Spirocyclic cladosporicin A and cladosporiumins I and J from a Hydractinia-associated Cladosporium sphaerospermum SW67

Here, we report the isolation and characterization of three new spirocyclic natural products named cladosporicin A (1), cladosporiumins I (2) and J (3) from the fungus Cladosporium sphaerospermum SW67. Cladosporicin A contains an unprecedented 2,7-diazaspiro[4.5]decane-1,4-dione skeleton conjugated with one 2,4-pyrrolidinedione moiety (tetramic acid). Cladosporiumins I and J are stereoisomers and are built of a tetramic acid core structure with a quaternary (C-3) center carrying a trans-hexylenic alcohol side chain and a six-membered lactone ring. The absolute structures of compounds 1–3 were elucidated by a combination of 1D and 2D NMR spectroscopy, modified Mosher's analysis, quantum chemical ECD calculations and computational NMR chemical shift calculations. Comparative genome sequence analysis led us to assign the putative new PKS-NRPS hybrid gene cluster (cls). Finally, we assessed bioactivities of compounds 1–3 in a series of pharmacological tests and found weak cytotoxicity against four human breast cancer cell lines (IC₅₀ ∼ 70–90 μM).

---

Introduction

Host-associated microbes have a long history of providing biologically and medically useful natural products such as antibiotics, as well as antifungal, anticancer and immune-modulatory agents. At the same time, natural products also play important roles in the defense and/or survival of the producing organisms and their evolutionary histories are the key distinguishing features that confer their remarkable success in biology and medicine.¹ Analyzing the chemical and biosynthetic repertoire of symbiotic microorganisms represents therefore an important general discovery paradigm for both biology and chemistry.

We recently started to investigate the metabolic potential of bacteria associated with the marine invertebrate Hydractinia echinata (Cnidaria), a colony-forming hydrozoan.² In continuation of this work, we analyzed eleven fungal co-isolates obtained from the tissue of H. echinata polyps for their metabolic potential as fungal metabolites exhibit an extraordinary range of molecular diversity and a remarkable ability to modulate or inhibit biological processes.^3,4

Thus, we selected isolate SW67, most closely related to the species Cladosporium sphaerospermum, for co-culture assays and metabolomic studies. NMR and MS-guided analysis of bacterium-fungus co-culture set-ups led to the isolation and structural characterization of three novel natural products (1–3, Fig. 1) containing a 2,4-pyrrolidinedione (tetramic acid) core and were named cladosporicin A and cladosporiumins I and J, respectively.

Fig. 1 Absolute structures of cladosporicin A [(E)-1 and (Z)-1], cladosporiumins I (2) and J (3).

The cladosporicin motif represents a unique hybrid scaffold and only recently several related cladosporiumin derivatives (5–7, Fig. 5) were reported.⁵ Based on our acquired whole genome sequence of SW67 and gene expression studies, we propose the putative PKS-NRPS hybrid gene cluster (cls) as genetic basis for the production of 1–3 and discuss the underlying biosynthetic steps responsible for the production of the growing family of tetramic acid derived metabolites.⁶

Results and discussion

Ten polyps of H. echinata (Alfred Wegener Institute, Sylt, Germany) were aseptically cut and homogenated. Samples were serially diluted with filtered sterile seawater and potato dextrose agar (PDA) plates were inoculated (100 μL of the dilution series) and incubated for 1–3 weeks at room temperature. Colonies with fungal morphologies were selected and transferred to new agar plates until pure cultures were obtained. Based on the internal transcribed spacer (ITS) gene sequence analysis of eleven isolated and distinct fungal phenotypes (GenBank accession: MH482916-MH482926),⁷ we determined that our isolates belong to five different genera of three distinct phyla (Aspergillus, Penicillium, Hortaea, Engyodontium, Cladosporium) (Fig. 2A). We then assessed the antimicrobial activities of culture extracts of all isolates.⁸ In short, methanolic extracts of 14 day old plate agar cultures (PDA) were purified using pre-packed C18 columns and were fingerprinted by LC-MS-UV analysis. Antibacterial and antifungal activities were assessed against a panel of bacterial and fungal human pathogenic test strains. As expected Penicillium and Aspergillus isolates, known to produce a broad range of bioactive metabolite,⁹ exhibited moderate to strong antimicrobial activity against several test strains (Fig. 2A). However, culture extracts of Engyodontium album MSW11-1, Hortaea werneckii MSW12-1B, and C. sphaerospermum SW67 exhibited only moderate antibacterial activity against Gram-positive bacteria (Fig. 2A). Interestingly, H. werneckii has been identified as an extreme halotolerant fungus from marine habitats and belongs to the black yeast group.^10,11 Current efforts in our group are now directed towards identifying key metabolites important in the interactions between Hydractinia and Hortaea. In contrast, isolate SW67 is most closely related to C. sphaerospermum, a species that has also been isolated from deep-sea environments and is known to produce several bioactive natural products of unique structure, including cytotoxic sporiolides,¹² antibacterial Sumiki's acid,¹³ and antifouling 3-phenyl-2-propenoic acid.¹⁴ Most recent examples include the PKS–NRPS hybrid family cladosporiumins A–H (5–13) and the closely related cladosins A–G (14–19) (Fig. 5).^15,16

