Introduction

Opportunistic infections by ubiquitous fungi represent a major challenge to immunocompromised patients. The basidiomycete yeast Cryptococcus neoformans, for instance, can cause life-threatening meningitis and affect the lungs and skin of patients with advanced acquired immunodeficiency syndrome (AIDS) [1,2,3,4,5,6]. Hippolachnin A (1) was recently isolated from the South China Sea sponge Hippospongia lachne. The molecule features six contiguous stereocenters embedded in a bicyclo[3.2.0]heptane framework, four of which bear ethyl groups. Biochemical assays, reported in the initial study, revealed high potency against several pathogenic fungi, including C. neoformans (MIC = 0.41 µM) [7]. Biosynthetically, 1 is part of the plakortin family [8] and, more precisely, part of the gracilioether family (Fig. 1). This family also includes the unnamed compound 2, the presumptive biosynthetic precursor of 1, as well as the graciloethers (3, 5, 6), and plaktorin (4).

Fig. 1
figure 1

Selected members of the plakortin family

Full size image

Attracted by its unusual structure combined with the potent bioactivity, numerous research groups, including ours, have developed unique synthetic solutions for this molecule. In the last three years, four total syntheses have been reported [9,10,11,12,13], including a recent biomimetic synthesis by Tang and Enders [14]. In 2014, we launched a program to develop a scalable and diversifiable approach to hippolachnin A (1). Herein, we show how our strategy evolved, discuss our studies on synthetic derivatives, and the results of our evaluation of their biological properties.

Results and discussion

As outlined in Scheme 1, we initially focused on a strategy based non-biomimetic [2 + 2]-cycloaddition of allylic ester 8, to simultaneously form the cyclobutane and the tetrahydrofuran of the natural product. The vinylogous carbonate should then be installed by Peterson olefination. Even though the desired [2 + 2]-cycloaddition of an acyclic α,β-unsaturated ester was anticipated to be difficult, we envisioned that, conducting the reaction in an intramolecular fashion, might allow for efficient trapping of the photochemically formed triplet diradical.

Scheme 1
scheme 1

First generation retrosynthesis of hippolachnin A based on a [2 + 2]-cycloaddition

Full size image

The synthesis commenced with the preparation of tertiary alcohol 13 from dicyclopentadiene (9) (Scheme 2). Conjugate addition from the convex side of enone 10, obtained from Alder-ene reaction of 9 with singlet oxygen, afforded ketone 11 [15]. Subsequent 1,2-addition of EtMgBr mediated by LaCl3.2LiCl yielded tertiary alcohol 12 as a single diastereoisomer [16]. While initial attempts to effect thermolysis resulted in decomposition, flash vacuum pyrolysis (FVP) cleanly provided sensitive tertiary alcohol 13 in good yield [17].

Scheme 2
scheme 2

Synthesis of allylic alcohol 13

Full size image

With 13 in hand, we envisioned esterification with the known carboxylic acid 14 (Scheme 3) [18]. Standard esterification conditions utilizing the acid chloride only resulted in isolation of the free acid after work up. Attempts to generate the acylium ion only led to decomposition of the starting materials [19]. Anhydride 15 was found to be stable to column chromatography but could not be used for ester formation, even at elevated temperatures. Interestingly, both the acid chloride and the mixed anhydride were able to react with t-BuOK to form the corresponding t-butyl ester 16.

Scheme 3
scheme 3

Attempted esterification of 8

Full size image

Since the sterical hindrance of the tertiary alcohol could not be overcome, we decided to install the fourth ethyl group at a later stage of the synthesis. The revised retrosynthesis is shown in Scheme 4. We envisioned formation of 1 by ethyl Grignard addition to cyclopentenone 17, followed by dehydration. The β-ketoester could be derived by Claisen condensation of methyl acetate with lactone 18. Lactone 18 would be traced back to an intramolecular [2 + 2]-cycloaddition of ester 19.

Scheme 4
scheme 4

Revised retrosynthesis of 1

Full size image

Epoxidation of cyclopentadiene [20] (20) followed by SN2׳ displacement with ethylcyanocuprate [21] provided allylic alcohol 21 in moderate yield (Scheme 5). Quantitative formation of the lithium alkoxide and subsequent quenching with acid chloride 14 gave rise to allylic ester 19.

Scheme 5
scheme 5

Synthesis of allylic ester 19 and formation of ester 22

Full size image

With 19 in hand we investigated the key [2 + 2]-cycloaddition [22]. Unfortunately, irradiation of 19 with 310 nm LEDs in the presence of either acetone or benzophenone in various solvents (MeCN, CH2Cl2, benzene) did not result in the desired cycloadditon. Instead, we isolated starting material as a mixture with its (E)-configured diastereomer 22. While excitation of the system proceeded smoothly as proved by the formation of 22, we wondered whether π-bond rotation was too fast to allow for capture of the diradical by the tethered olefin.

