Asymmetric, Stereodivergent Synthesis of (−)-Clusianone Utilizing a Biomimetic Cationic Cyclization

Abstract

We report a stereodivergent, asymmetric total synthesis of (−)-clusianone in six steps from commercial materials. We implement a challenging cationic cyclization forging a bond between two sterically encumbered quaternary carbons. Mechanistic studies point to the unique ability of formic acid to bring about successful cyclization to the clusianone framework.

Polyprenylated polycyclic acylphloroglucinols (PPAPs)^[1,2] are a structurally intriguing class of natural products containing a bicyclo[3.3.1]nonane-2,4,9-triketone core. (+)-Clusianone (1, Figure 1), a type B PPAP as indicated by the α-acyl β-hydroxy enone, ^[1f] was first isolated and characterized in 1976 from the bark of Clusia congestiflora.^[3a] Since the early 1990’s, there have been a handful of isolation studies that address the molecule’s structure.^[3] However, it was not until 20 years after its initial isolation that reports began to emerge on clusianone’s bioactivity.^[3c,4] In 2005, the compound was shown to possess notable anti-HIV activity (EC₅₀ = 20 nM) by inhibiting the gp120-sCD4 viral-receptor interaction suggesting its potential for preventing HIV infection.^[3c] Clusianone’s chemopreventative properties^[4c,d] are thought in part to arise from its ability to inhibit histone acetyltransferase (HAT) enzymes. Nemorosone (2), an isomeric type A PPAP as indicated by the bridgehead substituted acyl group α to a quaternary center,^[1f] has notable anticancer properties in part due to its ability to activate HAT enzymes^[4b,d,f] and serve as a protonophoric mitochondrial uncoupler.^[4g] Hyperforin (3), a type A PPAP found in St. John’s Wort, is used to treat depression and possesses anti-inflammatory and antibiotic properties.^[1,2d] Given the great diversity of bioactivities attributed to small structural changes among these PPAPs, it is not surprising that these compounds have become popular targets for synthetic chemists.^[5,6] For example, elegant asymmetric syntheses of clusianone have been reported by both the Simpkins^[6a] and Coltart^[6b,c] groups.

Figure 1.

Representative PPAP Natural Products.

Structurally, clusianone (1) and nemorosone (2) possess synthetically challenging highly oxygenated bicyclo[3.3.1]nonane cores adorned with isoprenyl side chains and three stereocenters that derive biosynthetically from the common cationic intermediate 6 (Figure 2).^[7] For more than a decade, synthetic groups have been largely unsuccessful at achieving similar cationic cyclizations to access the fully functionalized cores of type A and B PPAPs. Such cyclizations have been reported only in the case of systems lacking the full functionality of the natural products.^[5b,d,e,8,9] For example, Marazano and coworkers could not isolate C-cyclized products via cationic intermediates analogous to 6.^[8d] In a study by Couladouros and coworkers,^[8e] it was shown that unconjugated tertiary carbocations could not be cyclized to the [3.3.1]-bicyclic core of 3. The difficulty of such a transformation likely stems from a high degree of strain in the transition state and from the steric demands of forming a hindered carbon bond between two sterically congested quaternary carbons.

Figure 2.

Proposed Biosyntheses of (+)-clusianone (1) and (+)-nemorosone (2).

In line with our group’s interest in rapid access to PPAP natural products and derivatives,^{[2b,5d,8f,g,10]} we hoped to develop a route to 1 and/or 2 possessing the brevity and flexibility necessary for SAR studies. Herein, we report a stereodivergent, asymmetric synthesis of (−)-1 in only six steps from 5-methoxyresorcinol employing the first cationic cyclization to access the fully functionalized core of (−)-clusianone. Additionally, we reveal the selective synthesis of five novel architectures from the key cyclization substrate along with a new purification strategy for dearomatized phloroglucinols and type B PPAPs which should be of general utility for these types of compounds.

Inspired by the efficiency of their biosyntheses (Figure 2), we considered synthesizing 1 and/or 2 from 9, a common intermediate employed in our group’s total syntheses of both 7-epi-nemorosone^[2b] and plukenetione A.^[8f] As one possibility, we envisioned that (−)-7 could be obtained via cationic cyclization of dearomatized substrate 8 involving protonation of the 1,1-disubstituted olefin to generate a tertiary carbocation at C8 followed by intramolecular enol attack at C3 (Figure 3).

Figure 3.

Retrosynthetic Analysis to Access PPAP Core 7.