Fig. 2 (A) Maximum likelihood phylogenetic tree based on the Kimura 2-parameter model using the ITS gene sequence. Best DNA model was generated and robustness of interferes topologies was evaluated after 1000 bootstraps (>50% are shown). Right: Correlated heatmap showing antimicrobial activities against test strains (zone of inhibition in mm in standardized assay). (B) Axenic culture of SW67 and co-cultivation studies with (C) Bacillus sp. SW7 and (D) Aspergillus sp. MSW12-1A. (E) Comparative HPLC-based analysis of culture extracts obtained from bacterium-fungus co-cultivation [C. sphaerospermum SW67 against Bacillus sp. SW7, Cobetia sp. SW148 or P. luteoviolacea DSM 6842 (representative UV trace at 298 nm)]. Asterisk: Retention time of compounds 2 and 3. (F) Selected ion chromatogram of extracts (m/z = 350.15 [M + H]⁺ at 4.08 min).

C. sphaerospermum sp. SW67 (from now on named SW67) showed only minor antibacterial activities in our standard tests. To induce the production of antimicrobial “cryptic” and potentially novel metabolites,¹⁷ we pursued co-cultures of SW67 against co-occurring bacteria (mostly Gram-negative) and the herein reported fungal isolates that mimic natural stress factors. Co-culture assays (mycelium and the interaction zone) were subsequently analyzed by UHPLC-MS to detect metabolic changes.

Overall, three out of thirteen bacterial strains (Pseudoalteromonas sp. PS5, P. luteoviolacea DSM 6061, Bacillus cereus SW68) strongly inhibited growth of SW67 (Fig. S5†).¹⁸ Currently, we hypothesize that the observed antifungal activity of Pseudoalteromonas sp. PS5 and P. luteoviolacea DSM 6061 is due to the secretion of the reported cytotoxic compound tetrabromopyrrole and violacein, respectively.¹⁹ However, the elucidation of the chemical identity of other antifungal metabolites from Pseudoalteromonas remains a topic of current investigations. All other bacterial strains showed only minor influence on growth behaviour and morphology of SW67.

Subsequent fungus-fungus co-cultivation studies of SW67 and Hydractinia-associated fungi showed that SW67 was not able to inhibit growth of any fungal interaction partner and only overall reduced growth rates of both partners and green pigment formation in SW67 was observed. Here, it is noteworthy, that only one out of ten isolates (Aspergillus sp. MSW12-1A) was capable of inhibiting growth of SW67 (Fig. 2D, S6†).

Subsequently, the methanolic extracts of axenic cultures and co-culture assays (mycelium and the interaction zone) were analyzed by UHPLC-MS to detect induced metabolic changes within SW67. LC-MS analysis clearly revealed a reproducible, but unique metabolic spectrum for each co-culture set-up. But no general upregulation of single “cryptic” defensive metabolites in SW67 was observed (Fig. 2E and F, Fig. S9–11 and Table S4†).

Amongst several known and abundant metabolites produced by SW67 (e.g. cinnamic acid (m/z = 147.0 [M − H]⁻) and daidzein (m/z = 253.0 [M − H]⁻),^20,21 several so far unreported UV/VIS-m/z signals were detected during co-cultivation, including a distinct UV signal (298 nm) that correlated to a m/z = 350.15 [M + H]⁺.

To scale up the production of the molecular target for structural identification, we pursued a series of growth studies of SW67. Overall, our growth studies on different nutrient-rich media showed that at least two weeks incubation time was necessary to reach significant biomass of SW67 and metabolite production levels in liquid media or on plates (Fig. S7–S9†). When grown for longer than 4 weeks or when grown on saline minimal medium, metabolite production decreased significantly and the target ion was negligible (Fig. S9†). However, none of the tested media significantly induced production of the target ion compared to co-culture set-ups. Here, it is interesting to note that metabolite production in SW67 was downregulated significantly, when the strain was sub-cultivated twice as axenic culture for at least one month. Only co-cultivation with bacteria or reactivation from cryostocks fully restored metabolite production in SW67 to previous levels. Thus, we solely used “activated” SW67 culture for large scale cultivation on PDA or MEA agar.

To isolate the target ion, we performed large scale plate cultivation on PDA and MEA agar plates for 14 days at 25 °C in the dark. Mycelium-covered plates were then extracted with MeOH, filtrated and evaporate to dryness.^22,23 Subsequent solvent-based fractionation of crude extracts and MS-guided repetitive semi-preparative C18 reverse-phase HPLC yielded three new natural products 1–3 with very similar NMR signal pattern.