To prevent isomerization we decided to incorporate the enone double bond into a ring by stitching the ends of the ethyl groups together using a sulfur bridge. Ring expansion of tetrahydrothiopyranone 23 using ethyl diazoacetate gave rise to thiepanone 24 which, after reduction, mesylation, and elimination provided ethyl ester 26 [23]. Saponification followed by activation of the carboxylic acid as the acid chloride and subsequent esterification accessed enoate 28 in moderate yields (Scheme 6). To investigate the cycloaddition, we submitted 28 to the same conditions used for 19. After irradiation, the clear solution became cloudy and NMR analysis of the mixture indicated decomposition of the starting material.

Scheme 6
scheme 6

Synthesis of thiepan 28 and attempted [2 + 2]-cycloaddition

Full size image

While excitation of the molecule was possible, no productive pathways were observed, which might be attributed to a preferred conformation of the molecule where both olefins are pointing into opposite directions. In order to provide the molecule more flexibility, β-ketoester 32 was prepared (Scheme 7). Activation of acid 30 with CDI and subsequent Claisen condensation with t-BuOAc, followed by ketalization, gave dioxanone 31 [24]. Retro [4 + 2]-cycloaddition and trapping of the resulting ketene with alcohol 19 then afforded β-ketoester 32 [25]. Irradiation of 32 under the previously established conditions resulted in a complex mixture from which we were unable to isolate the desired bicyclo[3.2.0]heptane 33.

Scheme 7
scheme 7

Synthesis of β-ketoester 32 and attempted [2 + 2]-cycloaddition

Full size image

Frustrated by our inability to effect (2 + 2) cycloadditons we decided to turn to a different type of photochemistry (Scheme 8). In our new retrosynthetic analysis, we planned to close the tetrahydofuran ring in 1 by O-alkylation of an enolized β-ketoester 34 [26]. The vicinal ethyl groups should be introduced by addition of an ethyl nucleophile and an ethyl electrophile to the Michael acceptor 35. Compound 35 was derived from the known bicyclo[3.2.0]heptadiene 36, which can be obtained from tropolone ether 37 (Scheme 8) [27].

Scheme 8
scheme 8

Retrosynthetic analysis of 1 based on an excited state rearrangement

Full size image

Our synthesis commences with the photochemical conversion of 37 into methoxy bicyclo[3.2.0]heptadienone 36, which was originally reported by Dauben et. al [27]. Irradiation of 37 induced a disrotatory 4π-electrocyclization to give 38, which then undergoes an excited state rearrangement to yield 36. Notably, even though 38 is an isolatable intermediate, 36 was the only product obtained after full conversion of 37. Ethyl cuprate addition to 36 occurred exclusively from the convex side [15] of the molecule and gave, after Wittig olefination [28] and acid mediated cleavage of the enol ether [29] ketone 40 as a 10:1 mixture of E- and Z- isomers (major isomer shown). Formation of the vinyl triflate [30, 31] and subsequent carbomethoxylation [32] then gave methyl ester 41 (Scheme 9).

Scheme 9
scheme 9

Synthesis of methyl ester 41

Full size image

With 41 in hand, the stage was set for the installation of the vicinal ethyl groups (Scheme 10). We first envisioned the conjugate addition of an ethyl cuprate and subsequent alkylation at the α-position of the ester. While the conjugate addition occurred solely from the convex side of the molecule, acidic quenching of the resulting ester enolate gave an inconsequential mixture of diastereomers with respect to the ester group. Reformation of the enolate using LDA and subsequent trapping with ethyl iodide resulted in 2:1 mixture of diastereomers [33], favoring the desired one. We attributed this result to the opposing stereochemical bias provided by the bicyclic core and the newly introduced ethyl group. Although, the major isomer could be converted into an advanced intermediate 44 of the Wood-Brown synthesis [10], we thought that this low level of selectivity was unacceptable for an efficient synthesis.

Scheme 10
scheme 10

Synthesis of 44 via three component coupling

Full size image

To overcome this problem, we turned toward the 1,3-dipolar cycloaddition of a thiocarbonyl ylide followed by reductive desulfurization to add, in effect, two ethyl radicals across the strained and electron poor double bond of 41 [34,35,36,37,38]. The requisite thiocarbonyl ylide 48 could be generated in situ from sulfoxide 47 following Achiwa’s method [39,40,41,42,43,44]. Compound 47, in turn, was derived from (chloromethyl)trimethylsilane 45 by alkylation with methyl iodide followed by nucleophilic displacement with sodium sulfide and subsequent oxidation (Scheme 11).