At the outset of our investigation, we had three principal concerns regarding the success of a protonative, cationic cyclization to access the bicyclo[3.3.1]nonane core: 1) control of O- vs. C-selectivity in the cyclization, 2) whether the O-methyl protecting group at C4 could direct cyclization at C1 to the nemorosone core, and 3) if 8 would rearomatize under acidic conditions given reports that similar dearomatized intermediates have undergone rearomatization under acidic conditions.^[8a,d] We began our study by investigating an improved synthesis to acylphloroglucinol 9 which was previously prepared in 20 % yield from 11 by direct aromatic substitution.^[10] Pleasingly, we discovered that selective O-allylation of 11 proceeded in 71 % yield which was followed by a thermolytic Claisen rearrangement to produce 9 (92 %) on a multigram-scale.^[11] Racemic alcohol 12 was prepared in a single pot in 93 % yield via a modified procedure (Scheme 1).^[12] Triflation of 12 with triflic anhydride afforded 10 which was used in situ.^[11]Alkylative dearomatization of 9 proved challenging in initial studies as little was known regarding factors that may control C- vs. O-selectivity for dearomatization of phloroglucinol substrates.^[13] Due to the lack of literature precedent for alkylative dearomatizations involving unactivated alkyl electrophiles,^[8e,10,14] it was necessary to carry out a systematic evaluation of reaction conditions. Table 1 shows how base counterion plays a critical role in preventing undesired O-alkylation. The lithium counterion of the phenolate derived from 9 is likely tightly bound to the phenolate oxygen thereby preventing significant O-alkylation.^[15,16] Not surprisingly, KHMDS produced only O-alkylation products as the relatively non-coordinating potassium counterion will expose a naked oxygen enolate. We speculate that lithium coordination to both the sulfonate and enolate oxygen (not shown) helps to facilitate C-alkylation.^[17] A solvent screen revealed that the reaction does not proceed to full conversion in THF and is very slow in toluene; optimal results were achieved with a combination of THF/toluene (3:1, 0.057 M).^[11]

Scheme 1.

Synthesis of racemic triflate 10.

Table 1.

Optimizing C- vs. O-selectivity in Alkylative Dearomatization.

Entry	Base	Yield[a]
		%C	%O
1	KHMDS	0	53
2[b]	NaHMDS	47	31
3	LiHMDS	73	6

^[a]Isolated yield of product after silica gel chromatography.

^[b]Reaction was carried out in THF:benzene (3:1).

KHMDS = potassium bis(trimethylsilyl)amide

NaHMDS = sodium bis(trimethylsilyl)amide

LiHMDS = lithium bis(trimethylsilyl)amide

With ample quantities of dearomatized product 8 in hand, we evaluated conditions for cationic cyclization to 7 (Table 2). The success of our biomimetic cationic cyclization to achieve bond formation at C3 was revealed after investigating >70 different reaction conditions consisting of various Brønsted acids, Lewis acids, solvents, concentrations, temperatures, and work-up variations.^[11] Over 13 different products were isolated, five could be obtained selectively depending on the conditions, ^[8a,d,18] and only formic acid successfully formed a C-cyclized adduct. Highly polar solvents, such as TFA, HFIP, CH₃NO₂, and CH₃CN, favored the O-cyclized Cope rearrangement products 13 and 15. However, the unique combination of LiBr and p-TsOH^[19] in CH₂Cl₂ provided clean formation of 14 without Cope rearrangement. Using the latter combination in EtOAc, acetone, and 1,4-dioxane also afforded 14 but in significantly lower yields. Rearomatized product 16 was obtained in 97% yield under similar conditions in CH₂Cl₂ at slightly higher temperatures (Table 2, entry 7). We presume that LiBr increases the acidity of p-TsOH in CH₂Cl₂ via lithium coordination to the sulfonic acid such that cyclization is observed at temperatures below −40 °C.

Table 2.

Conditions Favoring Unique Cationic Cyclization Products.

Entry	Acid	Conditions	Adduct	Yield (%)[a]
1[b]	formic acid	Neat 10 °C to rt, 72 h	(+/−)-7	30
2	p-TsOH (10 equiv) LiOTf (10 equiv)	HFIP (0.015 M) 50 °C, 12 h	13	89
3	TFA (0.14 M)	Neat r.t., 24 h		97
4	p-TsOH (10 equiv) LiBr (10 equiv)	CH3NO2 (0.015 M) r.t., 10 min		55
5	p-TsOH (10 equiv) LiBr (10 equiv)	CH2Cl2 (0.015 M) −40 °C, 30 h	14	55[c]
6	p-TsOH (10 equiv) LiBr (10 equiv)	CH3CN (0.015 M) 60 °C, 10 min	15	35
7	p-TsOH (10 equiv) LiBr (10 equiv)	CH2Cl2 (0.015 M) −25 °C, 30 h	16	97[d]

^[a]Isolated yield of product after silica gel chromatography.