The first compound, cladosporicin A (1), was isolated as a yellowish oil and its molecular formula C₂₁H₂₇N₃O₅ was calculated from the deprotonated ion peak at m/z 400.1873 [M − H]⁻ (calcd 400.1872 for [M − H]⁻). The ¹H NMR spectrum (Table S7†) exhibited five methyls [δ_H 1.14 (d, J = 6.0 Hz, H-11), 1.83 (s, H-14), 1.89 (s, H-24), 2.16 (s, H-23), and 2.24 (s, H-13)], three methylenes [δ_H 2.86 (dd, J = 20.0, 11.0 Hz, H-7α), 3.74 (dd, J = 20.0, 5.5 Hz, H-7β), 1.12 (m, H-9a), 1.50 (m, H-9b), and 3.64 (m, H-20)], and two methines [δ_H 2.56 (m, H-8) and 3.90 (m, H-10)]. The ¹³C NMR data (Table S7†) showed a total of 21 carbon resonances attributable for five methyl (δ_C 18.7, 19.2, 21.1, 21.5, and 24.5), three methylene (δ_C 28.8, 41.1, and 45.5), two methine (δ_C 32.2 and 64.4), and eleven quaternary carbons (δ_C 52.3, 97.5, 122.4, 127.6, 130.5, 131.2, 169.2, 171.4, 173.4, 187.2, and 199.2). Analysis of 2D NMR spectra (¹H–¹H COSY, HSQC, and HMBC) led to the establishment of the planar substructures (parts A and B) (Fig. 3). The ¹H–¹H COSY correlations of H₂-7/H-8/H₂-9/H-10/H₃-11 suggested the presence of a spin system from C-7 to C-11 (part A) and a secondary alcohol at C-10 [δ_H 3.90 (1H, m); δ_C (64.4)], which was also verified by HMBC correlations from H-8, H₂-9, and H₃-11 to C-10. The highly substituted 2,4-pyrrolidinedione skeleton (part A) was verified by HMBC correlations of H₂-7/C-3, H₂-7/C-6, H₃-14/C-4, H₃-14/C-5, H₃-14/C-12, H₃-14/C-13, H₃-13/C-4, H₃-13/C-5, H₃-13/C-12. In part B, the second 2,4-pyrrolidinedione scaffold was determined based on the analysis of the ¹H–¹³C long-range correlations of H₂-20/C-16, H₂-20/C-17, H₂-20/C-18, H₃-24/C-18, H₃-24/C-19, H₃-24/C-23, H₃-23/C-18, H₃-23/C-19, and H₃-23/C-22. The HMBC correlations of H₂-7/C-3, H₂-7/C-6, H-8/C-6, H₂-20/C-6, and H₂-20/C-8 afforded the linkage between the parts A and B. The relative configuration of compound 1 was determined based on ¹H, HMBC and ROESY data. Intriguingly, 1D and 2D NMR data suggested that compound 1 exists as a non-separable mixture of E/Z isomers (tetra-substituted Δ^3/6 double bond, CD₃OD) in a ratio of 2 : 1 [(E)-1 : (Z)-1]). In particular, the down-field ¹³C shift of the hydrogen-bonded carbonyl group (with NH-21) compared to the free carbonyl group [(E)-1: δ_C 171.4 (C-2), 187.2 (C-4); (Z)-1: δ_C 173.2 (C-2), 185.2 (C-4)] was indicative for the stereochemistry of the Δ^3/6 double bond in 1. The relative configurations at C-8 and C-10 were assigned based the combination of ROESY, homonuclear J-resolved spectroscopy (JRES) and HSQC-HECADE data.²⁴ In particular, the observed ³J_HH and ³J_CH indicated a dominant gauche-type arrangement for the C-8 to C-11 spin system, and led us to propose the relative assignment of 8S* and 10R* (Fig. 3). To determine the absolute stereochemistry, Mosher's esterification protocol was pursued. However, all attempts to convert compound 1 failed and afforded only complex product mixtures. Here it is noteworthy that Mosher reactions for all three compounds (1–3) failed or were very low yielding, most likely because of hydrogen bonding properties between hydroxyl group at C-10 and ester groups in the 2,4-pyrrolidinedione skeleton as proposed in the calculation process of the GIAO magnetic shielding tensor values of compounds 2 and 3 (Fig. S52–58†).

Fig. 3 (A) Structures of compounds (E)-1 and (Z)-1. (B) Key COSY () and HMBC () correlations. (C and D) ROESY and J-based configuration analysis of compound 1 at C-9/C-10, and C-8/C-9, respectively. (E) Experimental (black) and calculated ECD spectra of compound 1.