Scheme 11
scheme 11

Synthesis of thiocarbonylylide 48 and 1,3-cycloaddition to form 49

Full size image

Dropwise addition of 48 to a hot solution of 41 indeed resulted in the clean formation of tetrahydrothiophen 49, which was obtained as a single diastereosiomer. Single crystal X-ray structure analysis showed that the methyl groups adopt a trans-configuration with respect to the heterocyclic ring. Notably, the reaction also proceeds under high pressure conditions (12 kbar) at room temperature, albeit with slightly lower yield.

With the key intermediate in hand we turned our focus to the homologation of the ester group to the corresponding β-ketoester. Unfortunately, all attempts [45,46,47,48] to effect a Claisen condensation failed, presumably due to steric hinderance (Table 1).

Table 1 Attempted Claisen condensation of 49.
Full size table

We sought to overcome this problem by using a less hindered electrophile settling on the cyano group with its sp-hybridized carbon. To this end we prepared nitrile 52 from vinyl triflate 51 by Pd-catalyzed cross coupling with sodium cyanide [49] (Scheme 12). To our delight, the 1,3-dipolar cycloaddition gave tricycle 53 as a single diastereomer, albeit in slightly lower yield. Blaise reaction of 53 with zinc organyl 54 then yielded the desired en-amine 55 [50]. Unfortunately, the vinylogous carbamate 55 proved to be completely resistant to hydrolysis with starting material recovered under standard conditions [51, 52]. Harsher conditions led to decomposition of 55.

Scheme 12
scheme 12

Synthesis of enamine 55 and attempted hydrolysis

Full size image

Although methyl ester 41 was resistant to a crossed-Claisen condensation, it could be reduced to the primary alcohol and mildly reoxidized. The resulting aldehyde 56 was able to undergo an aldol addition [53] with the enolate generated from methyl acetate and gave, after Dess Martin oxidaition, the desired β-ketoester 56. Formation of the tin enolate followed by trapping of the simultaneously generated tertiary carbocation [26] gave rise to (Z)-configured vinylogous carbonate 57 [54]. Its structure could be confirmed by X-ray structure analysis. Reductive desulfurization with Raney nickel in THF [55,56,57] gave access to the natural product, hippolachnin A, as a racemate (Scheme 13).

Scheme 13
scheme 13

Total synthesis of 1

Full size image

Piao et. al. reported strong antifungal activity of hippolachnin A [7]. With a good synthetic entry at hand, we decided to develop analogues that would allow us to map structure-activity relationships and would show improved physicochemical properties. Since 1 is poorly soluble in aqueous solutions, we chose to synthesize more polar derivatives by oxidizing the sulfur of the advanced intermediate 57 to either the sulfoxide or the sulfone (Scheme 14). Notably, treatment of 57 with excess m-CPBA at room temperature led to partial isomerization of the double bond to afford a mixture of 58 and 59. Partial oxidation of 57 gave sulfoxide 60 as a 5:1 mixture of diastereomers. We also synthesized the butyrolactone 61 by acid mediated cyclization of 41.

Scheme 14
scheme 14

Synthesis of synthetic analogs of 1

Full size image

With the racemic natural product and several synthetic analogous in hand, we evaluated the antifungal, antimicrobial and the nematicidal activity of 1, 57, 44, 61, 59, 58, and 60. We were surprised to find that (±)− 1 showed no antifungal activity, in particular against C. neoformans (Table 2). This observation was recently independently reported by Wood [58]. It seems unlikely that the unnatural enantiomer somehow neutralizes the biological effect of (+)-hippolachnin A. Our synthetic analogs of 1 were also inactive against the fungi listed in Table 2, except for butyrolactone 61, which showed modest activity against C. neoformans. Compounds 1, 57, 44, and 61 exhibited weak antimicrobial activity (Table 2 and supporting information) and no or very weak cytotoxicity (supporting information). Interestingly, all analogs, including the natural product, inhibited the growth of Caenorhabditis elegans, but again with weak potency (Table 2).

Table 2 Antifungal, antimicrobial, and nematicidal activity activity of 1 and its analogs
Full size table

Conclusion

Herein, we have presented the evolution of our campaign for the total synthesis of hippolachnin A. Starting from tropolone methyl ether 37, we developed a robust and scalable synthetic route which enabled us to synthesize not only more than 100 mg of the natural product but also a variety of synthetic derivatives. The bicyclic carbon core was constructed by photoisomerization of 37. The four ethyl groups were introduced by diastereoselective cuprate addition, Wittig olefination, and the 1,3-dipolar cycloaddition of a thiocarbonyl ylide. The latter represents a rarely used method to overcome steric hindrance and, in effect, link a nucleophile to an electrophile. The heterocyclic ring was closed via trapping of carbocation generated in situ by a tin enolate. Reports about the antifungal activity of hippolachnin A could not be confirmed. This is in keeping with several other recent cases where the purported bioactivity of natural products did not match the activity of their synthetic versions [59, 60]. Of course, this serves as yet another justification for total synthesis, which often provides material in purer form and on a larger scale than isolation from natural sources.