Synthesis and antifungal activity of (-)-(Z)-dysidazirine.

Invasive fungal infections, particularly those associated with Candida sp., pose a serious and increasing threat to human health. Widespread use of fluconazole since the mid-1990s has lead to the rise of resistant Candida strains (e.g. Candida glabrata and Candida albicans) and attendant increased mortality among immunocompromised individuals.1,2 New antifungal agents are urgently needed, and marine natural products represent an underexploited source of potential leads.2 (2R)-(E)-Dysidazirine, (−)-3, one of a few rare naturally occuring azacyclopropenes (2H-arizines), was isolated in 1987 from the marine sponge Dysidea fragilis (collected in Fiji).3 Compound (−)-3 exhibited potent anti-fungal activity against Candida albicans and Sacharamyces cerevisiae (4 µg/disk, disk-diffusion assay). More recently, brominated analogues 4 and (+)-6 (antazirines) were isolated along with (Z)-dysidazirine (1) and (2S)-(E)-dysidazirine [(+)-3] from the same sponge species collected in Micronesia.4 Antifungal activity for 4 and (+)-6 were not reported,4 and they were found to be inactive against a panel of bacteria. We recently isolated three new antazirine analogues ((+)-5, (+)-7, (−)-8) from a sample of D. fragilis that also provided the two known antazirines ((+)-4 and (+)-6).5 All five antazirines proved to be inactive against a panel of Candida and Cryptococcus spp., in stark contrast with (−)-3. This surprising result led us to hypothesize that terminal substitution (eg. Br, Cl, Me) attenuates antifungal activity in long chain 2H-azirines. In order to explore this hypothesis, we have developed a short, flexible synthesis of (−)-1 and a structural analogue (−)-2 that features a terminal tert-butyl group. There is only one published synthesis of a long chain 2H-azirine, that of E-dysidazirine, completed in 1995 by Davis and coworkers.6 In this case, the key azirine ring was formed by treating an appropriately substituted N-(p-tolylsulfinyl)-2-carbomethoxyaziridine with LDA at −78 °C. While this approach allowed for synthesis of (−)-3 with high optical purity, the necessity to preserve the Z configuration in (−)-1 required an alternate approach. We chose a strategy based on a variant of the asymmetric cinchona base-catalyzed Neber reaction of O-tosyloximes first reported by Zwanenburg and coworkers.7 Synthesis of (2R)-(Z)-dysidazirine [(−)-1] began with addition of the lithio-anion of pentadecyne (9) to methyl malonyl chloride (10, Table 1). The addition gave poor, variable yields when the deprotonation of 9 was effected with n-BuLi; using EtMgBr, however, led to clean formation of 11 in a reproducible 70% yield. Table 1 Optimization of pentadecyne addition to methyl malonyl chloride Keto-ester 11 was converted to the corresponding oxime (NH2OH•HCl, pyridine, EtOH, 55 °C, 30 min) which was tosylated immediately (p-toluenesulfonic anhydride, pyridine, DMAP, 1.5 h) giving 12 in two steps (72% yield). Treatment of 12 with quinidine (0 °C, 48 h) lead to clean, albeit slow, formation of the desired 2H-azirine ring.7 While the product azirine was formed with only modest enantioselectivity, this methodology is notable for its practical simplicity and high chemical yield. Partial hydrogenation of (−)-13 using Lindlar’s catalyst at ambient temperature in EtOH proved difficult due to facile reduction of the product alkene and the azirine ring within minutes. An improvement was found by lowering the temperature of hydrogenation (0 °C, hexanes) to give (Z)-dysidazirine [(−)-1] in 52% yield and optical purity (59% ee) comparable to the related natural products.5 In order to examine the effect of terminal substitution on antifungal activity, analogue (−)-2 was synthesized starting with Wittig reaction between phosphonium bromide 14 and 3,3-dimethylbutanal, giving alkene 15 as a mixture of double bond isomers. Hydrogenation of 15 (H2, Pd/C, EtOAc, 1.5 h) gave saturated alcohol 16 which was oxidized under Swern conditions to give the corresponding aldehyde.8 Treating the crude aldehyde immediately with PPh3/CBr4 (CH2Cl2, 0 °C, 1 h)9 afforded dibromoalkene 17 which was subsequently converted to terminal alkyne 18 (n-BuLi, THF, − 78 °C → rt) in 86% overall yield from 16. Addition of the lithiated acetylide ion of 18 to 10 proceeded reliably to give 19, albeit in moderate yield (45%). Keto-ester 19 was converted to the corresponding oxime-tosylate 20 (71%) which underwent quinidine-mediated cyclization to give the desired azirine (−)-21 in 80% yield. Cyclization of 20 proceeded even more slowly than the corresponding reaction with 12, and required brief warming to 10 °C to ensure complete consumption of starting material. Lindlar reduction of (−)-21 gave (−)-(R)-17,17-dimethyl-(Z)-dysidazrine [(−)-2] in 58% yield. Both (−)-1 and (−)-2 were screened for antifungal activity against a panel of Candida and Cryptococcus spp. (Table 2). Synthetic (Z)-dysidazirine [(−)-1] showed strong activity against each fungal strain (MIC50 2–8 µg/mL) with the exception of C. krusei (MIC50 16 µg/mL). This appears to be consonant with the activity originally reported for isomeric (−)-E-3 in 1988.3 In contrast, compound (−)-2 with the bulky tert-butyl chain terminus was effectively inactive across the entire panel. Table 2 Antifungal activity of (−)-1 and (−)-2a This surprising result points a subtle complexity in mechanism of action for (−)-1. The target of (−)-1 in fungi is unknown, however it is likely to involve essential metabolism of lipids. The electrophilic properties of the strained α,β-unsaturated azacyclopropene suggests, at first glance, that the anti-fungal activity of (−)-1 might arise from non-specific Michael addition to nucleophilic amino acid residues (e.g. cysteine, lysine). The differential activity observed for (−)-1 and (−)-2 however, argues for a more specific effect that involves metering of the lipid chain.10 We postulate that (−)-1 engages its target protein in a two-point binding motif in which the lipid chain occupies a hydrophobic pocket with stringent steric requirements while the electrophilic α,β-unsaturated azirine carboxylate terminus binds at a distal site. The specificity of this interaction with respect to chain length remains to be investigated, however it is notable that dysidazirine shares a common chain length (C18) with several sphingolipids important for fungal cell growth, such as phytosphingosine and ceramide.2 This suggests dysidazirine may be an ‘antimetabolite’ that interdicts fungal cell lipid metabolism, perhaps sphingolipid biosynthesis. A more extensive study of SAR and mechanism of action of (−)-1 is underway and our results will be reported in due course. In summary, we have completed the first synthesis of the marine natural product (−)-(Z)-dysidazirine4 and a structural analogue (−)-2 that features a terminal tert-butyl group. Compound (−)-1 shows significant antifungal activity toward a range of Candida and Cryptococcus species, but (−)-2 is almost completely inactive. This surprising result suggests that the activity of (Z)-dysidazirine is modulated by two-point binding to an as-yet unidentified target protein in fungal cells. Interaction between (−)-1 and its putative target likely involves sterically stringent interaction between the lipid side chain and a hydrophobic domain.