Therefore, the ECD spectrum of 1 was measured and the experimental data compared with quantum chemical ECD calculations of four possible diastereomers [1a(8S,10R,17S), 1b(8S,10R,17R), 1c(8R,10S,17S), and 1d(8R,10S,17R), Fig. 3]. Experimental ECD data of 1 matched well to the calculated ECD curves of 1a and 1c, indicating 17S as the most likely absolute configuration of the spiro-carbon center.

Subsequently, the absolute configurations of two stereogenic centers at C-8 and C-10, respectively, were achieved by the gauge-including atomic orbital (GIAO) NMR chemical shifts calculation including DP4 probability.²⁵ Conformers of both diastereomers 1a(8S,10R,17S) and 1c(8R,10S,17S) were obtained by the universal force field (UFF) and density functional theory (DFT) settings (B3-LYP functional/M3 grid size) and the basis set def-SV(P) for all atoms. These conformers were calculated at the B3LYP/def2-TZVPP theory level and the chemical shifts were calculated and Boltzmann-averaged at 298.15 K from the GIAO magnetic shielding tensor values. Comparison of the experimental NMR data of 1 with the calculated NMR values of the diastereomers 1a(8S,10R,17S) and 1c(8R,10S,17S) (Table S9, Fig. S53–58†), followed by the DP4 analysis suggested that diastereomer 1a(8S,10R,17S) exhibited a DP4 probability score of 95.5% given in both ¹H and ¹³C NMR chemical shift values.

Cladosporiumin I (2) was obtained as a yellowish oil and its molecular formula was deduced to be C₁₉H₂₇NO₅ from the molecular ion peak at m/z 348.1810 [M − H]⁻ (calcd. 348.1811) in HR-ESI-MS. The ¹H NMR spectrum (Table S8†) displayed the proton resonances of characteristic four methyl [δ_H 1.04 (d, J = 6.0 Hz, H-11), 1.31 (d, J = 6.0 Hz, H-20), 1.84 (s, H-14), and 2.17 (s, H-13)], four methylene [δ_H 1.46 (ddd, J = 13.5, 12.0, 12.0 Hz, H-18α), 1.81 (ddd, J = 13.5, 4.0, 2.0 Hz, H-18β), 1.99 (m, H-9a), 2.10 (m, H-9b), 2.42 (d, J = 6.0, H-6), 2.62 (dd, J = 18.0, 10.5 Hz, H-16α), and 2.69 (dd, J = 18.0, 7.0 Hz, H-16β)], and five methine signals [δ_H 2.46 (m, H-17), 3.63 (m, H-10), 4.39 (m, H-19), 5.24 (dt, J = 15.0, 7.5 Hz, H-7), and 5.53 (dt, J = 15.0, 7.5 Hz, H-8)]. The ¹³C NMR data (Table S8†) exhibited the carbon resonances of four methyl (δ_C 19.8, 22.0, 22.6, and 23.3), four methylene (δ_C 32.0, 33.4, 37.1, and 44.0), five methine (δ_C 37.9, 69.1, 78.6, 126.7, and 134.3), and six quaternary carbons (δ_C 59.2, 127.2, 132.0, 174.0, 175.9, and 202.1). The analysis of ¹H–¹H COSY, HSQC and HMBC spectra delineated the planar structures of three units (parts A–C). In part A, the ¹H–¹H COSY spectrum revealed the connectivity of H₂-6 to H₃-11, and the HMBC correlations from H₃-11 to C-10 and C-9 suggested the existence of secondary alcohol at C-10 (δ_C 69.1). For part B, the ¹H–¹H COSY spectrum indicated the correlations from H₂-16 to H₃-20 and the connectivity of C-16 to C-20. The existence of six-membered lactone ring was established by HMBC correlations of H₂-16/C-15, H₂-16/C-18, H₃-20/C-18, and H₃-20/C-19. Additionally, the relatively down-fielded spectroscopic values (δ_H 4.39/δ_C 78.6) of an oxygenated methine at C-19 confirmed an ester group in part B. In part C, HMBC correlations from H₃-13/H₃-14 to C-4, C-5 and C-12 led to the assignment of highly modified 2,4-pyrrolidinedione scaffold with Δ^5/12 double bond. The linkage among A–C parts was deduced by ¹H–¹³C long-range correlations of H₂-6/C-2, H₂-6/C-4, H-7/C-3, H-17/C-2, and H-17/C-4. The ROESY correlations between H-17 and H-19 indicated the co-facial orientation and its relative configuration was tentatively established as 17R* and 19S*. Modified Mosher's reaction and ¹H NMR and TOCSY spectra of the (R)- and (S)-MTPA ester derivatives suggested the absolute configuration of C-10 as S (Fig. 4).

