.14 In the model studies, however, a simplified substrate containing geminal dimethyl substituents at C8 was utilized. The effect of the C8 quaternary stereocenter in 10 on the reactivity and stereoselectivity of the Claisen rearrangement was a major concern in this system. Moreover, the success in the construction of the bicyclic core intimately depended on the substitution pattern and conformation of the substrate.15 In this sense, the C8 stereocenter could also affect the intramolecular aldol cyclization. From key intermediate 12, introduction of an oxygen functionality at the extremely congested C2-position and installation of a prenyl group at the C3-position would lead to 5.
Based on the synthetic plan, we first converted enantiomerically enriched oxazolidinone 9 to MOM ether 13 over three steps with high efficiency (Scheme 3). Since the C10 hydroxy group readily eliminated to give the corresponding enone under various reaction conditions, cleavage of the two TIPS groups was conducted by a two-step sequence, which afforded primary alcohol 14.16 Direct oxidation of 14 and subsequent addition of an isopropyl group to C10 were difficult because of the instability of the intermediate aldehyde derived from 14. Hence, 16 was synthesized via 15, which was produced from 14 by temporary protection of the primary alcohol with a TMS group, protection of the ketone as an enol silyl ether, and selective cleavage of the TMS ether. After oxidation of 15 with TPAP 17 followed by introduction of the isopropyl group under Barbier conditions (d.r.=5:1), hydrolysis of the enol silyl ether afforded ketone 16. Although multiple steps were required for the apparently simple conversion from 14 into 16, the overall yield was reasonable (58 % over 6 steps). After protection of the C10 hydroxy group of 16 with a TMS group, prenylation of the kinetically produced lithium enolate proceeded exclusively from the axial β face at C5 to give 17. The two diastereomers derived from the C10 stereocenter exhibited distinctly different reactivity in this step, and the ratio of the product diastereomers was enriched to 9:1.
Previous studies indicated that the configuration at C5 controls the approach of an allyl group to C1 in the Claisen rearrangement.14 Therefore, β-prenyl 17 was converted into α-prenyl 18 through a deprotonation/kinetic protonation sequence (Scheme 4). Cleavage of the TMS ether, Dess–Martin oxidation,18 and O-allylation produced 10, the precursor for the key Claisen rearrangement. The thermal Claisen rearrangement of 10 proceeded with high selectivity (12:1) from the β face, and 11, which contains the requisite three contiguous stereocenters (two of which are quaternary), was obtained with high fidelity. This excellent stereoselectivity was consistent with the model studies,14 and attributable to the pseudoaxial methyl group at C8 blocking the α face (19). The key bicyclic intermediate 12 was synthesized uneventfully from 11 through a selective hydroboration at the terminal double bond using disiamylborane [(Sia)2BH], Dess–Martin oxidation, intramolecular aldol cyclization of resulting aldehyde 20, and oxidation.
From 12, the remaining tasks were: 1) to convert the C7 MOM ether moiety into a prenyl group, 2) oxidize C2, and 3) install a prenyl group at C3. Of these tasks, conversion of the C7 MOM ether into a prenyl group was conducted first. Cleavage of the MOM ether under acidic conditions proceeded with concomitant protection of the homoprenyl group at C8 to give 21. This unplanned selective protection was desirable because the reactive homoprenyl group caused side reactions at a later metathesis stage. Swern oxidation of 21, followed by the addition of a vinyl Grignard reagent produced allylic alcohol 22 as a single isomer, which was deoxygenated through acetylation and a palladium-catalyzed allylic reduction.8e, 19 The subsequent cross-metathesis with isobutene using the Hoveyda–Grubbs catalyst20 afforded 23 containing the prenyl group at C7.
