Dereplication of Neonectria discophora SNB-CN63 metabolomes and Penicillium sclerotiorum SNB-CN111
The ICSN strain collection includes 130 strains of termite mutualistic microorganisms from French Guiana. Each strain was cultivated on solid PDA medium and then extracted by ethyl acetate. The specialized metabolomes of each extract were explored by reverse-phase liquid chromatography coupled with positive electrospray ionization tandem mass spectrometry (RPLC-ESI(+)-MS/MS). The MS/MS data were organized and visualized as a molecular network based on fragmentation spectra homology related to structural homology27 with MetGem software38 based on t-SNE visualization. We analyzed the production of specialized metabolites with high structural specificity in which nodes were clustered (Fig.1, top). We identified ilicicolins produced by Neonectria discophora SNB-CN63 (blue cluster in Fig.1) and azaphilones produced by Penicilliumsclerotiorum SNB-CN111 (red cluster in Fig.1).
Molecular network obtained for specific specialized metabolomes of P. sclerotiorum and N. discophora among 130 crude extracts of termite-associated microorganisms (top). MetGem software (https://metgem.github.io/). Chlorinated metabolites have been annotated by comparison of their MS/MS spectra with databases and are depicted with their typical isotopic pattern related to 37Cl (bottom).
In these two clusters, we observed a specific isotopic pattern at the MS level of chlorinated metabolites via the natural abundance of 37Cl isotopes (24.23%) versus 35Cl (75.77%)39. Thereafter, in the blue cluster, 12 molecules were annotated as ilicicolins by comparison with public or internal databases from previous studies (Table S1, Figure S1)17,18. Eight of these molecules bear a chlorine atom, such as LL-Z1272 or ilicicolinal (1), ilicicolinic acid A (2) and ilicicolinic acid C (3) (Fig.1). In the red cluster, twenty-three azaphilones were annotated using MS/MS spectral comparisons from public or in-house databases (Table S2, Figure S2)20,40,41. Among them, 14 are chlorinated, such as sclerotiorin (4), sclerotioramine (5) or 5-chloroisorotiorin (6), which were previously isolated in our group20,42,43,44.
Ilicicolins are reported as intermediates of halogenated metabolites named ascofuranone and ascochlorin, which are chlorinated by a FAD-dependent halogenase (AscD)45. In the literature, 28 ilicicolin or acid ilicicolinic scaffolds isolated from natural products have been reported, among which 22 are chlorinated46.
In a review article published in 2021, Pavesi et al. reported 676 azaphilones, among which 152 contain a chlorine atom47. These chlorinated metabolites are included in just four azaphilone subfamilies, i.e., chaetoviridins, falconensins, sclerotiorins and luteusins. Among these scaffolds, only the gene cluster involved in chaetoviridin biosynthesis has been elucidated. In that particular case, the enzyme involved in chlorine addition is also a FAD-dependent halogenase (CazI)48.
Hence, we hypothesized that FAD-dependent halogenases catalyze the halogenation of ilicicolins and azaphilones produced by N.discophora SNB-CN63 and P.sclerotiorum SNB-CN111, respectively45,48. To look for these enzymes, we sequenced these two fungal genomes using a combined long- and short-read sequencing approach followed by a hybrid assembly.