Fig. 4 (A) Structures of compounds 2 and 3. (B) Key COSY () and HMBC () correlations of parts A, B and C. (C) Analysis of Mosher derivatives 4a and 4b derived from 2 and 3, respectively; Δδ (δ_S − δ_R) values are shown. (D) Experimental (black) and calculated ECD spectra of compound 2.

The absolute configuration of the quaternary spiro-carbon center at C-3 was determined by quantum chemical ECD calculations. Comparison of theoretical spectra of four possible diastereomers 2a(3R,17R,19S), 2b(3S,17R,19S), 2c(3R,17S,19R), and 2d(3S,17S,19R) revealed two distinct signal pattern belonging to 2a/2c((3R,17R,19S)//(3R,17S,19R)) and 2b/2d((3S,17R,19S)//(3S,17S,19R)) respectively (Fig. 4D). Direct comparison revealed that experimental ECD spectrum of 2 was most similar to the spectral pattern of 2a and 2c; thus the absolute configuration of the quaternary spiro-carbon center was deduced to be 3R (Fig. 4).^5,26

The two observed ECD pattern for the diastereomer pairs 2a/c and 2b/d are most likely a result of a switch in chirooptic properties caused by the stereogenic changes in the conformational restrained spiro-center (C-3).²⁶ In contrast, configurational changes (C-17, C-19) of the freely rotating side chains had little effect on the calculated ECD spectra. The absolute configurations of C-17 and C-19 were deduced by GIAO NMR chemical shifts calculation described before. Comparison of experimental and calculated both ¹H and ¹³C NMR chemical shift values suggested the absolute configuration of 2 to be 3R,17R,19S (DP4 probability score of 100%, Fig. S53–58†) and collectively, the absolute structure of 2 was deduced as shown in Fig. 4.

The molecular formula of the third isolated compound, cladosporiumin J (3), was calculated from the deprotonated ion peak at m/z 348.1820 [M − H]⁻ (calcd for 348.1811) as C₁₉H₂₇NO₅, which was identical to that of 2. Comparison of 1D and 2D NMR spectra (Table S8†) revealed an almost identical spin and chemical shift pattern as compound 2, and only showed chemical shift variations for C-16 to C-19. Analysis of ROESY data revealed strong correlation between H-17 and H-19 suggesting either the relative configuration (17S,19R) or (17R,19S). Again, Mosher analysis indicated the absolute configuration of C-10 to be again (S)-C-10.

Then, comparative ECD analysis of four possible diastereomers revealed that 3 is most similar to 2b/2d((3S,17R,19S)//(3S,17S,19R)) indicating the absolute configuration to be S (C-3) (Fig. S40†).^5,26 These findings support the previous assignment of compound 2. Finally, NMR chemical shift calculation of the two most likely diastereomers 3b(3S,17R,19S) and 3d(3S,17S,19R) suggested the absolute configuration to be (3S,17S,19R) (Fig. S52–58†).

We were particularly intrigued by the fact that cladosporiumin derivatives 5–7 and 10–19 (Fig. 5) have been previously reported from related Cladosporium species (Cladosporium sp. SCSIO z025, C. sphaerospermum 2005-01-E3 and Cladosporium sp. OUCMDZ-1635) (Fig. 5 and Table S5†). Thus, we analyzed three unrelated Cladosporium species (C. sphaerospermum SF006509 and SF011511 (Jena Microbial Resource Collection) and C. pergangustum (CPC 18648)), isolated from a fungus-growing termite system, for the production of cladosporiumin-like compounds. Again, MEA plates were inoculated with the respective spore suspension and incubated for 14 days at 25 °C in the dark. Mycelium-covered plates were extracted with MeOH and purified using pre-packed C18 column and fingerprinted by LC-MS. Comparative MS-analysis revealed that neither C. sphaerospermum SF006509 and SF011511 nor C. pergangustum CPC 18648 produced compounds 1–3 (Fig. S6†).

Fig. 5 Overview of all reported cladosporiumins, cladosins and related structures isolated from Cladosporium spp.

Cladosporicin A, cladosins, and cladosporiumins are composed of a tetraketide unit and a valine residue, involving both polyketide synthase (PKS) and non-ribosomal peptide synthetase (NRPS) biosynthetic pathway.²⁷ In analogy to the currently accepted biosynthetic model for tetramic acid formation, the formation of 1–3 involves the condensation of an amino acid and an activated polyketide moiety catalyzed by a PKS–NRPS hybrid to yield the 1,3-dione-5,7-diol conjugate; formation of the 2,4-pyrrolidinedione skeleton occurs most likely in an off-loading step via a Dieckmann-type cyclization.