The oxidation of C2 was studied next; however, this task proved to be extremely difficult. After the conversion of 23 into 24 under the palladium-mediated conditions (Scheme 5),21 inter- and intramolecular conjugate addition of various heteronucleophiles (such as Si, O, and N reagents) was attempted, but without success.13a Finally, we attempted a [3,3] sigmatropic rearrangement of xanthate 25, which was produced efficiently from 24. Thermal rearrangement of 25 proceeded cleanly; however, the expected functionalization at C2 did not occur. Instead, dithioate 26 was obtained by a [1,3] rearrangement.22 This result again illustrated the highly congested nature of the C2-position. This finding, however, allowed us to attempt a vinylogous Pummerer rearrangement23 for the oxidation of C2. The intermediate thionium cation would be highly electrophilic, thus making it possible to introduce an oxygen functionality at the C2-position.
Based on this hypothesis, we studied the critical vinylogous Pummerer rearrangement intensively by using sulfoxide 32 as a model substrate (Table 1). Treatment of 32 with trifluoroacetic anhydride (TFAA) in the presence of NEt3 resulted in the vinylogous Pummerer rearrangement and normal Pummerer rearrangement proceeding at comparable rates, thereby affording, after hydrolysis, a 1:1.1 mixture of the desired allylic alcohol 34 and enone 35 (entry 1). Encouraged by this result, we then optimized the reaction conditions. The regioselectivity was greatly influenced by the base used. Among these examined, pyridine preferentially afforded 34 in moderate selectivity (entry 2; 34/35=3:1). The selectivity was improved by increasing the steric bulkiness of the pyridine-derived bases. 2,6-Di-tert-butylpyridine was finally found to be the optimum base, giving 34 as the major product with a selectivity of 8.3:1 (entry 4).
Having optimized the vinylogous Pummerer rearrangement with model substrate 32, we applied the conditions to the actual substrate 27, which was synthesized from 26 through thiolysis followed by S-methylation and S-oxidation24 (Scheme 5). As expected, the vinylogous Pummerer rearrangement of 27 proceeded preferentially (4:1) to the normal Pummerer rearrangement under the optimized conditions, thereby providing the desired allylic alcohol 28 in 65 % yield.
The final task was the installation of the prenyl group at C3. S-Oxidation using H2O2 in hexafluoroisopropanol (HFIP)25 followed by Dess–Martin oxidation of the allylic alcohol afforded sulfoxide 29. After deprotection of the homoprenyl group through elimination of the methoxy group by treatment with an acidic resin, an addition/elimination sequence using allyl alcohol afforded allyl ether 30. The catalyzed intramolecular allyl transfer presumably proceeded via a π-allyl-palladium intermediate, and enol acetate 31 was obtained in 50 % yield after O-acetylation in a one pot reaction. It is noteworthy that thermal, microwave-assisted, and Lewis acid mediated Claisen rearrangement of 30 only produced a trace amount of the product (giving either complex mixtures or no product). Finally, cross-metathesis to introduce the prenyl group at C3, and methanolysis of the acetate under basic conditions completed the total synthesis of ent-hyperforin (5). 1H, 13C NMR, and IR spectroscopic data as well as mass spectrometric data were all identical with the reported values. The optical rotation of synthesized 5 was opposite to that of the natural isomer (=−36.8 (c=0.38, EtOH); Lit. +41).2
In conclusion, we have achieved the first catalytic asymmetric total synthesis of ent-hyperforin. The key reactions were: 1) an iron-catalyzed asymmetric Diels–Alder reaction to produce contiguous C7 and C8 stereocenters; 2) a stereoselective Claisen rearrangement to produce the bridgehead quaternary carbon atom at C1; 3) an intramolecular aldol reaction to produce the highly substituted bicyclic core; and 4) a vinylogous Pummerer rearrangement to install the oxygen functionality at the C2-position. These basic methods are applicable to the asymmetric synthesis of other PPAPs and analogues of hyperforin. However, further improvements in the efficiency of the reactions may be necessary for such aplications. Studies are ongoing and will be reported in due course.