The N. discophora genome (ENA accession number: GCA_911649645) was obtained in 26 contigs covering 41.6 Mbp and characterized by a GC content of 54.2%. The completeness of the genome was established at 99.3% using the BUSCO score, and 12,267 genes were predicted (Tables S3, S4)49. The contiguity of the genome assembly is characterized by an N50 length of 4.04 Mbp and an L50 of five. We used the antiSMASH pipeline to predict the existence of a biosynthetic gene cluster31. Two biosynthetic gene clusters were predicted to include a halogenase as a tailoring enzyme. Sequence comparisons using halogenases from the SwissProt and UniProt databases confirmed that no other putative halogenase was present in the N. discophora SNB-CN63 genome. Four genes, a non-reducing polyketide synthase, a prenyltransferase, a non-canonical non-ribosomal peptide synthetase and a halogenase, from candidate cluster Ndi_Ili are similar to the ascofuranone biosynthetic gene cluster (59 to 79% similarity, Table S5). Only the halogenase from candidate cluster Ndi_WSC72 shares 70% similarity with ascD, the ascofuranone halogenase (Table S6). A synteny analysis visualized with clinker50 further confirm the stronger similarity between Ndi_Ili and the ascofuranone biosynthetic gene cluster (Figure S3). Thus, we concluded that Ndi_Ili was the best candidate cluster for ilicicolins biosynthesis45. This ilicicolin biosynthesis cluster is composed of 10 genes named Ndi_Ili_A-J (Fig.2, Table S5).
Annotated biosynthetic gene cluster of N. discophora SNB-CN63 and related biosynthetic pathways. NR-PKS: nonreducing polyketide synthase, NRPS: nonribosomal peptide synthase. KS: keto-synthase, AT: acyltransferase, PT: product template, ACP: acyl carrier protein, TE: thioesterase. Chemdraw software.
The nonreducing polyketide synthase Ndi_Ili_A is composed of five domains, keto-synthase, acyltransferase, product template, acyl carrier protein, and thioesterase, and it is predicted to catalyze orsellinic acid formation (7) since it shares 56% identity and 70% similarity with AscC, as described in Acremonium egyptiacum45. The prenyltransferase Ndis_Ili_B produces grifolic acid (8) from orsellinic acid by the addition of a farnesyl group (observed at [M+H]+, m/z 373.2373, err. 0.0ppm, C23H32O4), and then a noncanonical nonribosomal peptide synthase Ndi_Ili_C reduces the carboxylic acid function to form ilicicolin B, also named LL-Z 1272 (9) ([M+H]+, m/z 357.2422, err. 0.6ppm, C23H32O3). Finally, the FAD-dependent halogenase Ndis_Ili_D adds a chlorine atom to the orsellinic scaffold to form Compounds 1 and/or 2 ([M+H]+, m/z 391.2028, err. 1.7ppm, C23H31ClO3 and [M+H]+, m/z 407.1979, err. 1.1ppm, C23H31ClO4, respectively), indicating some flexibility of halogenase regarding its substrates (Table S7). It is likely that the formation of other compounds previously described in the literature from this strain45, such as ilicicolinals and ilicicolinic acids, involves the action of monooxygenase, epoxidase and terpene cyclase acting on the prenyl chain. These enzymes are probably located outside the Ndis_Ili gene cluster, as determined by Araki et al. for ascochlorin and ascofuranone45.
The genome of P. sclerotiorum SNB-CN111 (ENA accession number: GCA_911649655) was obtained in 10 contigs covering a total size of 34.7 Mbp and a GC content of 48.3%. The completeness of the genome was established at 98.3% using the BUSCO score, and 12,582 genes were predicted (Tables S3, S4)49. The contiguity of the genome assembly is characterized by an N50 length of 4.34 Mbp and an L50 of four. We used the antiSMASH pipeline to annotate a biosynthetic gene cluster. We predicted 15 clusters with polyketide synthase as core enzymes, with three of them containing two polyketide synthases. Only one of these three clusters with two polyketide synthases included a halogenase31. This putative azaphilone biosynthesis gene cluster is composed of 13 genes named Psc_Aza_A-M (Fig.3, Tables S8, S9). The Psc_Aza_A protein was identified using a Pfam search as a highly reducing polyketide synthase composed of seven modules: -ketoacyl