To support the general assumption that compounds 1–3 are PKS–NRPS derived hybrid molecules, we performed ¹³C labelling experiments using growth media supplemented with ¹³C-valine and ¹³C-acetate, respectively. Subsequent HRMS-analysis of enriched compound mixtures revealed the incorporation of one valine unit and a PKS-typical CH₃¹³CO₂-dependent HRMS (m/z) pattern of cladosporiumin derivatives (Fig. S11 and S12†).²⁸ Furthermore, ¹³C-labeling pattern and NMR-analysis indicated a low abundance of the structurally related cladodionen 9 and cladosin derivatives of type 14–18 (Fig. S13–S15†).²⁹ Determination of their absolute structures is currently ongoing.

To identify putative biosynthetic gene clusters encoding for the production of the isolated compounds, the genome of SW67 was sequenced using Illumina MiSeq paired-end v3 chemistry (300 cycles per read) and a draft genome was assembled using SPAdes v.3.6.2. Secondary metabolism-related genes were predicted using FungiSmash³⁰ and NRPS predictor.³¹ In silico analysis of the fungal genome revealed numerous cryptic gene loci putatively coding for terpene cyclases, NRPS, PKS, or in one case a PKS-NRPS hybrid cluster (cls) (Table S1†).

Based on the biosynthetic logic for tetramic acid formation, we currently assume that the PKS–NRPS hybrid cluster (cls) represents the genetic basis for cladosporicin A, cladosporiumins and cladodionen production in SW67. The cluster includes twelve genes, annotated as clsA – clsL (Fig. 6A and Table S2†), encoding for a siderophore esterase (ClsA), an AMP dependend synthetase/ligase (ClsD), a γ-glutamyl transferase (ClsF), one PKS-NRPS hybrid (ClsI), a cytrochrome P450 (ClsK), two transport related proteins (ClsB and ClsG) and four genes coding for putative proteins with unknown function (ClsC, ClsE, ClsH, ClsJ). Subsequent gene cluster homology searches and a blast query of our proposed cls gene cluster resulted in the identification of an almost identical non-characterized gene cluster, named clu within the genome of another C. sphaerospermum UM 843 (PRJNA85131) (Fig. S3†). However, it is currently unknown if isolate UM 843 is also able to produce compounds 1–3 or related derivatives.

Fig. 6 (A) Gene map of cls (PKS-NRPS hybrid) gene cluster. KS: Ketoacyl synthase; AT: acyltransferase; KR: ketoreductase; C: condensation domain; A: adenylation domain; PP: peptidyl carrier protein; TD: terminal reductase. The two unidentified domains (X, Y) show moderate similarities to a dehydratase and enoylreductase. (B) Representative gene expression study of clsI (PKS-NRPS) and β-actin (housekeeping gene) using conventional reverse transcriptase PCR. Different cultivation conditions were tested: plate culture (MEA), static liquid culture (MEB), control: RT-(gDNA control without reverse transcriptase enzyme); positive control: gDNA; negative control without DNA or RNA template: C−.

To support our biosynthetic proposal, we performed a classical reverse transcriptase PCR of several biosynthetic genes encoded within the cluster cls. As depicted in Fig. 6 (and Fig. S15 and S16†) clsA, clsD, clsF, clsI, clsK were constantly expressed in SW67 independent of the tested cultivation condition and correlates with constant production of 1–3.

Based on the composition of the putative gene cluster cls and the biosynthetic logic of tetramic acid formation, we propose four likely scenarios resulting in the formation of cladosins, cladodionen, cladosporicin A, or cladosporiumins (Fig. 7). We currently assume that the proposed iterative PKS might act more promiscuous compared to others by providing different and partially reduced polyketide chains. However, the involvement of a second homologous PKS cannot fully be excluded. In all four proposed scenarios (A–D) the condensed hybrid molecule is released as a tetramic acid core from the hybrid synthase via Dieckmann cyclization. Due to the observed multitude of stereoisomers, several spontaneous/non-enzymatic steps without stereo-control are likely to occur. Here it needs to be acknowledged that more detailed biosynthetic studies to prove the outlined proposals are necessary and require the determination of the respective gene cluster boundaries using, e.g. transcriptome analysis.

Fig. 7 Proposed biosynthetic steps for the formation of the core structure of (A) cladosins, (B) cladodioden, (C) cladosporiumins, and (D) cladosporicin A.

In a first scenario (Fig. 7A), the iterative PKS generates an unsaturated polyketide chain. Dieckmann cyclization will then result in the formation of the highly reactive tetramic acid derivative 20, which likely exists in both, keto and enol form. Subsequent oxidation and elimination of H₂O yields derivatives 21a and 21b, respectively, and reaction with NH₃ results in imine/enamine formation as reported for cladosins (22). In a second scenario (Fig. 7B), the PKS generates a partially reduced polyketide chain that allows an intramolecular condensation reaction yielding cladodionen (9). In a third scenario (Fig. 7C), the PKS generates an almost fully reduced polyketide chain that after Dieckmann-cyclization lacks the carbonyl group and is not able to undergo imine formation (A) nor cyclizations (B). Instead, deprotonation of the C–H acidic position at C-3 under physiological conditions could induce further (unselective) addition reactions. Amongst others, a (spontaneous) nucleophilic 1,4 addition reaction to a PKS-derived unsaturated C-6 unit 25 could result in the addition product 26. Additional intramolecular (spontaneous) esterification would yield the stable 6-membered lactone moiety (cladosporiumin core structure). Here, it needs to be acknowledged that a spontaneous (non-enzymatic) addition reaction sequence does not explain the observed dominant formation of 27 derivatives, but could explain the unselective formation of C-3 and C-17 stereoisomers.

