It has been estimated that 700,00 people a year die from infections with microbial pathogens that are resistant to antibiotic treatment and the problem continues to grow due to a lack of new antibiotics being developed for clinical use [1]. Staphylococcus aureus is the most common cause of human bacterial infections in hospitals, long-term care centers, and community settings [2]. Methicillin-resistant S. aureus (MRSA), one of the “ESKAPE” pathogens [3], is resistant to all known β-lactam antibiotics, and clinical isolates of MRSA have shown reduced susceptibility to vancomycin and resistance to linezolid and daptomycin [4]. This perilous resistance landscape has created an urgent need for new antibiotics to treat MRSA infections. The majority of antibiotics available for use by clinicians are microbial natural products or synthetic analogs of microbial natural products [5]. For example, the widely used semisynthetic β-lactam antibiotic methicillin was inspired by penicillin G, a metabolite of the fungus Penicillium chrysogenum. Fungi and other microbial natural product sources continue to be a major focus of new antibiotic discovery efforts in our laboratories [6,7,8] and elsewhere [5].

In 2015, we reported the production of the new antibiotic solanioic acid (1) (Fig. 1) by cultures of the fungus Rhizoctonia solani isolated from the tubers of the Sri Lankan weed Cyperus rotundas [9]. Solanioic acid (1) is a degraded and rearranged steroid with an unprecedented carbon skeleton featuring a highly functionalized conjoint ring system. It showed potent in vitro antimicrobial activity against a small panel of Gram-positive bacteria, including MRSA, with MICs of 1 μg ml−1, but it was only weakly active against the yeast Candida albicans (MIC 16 μg ml−1) and inactive against Gram-negative bacteria at 64 μg ml−1. Solanioic acid (1) represents a novel antibiotic scaffold with significant in vitro activity against MRSA that warrants further investigation as an antibacterial drug lead.

Fig. 1
figure 1

Compounds 1 to 4

Steroids with contracted B rings are rarely encountered as natural products, with less than a dozen examples reported from human plasma [10], plants [11, 12], sponges [13, 14], and starfish [15]. Nearly all of these natural products, exemplified by 2 and 3 (Fig. 1) isolated from a plant and a sponge, respectively, have five-membered B rings containing an aldehyde appendage with a β-tertiary alcohol indicating that the ring contraction arose from oxidative cleavage of a B ring double bond followed by an aldol condensation to give the new five-membered ring. The location of the tertiary alcohol at C-5 or C-8 presumably predicts whether the contraction was a 5(6→7)abeo (e.g. 2) or 8(7→6)abeo (e.g. 3) process, but there are no reported stable isotope feeding studies that confirm this expectation. Solanioic acid (1) appears to be the only B ring contracted steroid isolated to date from a fungus and since it lacks a tertiary alcohol at either C-5 or C-8 there is no telltale marker unambiguously predicting whether its biogenesis involved a C-5/C-6 or C7/C-8 alkene bond scission. The position of the alkene in the B ring of solanioic acid (1) might suggest a 5(6→7)abeo contraction; however, the configuration at C-8 in 1 is the opposite of that found in a normal steroid skeleton indicating that the precursor to 1 may have had a Δ [7, 8] alkene that was cleaved as we originally proposed (Scheme 1) [5].

Scheme 1
scheme 1

Proposed biogenesis of solanioic acid (1) from fungisterol (5) [5].

In order to distinguish between the two possible ring B contraction routes, we have carried out feeding experiments with [1-13C]-acetate, [2-13C]-acetate, and [1,2-13C]-acetate. Unexpectedly, workup of one of the feeding study cultures yielded not only solanioic acid (1) but also its C-9 epimer 4 (Fig. 1). The details of the isolation and structure elucidation of 9-epi-solanioic acid (4) and the results of the 13C labeling studies are described below.

As part of our report of solanioic acid (1), we proposed a biogenetic route to its new carbon skeleton that involved an oxidative cleavage of the C-7/C-8 alkene bond in fungisterol (5) to give a C-8 ketone and a C-7 aldehyde that underwent an intramolecular C-6/C-8 aldol condensation to give a contracted five-membered ring B with aldehyde and tertiary alcohol substituents (Scheme 1) [5]. An alternative route to the contracted ring B, that was not mentioned in our original report, would involve oxidative cleavage of the C-5/C-6 alkene bond in ergosta-5,22-dien-3-ol (6) to give a C-5 ketone and a C-6 aldehyde followed again by an intramolecular C-7/C-5 aldol condensation to give a contracted ring B with aldehyde and tertiary alcohol substituents (Scheme 2). Both ring B contracted intermediates could be elaborated with virtually identical common biosynthetic transformations to solanioic acid (1) (Schemes 1 and 2) so that these proposals do not favor one alkene cleavage over the other.