^[b]Product 13 also produced in 34 % yield.

^[c]Product isolated as a mixture of cis and trans isomers.

^[d]Product obtained pure (¹H NMR) without need for further purification.

In rationalizing the various O-cyclization outcomes, we observed that pyranodienone 13 could be demethylated to afford product 15 (79 %, Scheme 2). Rearomatization to product 16 proceeds strictly from intermediate 14, while 15 is obtained directly from demethylation of O-cyclized Cope product 13. In all cases, demethylation and Cope rearrangement occurred after cyclization to 14. Based on these observations, we propose a mechanism that accounts for the formation of O-cyclized products 13–16 (Scheme 2). In contrast to the above results, neat formic acid uniquely led to diastereoselective cyclization of 8 which proceeded to afford both allyl clusianone 7 (30 %) and Cope product 13 (34 %, Table 2, entry 1). Moreover, we observed that C-cyclization proceeded less efficiently in the absence of an enol methyl ether at C4 providing decomposition and small quantities of O-cyclization products including 15 in 23 % yield.

Scheme 2.

Mechanistic Rationale for O-Cyclized Products. a) LiBr, p-TsOH, CH₂Cl₂, −40 °C, 30 h, 55 %; b) formic acid, 10 °C to r.t., 12 h, quant.; c) LiBr, p-TsOH, CH₂Cl₂, −25 °C, 30 h, 97 % d) LiBr, p-TsOH, CH₂Cl₂, r.t. 5 min, 79 %.

Alkylative dearomatization of 9 with chiral triflate (−)-(R)-10^[12] (Figure 3) led to the production of (−)-(S,S)-8 and (+)-(R,S)-8 as a 1.3:1 mixture of diastereomers which were separated by preparative thin layer chromatography. When each was individually subjected to neat formic acid, diastereomer (−)-(S,S)-8 converted in a stereodivergent manner to (−)-allyl clusianone (−)-7 (72 %) and Cope product (−)-(S)-13 (13 %),^[20] while diastereomer (+)-(R,S)-8 exclusively afforded (−)-(S)-13 (84 %, Scheme 4). (−)-Clusianone (−)-1 was obtained following cross metathesis of (−)-7, and each diastereomer of 8 could be selectively converted into their corresponding O-cyclized adducts, (R,S)-14 and (S,S)-14, in reasonable yields utilizing LiBr/p-TsOH at −40 °C.

Scheme 4.

Stereodivergent Syntheses of (−)-Clusianone (−)-1 and O-Cyclized Products.

As formic acid was the only condition out of >70 others that led to C-cyclization products, it seemed likely that it may mediate cyclization via a unique mechanism.^[21] To simplify our analysis of this mechanism and the observed stereodivergency, we considered the possibility that one tautomer of methyl enol ether 8 might be responsible for the majority of C-cyclization to allyl clusianone 7. We sought to test this hypothesis by first methylating 8 with TMSCHN₂^[11] which was followed by treatment of the resulting mixture of permethylated isomers 19 and 20 with formic acid (Scheme 5). Indeed, methyl ether (+/−)-21^[22] was the only clusianone-type product isolated from the reaction in significant yield thereby indicating that cross-conjugated isomer 20 was largely responsible for C-cyclization. Only O-cyclization products derived from product 19 were isolated from the reaction.

Scheme 5.

Cross-Conjugated Isomer 20 is Largely Responsible for C-Cyclization.

Contrary to other carboxylic acids (e.g. acetic acid), formic acid can add efficiently to 1,1-disubstituted olefins under ambient conditions^[23] which we have demonstrated for the case of 2-methyl-1,5-hexadiene.^[11] This could influence the cyclization of (−)-(S,S)-8 via stabilization of the carbocation forming a tight ion pair in solution. It is also known that formic acid can add to electron-deficient and strained bridgehead ketones.^[24] If we consider the possibility of formate addition to (−)-(S,S)-8 at C2 of the cross-conjugated carbonyl (alternative 1, Figure 4A), this would render C3 more electron-rich relative to its vinylogous ester counterpart.^[11] Although alternative 1 may improve the probability for cyclization at C3, it is also possible that formate addition at C2 may not be necessary for cyclization (alternative 2, Figure 4A). Cross-conjugated ketone 23a is more electron rich at C3 than its conjugated tautomer and may be sufficient to cyclize to (−)-7.