Finally, a fourth scenario (Fig. 7D) could explain the formation of cladosporicin A. Here, the key building block could be derived from propionic acid and valine, which forms after Dieckmann cyclization and oxidation the highly reactive putative intermediate 28. A stepwise or concerted addition reaction of 28 and cladosins (22) would yield the spirocyclic derivative 1.

Here, we acknowledge that heterologous expression of the respective hybrid gene cluster is necessary to proof the proposed pathway. Recently, it was shown that natural products containing tetramic acids exhibit a remarkable diversity of biological activities, including antimicrobial activity (e.g. nocamycin, oleficin, PF1052, BU-2313A and BU-2313B, melophlins), antiviral, antiulcerative as well as cytotoxicity properties (aurantoside, equisetin, discodermide, cylindramide, cladosins).^23,32 Thus, we tested enriched methanolic extracts of SW67 and pure compounds for their bioactivities including their insulin secretion effects, the inhibition of nitric oxide production, renoprotective effects, protease inhibition activities and their antimicrobial properties against human standard pathogens.

Overall, no significant activities were measurable and only weak inhibitory effects of compounds 1–3 on tumor cell growth in human breast cancer cells (Bt549, HCC70, MDA-MB-231, and MDA-MB-468) by using an SRB bioassay were detected in all four cell lines with IC₅₀ values in the range of 70–85 μM (Table 1).³³

Table 1 Cytotoxicity of compounds 1–3 against four cultured human breast cancer cell lines using the SRB bioassay in vitro

Compound	IC50 ^a (μM)
	Bt549	HCC70	MDA-MB-231	MDA-MB-468
a IC50 value of compounds against each tumor cell line, defined as the concentration (μM) that caused 50% inhibition of cell.b Etoposide served as positive control.
1	70.88	74.48	75.54	79.36
2	76.18	85.29	82.37	81.44
3	78.96	76.41	79.27	74.64
Etoposide^b	1.82	1.76	2.27	2.08

---

Conclusions

The analyses of host-associated microorganisms have resulted in the identification of many new microbial species and characterization of novel natural product skeletons. Here we isolated eleven fungi from the tissue of the hydroid polyp of H. echinata and evaluated their antimicrobial activity against a panel of standard test strains. Co-cultivation studies with Hydractinia-associated bacteria and fungi revealed highly dynamic UV-VIS/MS spectra and comparative metabolomic analysis led to the prioritization of several so far unreported target ions. MS-based isolation procedure of one of the identified target ions resulted in the characterization of three previously unreported PKS–NRPS hybrid molecules expanding the cladosin-like family. Whole genome sequencing and classical reverse transcriptase PCR allowed us to identify a putative PKS–NRPS hybrid gene cluster likely responsible for the production of the described compounds and to formulate a biosynthetic proposal for the formation of cladosin-like natural products.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (2018R1A2B2006879). MR and CB are supported by the Jena School for Microbial Communication, and the German Research Council (BE4799/2-1 EXSPHINGO and CRC ChemBioSys 1127). The authors thank M. Küffner, L. Drechsel and A. Schmalzl, for assisting in the cultivation of Cladosporium sphaerospermum SW67. We thank A. Perner, H. Heinecke, and C. Weigel for their help with MS and NMR measurements and antimicrobial activity assays.