synthase, acyltransferase, dehydratase, methyltransferase, enoylreductase, keto-reductase and acyl carrier protein (phosphopantetheine attachment site). This sequence is typical of highly reducing polyketide synthases such as ATEG_07659 (65% identity, 77% similarity) involved in the biosynthesis of azaphilones such as asperfuranone51. The Psc_Aza_B gene is predicted to encode a nonreducing polyketide synthase composed of five domains: -ketoacyl synthase, acyltransferase, acyl carrier protein (phosphopantetheine attachment site), methyltransferase and a terminal domain. The enzyme CazM, a nonreducing polyketide synthase involved in chaetoviridin synthesis, is the closest homolog (67% identity, 78% similarity) to Psc_Aza_B52. Other azaphilone biosynthetic pathways involving both highly reducing and nonreducing polyketide synthases are described in the literature47,53. A synteny cluster comparison using clinker50 revealed that the polyketide synthase pair involved in other azaphilone biosynthetic pathways (i.e., chaetoviridin, asperfuranone, azanigerone, mitorubrinol and ankaflavin) is conserved and homologous to Psc_Aza_A and Psc_AzaB (Figure S4). These polyketide pairs can operate sequentially, with a highly reducing polyketide synthase producing the first precursor, which is then transferred to nonreducing polyketide synthase and extended (asperfuranone)54. They may also operate in a convergent manner with both enzymes being responsible for the biosynthesis of polyketides that are assembled together at later steps (azanigerone)56 or in a hybrid manner with both sequential and convergent modes (chaetoviridin)55. Finally, Psc_Aza_C, a FAD-dependent monooxygenase, is found in the azaphilone biosynthetic gene cluster47. FAD-dependent monooxygenases play a key role in azaphilone synthesis as they are required for the cyclization of the pyran ring. The high sequence similarity of the Psc_Aza_A/Psc_Aza_B polyketide synthases with the CasF/CazM and ATEG_07659/ATEG_07661 couples suggests that the azaphilone biosynthetic pathway is similar to that of asperfuranone or chaetovirin (Figure S4). Moreover, Psc_Aza_A/B/C/D/E/G/H/L are similar to other proteins involved in other azaphilone biosynthetic pathways (Figure S4), strengthening our annotation identification of Psc_Aza as a putative azaphilone biosynthetic pathway responsible for the production of sclerotiorin.
Annotated biosynthetic gene cluster of P. sclerotiorum SNB-CN111 and related biosynthetic pathways. NR-PKS: nonreducing polyketide synthase, HR-PKS: highly reducing polyketide synthase. KS: keto-synthase, AT: acyltransferase, DH: dehydratase,MT: methyltransferase, ER: enoylreductase, KR: ketoreductase, ACP: acyl carrier protein, TD: terminal domaine Chemdraw software.
Psc_Aza_A catalyzes the elongation of the 4,6-dimethyl-2,4-octadienal unit, and cyclization is then performed by Psc_Aza_B. The monooxygenase Psc_Aza_C then catalyzes the cyclization of the pyran ring and the formation of the azaphilone scaffold (Fig.3). This hypothesis about the initiation of the azaphilone biosynthetic pathway is strengthened by the detection in the molecular network of an ion of m/z 317.1747 (err. 0.1ppm) corresponding to the molecular formula C19H24O4, whose fragmentation spectrum agrees with the structure of metabolite 10 resulting from Psc_Aza_B catalysis (Figure S5). The [M+H]+ ion of compound 11 is observed at m/z 333.1699 (err. 0.8ppm), which corresponds to the molecular formula C19H24O5 expected for the biosynthetic intermediate resulting from the biotransformation of metabolite 10 by Psc_Aza_C. The fragmentation spectrum of this ion at m/z 333.1699 confirms the proposed structure of intermediate compound 11 (Figure S6). Metabolite 12, originating from the spontaneous conversion of metabolite 11, was detected as an [M+H]+ ion at m/z 315.1594 (err. 1.0ppm, C19H22O4), leading to a typical fragmentation spectrum from the sclerotiorin scaffold with a fragment at m/z 147.0457 (err. 7.9ppm) and has not been observed until now for compounds 10 and 11 (Figure S7)20. Molecule 12 is also described as an azaphilone intermediate involved in the asperfuranone biosynthesis pathway56.