Figure 4.

Stereochemical Rationale for Observed Stereodivergency.

Evidence for the proposed formate adduct intermediates in our cyclization was obtained from Ultra High Performance Liquid Chromatography (UPLC) measurements in the reaction medium (98 % formic acid) at various time intervals for both (−)-(S,S)-8 and (+/−)-(S,S)-20.^[11] The [M+H⁺] of the observed reaction components in the Evaporative Light Scattering Detector (ELSD) trace corresponded to monoformate and triformate adducts in the first 30 minutes of reaction; this indicates possible addition to C2 and/or C13 in addition to the 1,1-disubstituted double bond prior to cyclization. Substrate (+/−)-(S,S)-20 was routinely monitored every few hours to observe the formation of monoformate and diformate adducts in the ELSD trace; the formate peaks gradually disappeared as the reaction proceeded to completion. For both substrates, the disappearance of all formate intermediates was observed after 24 h with only product remaining.^[11]

The observed stereodivergent outcome (Scheme 4) may be governed by steric interactions as represented in Figure 4. Unfavorable steric interactions in cations 22b/23b, deriving from (−)-(S,S)-8, between the allyl group and the cyclohexadiene ring system likely cause the equilibrium to favor cations 22a/23a (Figure 4A). Because the cations are proximal to the nucleophilic carbon at C3 in 22a/23a, we observe dominant formation of the C-cyclized product. The preference for O-cyclization in the case of (+)-(R,S)-8 may be rationalized by close proximity of the cation to the enol depicted in the sterically favored conformation 24b (Figure 4B). The configuration depicted in 24a experiences destabilizing steric interactions between the allyl substitutent and the cyclohexadiene core, thereby pushing the equilibrium to favor 24b. Hence, 24b should react to give O-cyclization as the preferred stereodivergent outcome for diastereomer (+)-(R,S)-8.

To conclude our study, we wished to demonstrate that our synthesis of (+/−)-1 could be achieved on a gram-scale which is not commonly reported for this class of compounds. The difficulty associated with synthesizing large quantities of PPAP natural products derives from their tendency to bind to silica and form salts.^[8d,11] Our final synthesis of racemic clusianone (+/−)-1a utilized a general purification strategy for dearomatized phloroglucinols and type B PPAPs, ^[11] where the synthesis of acylphloroglucinol 9 and the subsequent alkylative dearomatization were achieved on a multigram-scale to provide 8 (30 - 37 % over four steps, 1.3:1 d.r., Scheme 6). Without the need to separate diastereomers, the key cyclization in formic acid was conducted on scales up to 2 g where products were separated by basic extraction to provide the potassium salt of allyl clusianone (+/−)-7a (26 – 32 %) as a single isomer.^[11] Olefin metathesis proceeded in 81 % yield using the Grubbs II catalyst^[2c] which was followed by basic workup to isolate more than 210 mg of racemic clusianone potassium salt (+/−)-1a. All salt products could be converted to their conventional protonated, tautomeric forms via simple extraction with 1 M HCl.

Scheme 6.

Large-Scale Synthesis of (+/−)-Clusianone Potassium Salt (+/−)-1a. a) AlCl₃, BzCl, 0 °C to r.t., 3h, 69 %; b) K₂CO_3, nBu₄NI, allyl bromide, acetone, 70 °C, 71 %; c) 1,2-dichlorobenzene, 210 °C, 12 h, 92 %; d) LiHMDS, 10, THF:Tol = 3:1, −20 °C to r.t., 2 h, 67 – 83 %, 1.3:1 d.r.; e) formic acid, 10 °C to rt, 48 h, 26 – 32 %; f) Grubbs II (20 mol %), isobutylene, −78 to 60 °C, 24 h, 81 %.

In conclusion, we have developed a scalable, asymmetric, and stereodivergent synthesis of (−)-clusianone (−)-1 in only six steps from commercial starting materials. Protonative cationic cyclization of 8 allowed selective access to five novel architectures. Mechanistic studies^[11] are described that underscore the ability of formic acid to mediate a unique biomimetic cyclization to access allyl clusianone 7. Finally, we developed a general purification strategy for dearomatized phloroglucinols and type B PPAP derivatives, rendering our entire synthesis column-free from intermediate phloroglucinol 9.^[11] Further studies regarding the synthesis and biological activity of PPAP natural products and derivatives are in progress and will be reported in due course.

Scheme 3.

C4 Methyl Ether Proved Necessary for Efficient C-cyclization.