References

1. N. Adnani, S. R. Rajski and T. S. Bugni, Nat. Prod. Rep., 2017, 34, 784–814.
2. H. Guo, M. Rischer, M. Sperfeld, C. Weigel, K. D. Menzel, J. Clardy and C. Beemelmanns, Bioorg. Med. Chem., 2017, 25, 6088–6097.
3. P. Bhadury, B. T. Mohammad and P. C. Wright, J. Ind. Microbiol. Biotechnol., 2006, 33, 325–337.
4. J. F. Imhoff, Mar. Drugs, 2016, 14, 19.
5. Z.-h. Huang, X.-h. Nong, X. Liang and S.-H. Qi, Tetrahedron, 2018, 74, 2620–2626.
6. T. L. S. Kishbaugh, Curr. Top. Med. Chem., 2016, 16, 3274–3302.
7. R. H. Nilsson, E. Kristiansson, M. Ryberg, N. Hallenberg and K.-H. Larsson, Evol. Bioinf. Online, 2008, 4, 193–201.
8. J. H. Jorgensen, J. F. Hindler, L. B. Reller and M. P. Weinstein, Clin. Infect. Dis., 2007, 44, 280–286.
9. D. J. Faulkner, Nat. Prod. Rep., 2001, 18, 1R–49R.
10. N. Gunde-Cimerman, J. Ramos and A. Plemenitaš, Mycol. Res., 2009, 113, 1231–1241.
11. M. Lenassi, C. Gostinčar, S. Jackman, M. Turk, I. Sadowski, C. Nislow, S. Jones, I. Birol, N. G. Cimerman and A. Plemenitaš, PLoS One, 2013, 8, e71328.
12. H. Shigemori, Y. Kasai, K. Komatsu, M. Tsuda, Y. Mikami and J. I. Kobayashi, Mar. Drugs, 2004, 2, 164–169.
13. R. Jadulco, P. Proksch, V. Wray, Sudarsono, A. Berg and U. Gräfe, J. Nat. Prod., 2001, 64, 527–530.
14. S.-H. Qi, Y. Xu, H.-R. Xiong, P.-Y. Qian and S. Zhang, World J. Microbiol. Biotechnol., 2009, 25, 399–406.
15. G. Wu, X. Sun, G. Yu, W. Wang, T. Zhu, Q. Gu and D. Li, J. Nat. Prod., 2014, 77, 270–275.
16. G.-H. Yu, G.-W. Wu, T.-J. Zhu, Q.-Q. Gu and D.-H. Li, J. Asian Nat. Prod. Res., 2015, 17, 120–124.
17. F. Y. Lim, J. F. Sanchez, C. C. Wang and N. P. Keller, Methods Enzymol., 2012, 517, 303–324.
18. C. Holmström and S. Kjelleberg, FEMS Microbiol. Ecol., 1999, 30, 285–293.
19. S. Yada, Y. Wang, Y. Zou, K. Nagasaki, K. Hosokawa, I. Osaka, R. Arakawa and K. Enomoto, Mar. Biotechnol., 2008, 10, 128–132.
20. L. Liu, W. R. Hudgins, S. Shack, M. Q. Yin and D. Samid, Int. J. Cancer, 1995, 62, 345–350.
21. Y.-C. Chang, M. G. Nair and J. L. Nitiss, J. Nat. Prod., 1995, 58, 1901–1905.
22. K. H. Kim, T. R. Ramadhar, C. Beemelmanns, S. Cao, M. Poulsen, C. R. Currie and J. Clardy, Chem. Sci., 2014, 5, 4333–4338.
23. H. Guo, R. Benndorf, D. Leichnitz, J. L. Klassen, J. Vollmers, H. Görls, M. Steinacker, C. Weigel, H. M. Dahse, A. K. Kaster, M. Poulsen and C. Beemelmanns, Chem. – Eur. J., 2017, 23, 9338–9345.
24. N. Matsumori, D. Kaneno, M. Murata, H. Nakamura and K. Tachibana, J. Org. Chem., 1999, 64, 866–876.
25. S. G. Smith and J. M. Goodman, J. Am. Chem. Soc., 2010, 132, 12946–12959.
26. D. Padula and G. Pescitelli, Molecules, 2018, 23, 128.
27. B. J. L. Royles, Chem. Rev., 1995, 95, 1981–2001.
28. C. Hertweck, Angew. Chem., Int. Ed., 2009, 48, 4688–4716.
29. G. Zhu, F. Kong, Y. Wang, P. Fu and W. Zhu, Mar. Drugs, 2018, 16, 71.
30. K. Blin, T. Wolf, M. G. Chevrette, X. Lu, C. J. Schwalen, S. A. Kautsar, H. G. Suarez Duran, E. L. De Los Santos, H. U. Kim and M. Nave, Nucleic Acids Res., 2017, 45, W36–W41.
31. M. Röttig, M. H. Medema, K. Blin, T. Weber, C. Rausch and O. Kohlbacher, Nucleic Acids Res., 2011, 39, W362–W367.
32. R. Schobert and A. Schlenk, Bioorg. Med. Chem., 2008, 16, 4203–4221.
33. P. Skehan, R. Storeng, D. Scudiero, A. Monks, J. MaMahon, D. Vis-tica, J. T. Warren, H. Bokesch, S. Kenney and M. R. Boyd, J. Natl. Cancer Inst., 1990, 82, 1107–1112.

---

Footnotes

† Electronic supplementary information (ESI) available: Fermentation procedures; isolation procedures, ESI-HRMS, ¹H NMR, ¹³C NMR, and 2D NMR spectra as well as chemical modifications. See DOI: 10.1039/c8qo01104d

‡ These authors equally contributed to the paper.

---