Three biosynthetic pathways are possible with this azaphilone scaffold (12). The first pathway leads to the production of sclerotiorin (4). It is mediated by the action of an acyl transferase Psc_Aza_D to form compound 13 ([M+H]+, m/z 357.1705, err. 2.4ppm, C21H24O5). Then, the FAD-dependent halogenase Psc_Aza_H leads to the production of molecule 4 ([M+H]+, m/z 391.1309, err. 0.6ppm, C21H23ClO5). The second biosynthetic pathway leads to the formation of isochromophilone I (17) and is also initiated by the action of an acyltransferase, probably Psc_Aza_D, which adds an acetoacetate to form compound 14 (not detected) that is spontaneously converted by Knoevenagel condensation into compound 15 ([M+H]+, m/z 381.1691, err. 1.5ppm, C23H24O5)42. The angular lactone is then hydrogenated by the action of Psc_Aza_F or G to form compound 16 ([M+H]+, m/z 383.1859, err. 1.6ppm, C23H26O5), which is then chlorinated by the action of Psc_Aza_H. The third biosynthetic pathway leads to the formation of compound 20. The formation of molecule 20 requires the action of an oxidoreductase to reduce C-6 ketone and form a hydroxyl, a dehydrogenase (to hydrogenate the C-1/C-8a bond), an acyltransferase and a halogenase. However, without the detection of intermediates between compounds 12 and 19, it is not possible to define which enzymes are involved and in which order. The enzymes Psc_Aza_D, E, and F/G could be involved in the biosynthesis of metabolites 12 to 19 because of their Pfam domains and their inclusion in the azaphilone biosynthetic gene cluster. Finally, Psc_Aza_H catalyzes the chlorination of molecule 19 ([M+H]+, m/z 361.2021, err. 3.2ppm, C21H28O5) to form compound 20 ([M+H]+, m/z 395.1626, err. 1.6ppm, C21H27ClO5). Notably, compound 21, the chlorinated analog of intermediate 12 (m/z 349.1205, err. 1.1ppm, C19H21ClO4) was also detected (Figure S8), suggesting that the FAD-dependent halogenase Psc_Aza_H may catalyze chlorination as soon as the azaphilone scaffold is formed.
Thus, we completed the annotation of the biosynthetic gene cluster of sclerotiorin (4) and isochromophilone I (17). Both compounds originated from an intermediate with a minimal azaphilone scaffold with a 3,5-dimethyl-1,3-heptadienyl chain (molecule 12) that is typical of sclerotiorin and its analogs (Fig.3). The gene coding for a halogenase, Psc_Aza_H, was identified, as well as numerous chlorinated azaphilone intermediates or analogs (Tables S2, S8). The FAD-dependent halogenase may be involved as early as the formation of the azaphilone scaffold since intermediate 21 is the chlorinated analog of compound 12.
In summary, we identified two clusters of biosynthetic genes that may be responsible for the production of two chlorinated polyketide families: ilicicolins and azaphilones. For each cluster, we annotated a FAD-dependent halogenase. We then applied the OSMAC method to generate structural diversity and to confirm the ability of halogenases in the biosynthetic pathways to introduce various halogens (Cl, Br and I).
Several studies have highlighted the ability of FAD-dependent halogenases to introduce different halogens, such as Cl, Br and I, into various chemical scaffolds57,58,59. Therefore, we sought to generate new compounds taking into account this biosynthetic possibility from our two sequenced strains: N. discophora SNB-CN63 and P. sclerotiorum SNB-CN111. For this purpose, we cultivated strains on PDA media supplemented with 10gL1 NaCl, KBr or KI without affecting microorganism growth. We further analyzed crude extracts by RPLC-ESI(+)-MS/MS to highlight and annotate the major halogenated biosynthesized analogs (Fig.4a).
Generation of halogenated azaphilones and ilicicolins by the OSMAC method. (a) Extracted ion chromatograms of halogenated azaphilones from scaffold A (H, Cl, Br and I) with their isotopic patterns. (b) Identified halogenations from Ndi_Ili_D on ilicicolin scaffolds A and B. (c) Identified halogenations from Psc_Aza_H on azaphilone scaffolds B, C, D and E.
