Meroterpenoids such as the andrastins and the citreohybridones are interesting as examples of mixed polyketide-terpenoid biosynthesis [1,2,3] and they often have biological activities [3,4,5,6,7]. The andrastins and citreohybridones were originally isolated from hybrid strains of Penicillium sp. FO-3929 [1,3] and P. citreo-viride B [4]. In the course of our analytical HPLC screening of Penicillium species, the characteristic UV-spectrum of andrastin A (1) (Figure 1) was detected in many known Penicillium species such as the blue cheese mold P. roqueforti [8,9], the onion associated mold P. albocoremium [10], and psychrotolerant strains of P. crustosum from arctic ice [11]. The newly arctic described fungus P. ribeum [12,13], however, was found to produce a metabolite with the UV-spectrum of andrastin A, but with a different retention time and molecular mass. The data for the compound matched those of the newly reported compound atlantinone A [14], which we have confirmed by isolation and structure elucidation. HPLC-UV-MS analysis demonstrated that several species within genus Penicillium produce the compound under standard laboratory conditions.

An EtOAc extract of Penicillium ribeum (IBT 16537) was separated by high-speed countercurrent chromatography (HSCCC), and through UV-guided fractionation the fifth HSCCC fraction was further purified by HPLC to afford pure 2 (6.8 mg).

Atlantinone A (2) exhibited a molecular ion in HREIMS corresponding to the molecular formula C₂₆H₃₄O₆(10 degrees of unsaturation). The 2-dimensional structure of 2 was established by interpretation of 1- and 2-dimensional NMR data (see Supplementary data). The relative configuration of 2 was established by NOE experiments (Figure 2) indicating the presence of an ent-5α,14β-androstane skeleton with a 10,3α lactone bridge in agreement with the X-ray structure published by Cichewicz’s group [14]. The unusual configuration of 2 is undoubtedly caused by the lactone bridge forcing the oxygen to occupy an α-position.

We propose that atlantinone A (2) is derived from andrastin A (1), which on enzymatic oxidation at C-3 and C-10 could cause the oxygen at C-3 to inverse from β- to α (Figure 3). This reduction may be either stereospecific or non-stereospecific. In the first case the lactone formation would arrest the configuration in the α configuration. In the second type of reduction the α configuration would be effectively removed by lactone formation and would on repetitive oxidation reduction cycles end up completely in the α state. Atlantinone A (2) is to our knowledge the first meroterpene with an α configuration in position 3. The citreohybridones [4] and andrastins A-C [1,3] are all reported with β configuration in position 3.

In our previous studies of cheese associated fungi [15] we had observed an unknown peak in several fungal extracts with a UV spectrum very similar to that of atlantinone A. Recultivation on standard cultivation media, followed by extraction and analysis by LC-DAD-MS, indeed confirmed the following species to be atlantinone A producers; P. solitum, P. discolor, P. commune, P. caseifulvum, P. palitans, P. novae-zeelandiae and P. monticola, as evident by detection of [M+H]⁺ and [2M+H]⁺ in ESMS in the positive mode similar to those of the authentic standard. We also screened several isolates of P. hirayamae for atlantinone A, but were not able to detect it in any of the isolates NRRL 3588, FRR 143, FRR 3330, FRR 1835, or IBT 30340 neither on CYA nor on YES agar.

Since the former five species are often found as contaminants of various types of cheeses [15], one might speculate that atlantinone A can also be detected in cheese products. The compound did not show any inhibitory activities against a panel of bacteria and fungi [16], however it’s structural similarity to the andrastins, previously shown to have potent anticancer drug properties as farnesyltransferase inhibitors [1], indicates that further bioactivity testing is needed in order to establish the potential risk, or possible health beneficial effects of atlantinone A.

While P. solitum, P. discolor, P. commune and P. caseifulvum (all members of the section Fasciculata [17]) readily produce atlantinone A , P. hirayamae (previously incorrectly called P. citreonigrum, see [17,18]) from the unrelated section Sclerotiora will apparently only produce atlantinone A after treatment with epigenetic modifiers. This difference in ability to express certain secondary metabolites is parallel to our recent findings of orsellinic acid and related compounds being produced at standard conditions by Aspergillus nidulans [19], while Schroeckh et al. [16], only saw production of these compounds when the fungus was grown in a bacterial co-culture, but in the latter case the same fungus was examined in different laboratories under different conditions.

Altogether we have demonstrated that what appears to be a product of a silent pathway in one species, can be a major compound in other species when grown on standard laboratory conditions. Thus, we have shown that atlantinone A previously only expressed by the use of epigenetic modifiers in P. hirayamae is readily produced by P. solitum, P. discolor, P. commune, P. caseifulvum, P. palitans, P. novaezelandiae and P. monticola on the common media CYA and YES.