The research and collection of biological samples was authorized by the Brazilian government (SISBIO authorization 46555-6, CNPq process 010936/2014-9). Collections were performed in October 2015 and December 2018 in the Atlantic Forest at Itatiaia National Park, RJ, Brazil. GPS coordinates, information of isolated bacteria and Cecropia trees are in Supplementary Table 1. Cecropia species were identified using Gaglioti and Romaniuc-Neto (2012), and Brazil Flora Group (BFG) (2015). Vouchers were deposited at the Herbarium of the Department of Biology, FFCLRP-USP (Herbarium SPFR).

Samples of Cecropia were collected from the internodal space that was associated with Azteca ants. Fungal patches samples, worker ants, pupae, and larvae were also collected. For the isolation of bacteria, parts of the collected samples were placed in Petri dishes containing ISP-2 medium (4 g of yeast extract, 4 g of dextrose, 10 g of malt extract and 20 g of agar, and 1 L of distilled water), supplemented with the antifungals nystatin (40 mg/L) and cycloheximide (50 mg/L). Plates were incubated at 30°C and the appearance of colonies was monitored for four weeks. For the isolation of actinobacteria, fragments of the collected samples were placed on chitin plates (in one liter: 5.33 g of chitin, 0.767 g K₂HPO₄, 0.367 g KH₂PO₄, 0.244 g MgSO₄, 0.01 g FeSO₄⋅7H₂O, 0.001 g ZnSO₄⋅ 7H₂O, 0.001 g MnCl₂ ⋅ 4H₂O, 20 g of agar) supplemented with the antifungals nystatin (40 mg/L) and cycloheximide (50 mg/L) (Hsu and Lockwood, 1975). After four weeks of growth at 30°C, bacterial colonies were subcultured onto ISP-2-agar plates, and serial culturing was done until the obtention of pure cultures.

The bacterial strains were grown in ISP-2 medium and incubated at a controlled temperature of 30°C for 2–3 days. After this the DNA extraction of the bacteria was carried out using the microLYSIS^®-Plus solution according to the manufacturer’s instructions.

Genomic DNA from actinobacteria was extracted using an adapted protocol (Kumar et al., 2010). Strains were grown in Tryptone Soy Broth (BD, United States) at 30°C and 300 rpm for 3–5 days. After this period, the culture was transferred to Eppendorf microtubes and centrifuged at 10,000 × g for 10 min. The supernatant was removed, and the pellet frozen at -20°C. The pellet was washed in 500 μL of sucrose 10.3%, centrifuged for 1 min at 10,000 × g, and the supernatant was discarded. Then 450 μL of TSE (50% Sucrose; 0.5 M EDTA pH 8 and 1 M Tris Buffer pH 8) were added with lysozyme and incubated for 20 min at 37°C. After that period, 13 μL of proteinase K were added and incubated for another 15 min at 55°C. At the end of the incubation time, 250 μL of 2% SDS was added, and then gently stirred until the formation of a clear solution was observed. Then, 300 μL of phenol: chloroform pH 8.0 were added, mixed, and then centrifuged for 10 min at 4°C. The supernatant was transferred to another tube, and 60 μL of 3M NaOAc, pH 6.0, and 700 μL of isopropanol were added. The contents were mixed until the appearance of “white strings”. The tube was centrifuged down briefly and poured off the supernatant. The DNA was resuspended in 100–200 μL of autoclaved deionized H₂O and quantified using NanoDrop^®.

PCR amplification of the 16S rRNA gene of bacteria were performed using two primers: 27F (5′-AGAGTTTGATCMT GGCT-3′) and 1492R (5′-TACGGYTACCTTGTTACGACTT-3′). The final reaction volume of 15 μL contained: 8 μL EconoTac^® DNA Polymerase (Lucigen, United States), 0.5 μL of each primer 27F and 1492R, 0.5 μL DMSO, 4.5 μL deionized H₂O, and 1 μL DNA (10 ng/μL). An initial denaturation step at 94°C for 3 min was followed by 32 cycles of amplification of 94°C for 30 s, 60°C for 30 s, and 72°C for 2 min, and a final extension step of 72°C for 5 min. Amplicons were detected by agarose gel electrophoresis and visualized by ultraviolet (UV) fluorescence after staining with ethidium bromide. The Big Dye sequencing was done using 1.5 μL 5X buffer, 1 μL primer (10 μM), 1 μL BigDye 3.1 (Applied Biosystems), 0.5 μL DMSO, 1 μL PCR product DNA, and deionized water to make up the total volume of 10 μL. The reaction started with 95°C for 3 min, followed by 35 cycles of 96°C for 10 s, 58°C for 3 min and a final extent of 72°C for 7 min. The sequencing reaction was purified with the Axyprep Mag Dyeclean purification kit (Axygen). The DNA was resuspended in 25 μL of deionized H₂O before submission to the Center for Genetics and Biotechnology at the University of Wisconsin (Biotech Center). The 16S rRNA sequences were searched for homologous sequences with BLASTn in the NCBI database. The sequences are deposited at NCBI GenBank under Accession numbers: MW139977 - MW140014.