Scheme 2
scheme 2

Proposed biogenesis of solanioic acid (1) from ergosta-5,22-dien-3-ol (6)

Figure 2 illustrates that the aldehyde substituent at C-7 should be labeled by [1-13C]-acetate if the ring B cleavage involves the C-5/C-6 bond and by [2-13C]-acetate if the cleavage involves the C-7/C-8 bond. In addition, C-5 and C-6 should be part of the same intact acetate unit only if the cleavage is at C-7/C-8. Therefore, the feeding studies were expected to clearly distinguish between the two pathways.

Fig. 2
figure 2

Predicted 13C labeling patterns from feeding [1,2-13C]acetate. Panel A illustrates labeling resulting from cleavage of C-7/C-8 and Panel B illustrates labelling from C-5/C-6 cleavage

Stable isotope feeding experiments were carried out by growing cultures of R. solani on the surface of solid agar containing potato dextrose media and either [1-13C]-acetate, [2-13C]-acetate, or [1,2-13C]-acetate. The cultures were allowed to grow at RT for 7 days before harvesting by cutting the solid agar and attached fungal mass into small pieces and extracting them by soaking in ethyl acetate. Crude ethyl acetate extracts were concentrated in vacuo and fractionated using gradient Si-gel and Sephadex LH20 chromatography, and reversed phase HPLC, to give pure 13C-labeled solanioic acid (1) from all three cultures and 9-epi-solanioic acid (4) only from the [1-13C]-acetate feeding experiment.

9-Epi-solanioic acid (4) was isolated as a clear viscous oil that gave an [M + Na]+ ion at m/z 479.2765 (calcd for C28H40O5Na, 479.2773) in the HRTOFESIMS appropriate for a molecular formula of C28H40O5 identical to that of solanioic acid (1). The 1D 13C and 1H NMR data obtained for 4 showed a close resemblance to the data obtained for solanioic acid (1) (Table 1) and analysis of the 2D NMR data obtained for 4 (Supporting Information) showed that it had the same planar structure as 2 (Fig. 3a). Therefore, 4 was presumed to be a stereoisomer of 2. The largest deviations in the NMR data for 1 and 4 were observed in the 1H and 13C chemical shifts for H-7 (δ 1: 9.73; 4: 9.61), H-8 (δ 1: 3.86; 4: 3.73), and H-9 (δ 1: 2.66; 4: 2.51) and C-8 (δ 1: 40.8; 4: 43) and C-9 (δ 1: 65.4; 4: 60.1) indicating that the configurational differences were located in this region of the stereoisomers. ROESY correlations were observed between the H-9 (δ 2.66) and Me-18 (δ 2.51) resonances in 1 consistent with H-9 being cis to Me-18, while ROESY correlations were observed between the H-9 and H-1ax (δ 1.25) resonances in 4 consistent with H-9 being cis to H-1ax and trans to Me-18 (Fig. 3b). The NMR data described above were consistent with 4 being the C-9 epimer of 1. Upon sitting in the NMR tube in DMSO-d6, 4 spontaneously and quantitatively converted to 1 confirming our structure assignment.

Table 1 NMR data for 9-epi-solanioic acid (4) and comparison with the NMR data of solanioic acid (1) recorded in DMSO-d6
Fig. 3
figure 3

COSY (Panel A), HMBC (Panel A), tROESY correlations (Panel B), and J couplings (Panel B) for H-8 and H-9 of 1 and 4 used to assign the planar structure and configuration of 9-epi-solanioic acid (4)

Examination of the 13C NMR spectrum of 1 and 4 isolated from the [1-13C]-acetate and [2-13C]-acetate feeding experiments showed that C-7 (1: δ 188.9), C-11 (1: δ 202.9), and C-12 (1: δ 177.4) were labeled by [1-13C]-acetate and not labeled by [2-13C]-acetate as shown in Fig. 4 and Supporting Information Table SI-1. The 13C NMR of solanioic acid (1) isolated from the [1,2-13C]-acetate feeding experiment showed intense flanking doublets for C-11 and C-12 indicative of these carbons coming from intact acetate units, while the C-5 (δ 163.8) resonance showed no strong flanking doublet but rather just an enriched natural abundance singlet and a series of weak unresolved doublets at its base indicating that it was not part of an intact acetate unit but had multiple long-range couplings to other acetate units in the same molecules resulting from very high incorporation of [1,2-13C]-acetate. The complete incorporation pattern resulting from the three feeding experiments is shown in Fig. 4. This labeling pattern is exactly what is expected for cleavage of a C-5/C-6 alkene bond followed by an intramolecular aldol condensation between C-7 and C-5 to give the contracted ring B as outlined in Scheme 2 and Fig. 2b. The isolation of 9-epi-solanioic acid (4) suggests that first natural product formed by this pathway is 4, that has the normal C-9 steroid configuration. 9-epi-solanioic acid (4) is then either biosynthetically or spontaneously epimerized to solanioic acid (1), which is the thermodynamically most stable epimer.

Fig. 4
figure 4

Labeling pattern observed in solanioic acid (1) resulting from separate feeding experiments with [1-13C]-acetate, [2-13C]-acetate, and [1,2-13C]-acetate