We searched the m/z values of the protonated species corresponding to the halogenated (Cl, Br and I) metabolites in the MS data. We also examined the isotopic profiles and the retention time (RT), which evolves with the sizes of the halogen substituting H, i.e., H We performed a scale-up culture to confirm our structural annotations and to demonstrate that halogenase promiscuity can be exploited to produce sufficient quantities of new and isolable compounds. We used Czapek medium (Czk) to scale up the culture of P. sclerotiorum because no organic nitrogen is provided in this medium, thereby leading to a reduced number of produced metabolites. As brominated molecules from scaffolds B and E have already been described, we focused on scaffold A60,61,62. Furthermore, nitrogenated azaphilone-like molecules bearing scaffold D were not produced in Czk medium, and bromine molecules related to scaffold C were not abundant enough. Therefore, compound 22, which is related to scaffold A and has incorporated bromine, was produced and isolated. Compound22 was obtained as an orange oil and its molecular formula was determined to be C23H23BrO5 based on the ESI-HRMS experiment ([M+H]+ peak at m/z 459.0798 calcd for C23H23BrO5H+, err. 0.7ppm) (Fig.5). The azaphilone scaffold was identified by 13C NMR of carbon with chemical shifts at C of 153.2, 159.7 and 184.5 (C-1, C-3 and C-6, respectively) and validated by HMBC correlations of H1/C-3, C-4a, and C-8a, H4/C-3, C-5, C-8, C-8a and finally H18/C-6, C-7 and C-8. In addition, the HMBC correlations of H-9 and H-10 with C-3 and H-4 with C-3 allowed the side chain to be connected. The lactone moiety was confirmed by the presence of 4 carbons, including 2 carbonyls observed at C 195.1 and 169.4, one methene (at C 124.8), one methyl group (C 30.9) and by HMBC correlations between H-5 and C-3 and C-4. The 3,5-dimethyl-1,3-heptadienyl unit was established by typical correlations of trans-coupled olefinic protons observed in COSY with correlations between H-9/H-10, H-12/H-13/H-16 and H-13/H-14/H-15. As no proton was observed at the C-5 position on the HSQC spectrum, the bromine atom was positioned there (Figures S15-20). This attribution is in accordance with the previously-reported NMR characterization of compounds with the same scaffolds as 15 and 6 (Tables S10, S11)44,60. Compound 22 was described for the first time and was named 5-bromoisorotiorin (Figure S21). Structural elucidation of compound 22 and 2D RMN correlations observed: 1H-1H COSY (in bold) and 1H-13C HMBC (arrows). To date, the isolation and identification of brominated azaphilones has been described in only three publications, and all of them reported the isolation from marine sponge-derived fungi of the Penicillium genus (P. canescens and P. janthinellum)62,63. The authors cultivated these strains with NaBr to obtain these five brominated azaphilones. The new brominated metabolite 22 was also obtained by an OSMAC method using halogenase promiscuity but for the first time from a terrestrial fungus. In a previous study concerning mutualistic strains isolated from termites, we showed that the PDA extract of SNB-CN111 had antifungal activity against Trychophyton rubrum20. Therefore, we compared the antimicrobial activity of previously isolated azaphilones 5-chloroisorotiorin (15) and sclerotiorin (4) with the newly characterized azaphilone 5-bromoisorotiorin (22) on the same human pathogen, i.e., Tricophyton rubrum. We obtained a minimal inhibitory concentration (MIC) value of 32gmL1 for the azaphilone extract and the three individual compounds. Therefore, halogens on azaphilone scaffolds do not seem to modulate the antimicrobial activity of azaphilone, but the promiscuity of Psc_Aza_H halogenase offers the opportunity to generate undescribed natural compounds. Originally posted here:
Combining OSMAC, metabolomic and genomic methods for the production and annotation of halogenated azaphilones and ilicicolins in termite symbiotic...