For a strain of the fungus Pestalotiopsis clavispora FB1 isolated from Cecropia leaves (Supplementary Figure 1) the ITS region (internal transcribed spacer) was amplified using the primers TS1 (5′-TCCGTAGGTGAACCTGCGG-3′) and ITS4 (5′-TCCTCCGCTTATTGATATGC-3′) (White et al., 1990). The PCR consisted of 4.0 μL of dNTPs (1.25 mM each), 2.5 μL of buffer 10X, 1.0 μL of MgCl₂ (50 mM), 0.2 μL de Taq-polymerase (5.0 U/μL), 2,0 μL of each primer (10 μM), 10.3 μL of ultrapure water and 2.0 μL DNA (10 ng/μL). An initial denaturation step at 94°C for 3 min was followed by 35 cycles of amplification of 94°C for 1 min, 55°C for 1 min and 72°C for 2 min, and a final extension step of 72°C for 5 min. Sequencing was performed as described above for bacteria.

Genomic DNA from strain Pseudomonas sp. ICBG1301 was obtained as described previously in the 16S rRNA extraction from actinobacteria with additional steps. After the appearance of “white strings” it was centrifuged down briefly, poured off the supernatant, and redissolved in 500 μL TE. 10 μL of RNase A (20 mg/mL) were added and rocked at room temperature until dissolved. 300 μL phenol were added and mixed. 150 μL of chloroform were also added and mixed. After that, the sample was spun at 4000 x g for 10 min and the top layer was transferred to a new tube. 300 μL chloroform were added and the solution was spun for 3 min at 4000 × g. The top layer was transferred to a tube holding 50 μL 3M NaOAc pH 6 and 350 μL isopropanol, and the tube was inverted until the DNA appeared as a stringy clump. The supernatant was poured off and the DNA was dissolved in deionized H₂O. The genomic DNA was sequenced by Illumina MiSeq^TM System 2 × 300 bp at the Center for Genetics and Biotechnology at the University of Wisconsin (Biotech Center). The assembly method used was SPAdes V3.11.0, the genome coverage was 284.5 X and 29 contigs were obtained. The genome was deposited into Genbank, and its accession number is JAEKGB000000000.

We used the 16S rRNA gene of the four Pseudomonas strains isolated and others present in the NCBI database, that are known to be viscosin-producers, to construct a phylogenetic tree. The sequences were aligned using the MUSCLE alignment algorithm, trimmed manually and the alignments were refined using Geneious Prime 2020.0.4 (https://www.geneious.com). The construction of Maximum-likelihood phylogenetic trees was performed in the IQ-tree 1.6 software using ModelFinder to figure out the best nucleotide replacement model. The TPM3 + F + R2 model was used and branch support was assessed by ultrafast bootstrap UFBoot (1000 bootstrap replicates) (Minh et al., 2020). Phylogenetic tree was visualized with iTol v5 (Letunic and Bork, 2019). The analysis involved 33 nucleotide sequences and the phylogenetic tree was rooted by Pseudomonas aeruginosa PAO1.

Four conserved “housekeeping” genes were used for the multilocus sequence analysis (MLSA): 16S rRNA, gyrB, rpoB, and rpoD. All these genes were extracted from the genome of ICBG1301 and 30 other Pseudomonas spp. deposited in the NCBI database and the phylogenetic tree was rooted by Pseudomonas aeruginosa PAO1.

The sequences belonging to each gene were aligned using MUSCLE, trimmed manually and the alignments were refined using the same algorithm on Geneious Prime 2020.0.4 (https://www.geneious.com). Then the sequences were concatenated. The construction of Maximum-likelihood phylogenetic trees were performed in the IQ-tree 1.6 software using ModelFinder to figure out the best nucleotide replacement model. The TIM + F + I + G4 model was used and branch support was assessed by ultrafast bootstrap UFBoot (1000 bootstrap replicates) (Minh et al., 2020). Phylogenetic tree was visualized with iTol v5 (Letunic and Bork, 2019).

The same method was used for the construction of phylogenetic trees of adenylation and condensation domains. Amino acid sequences of these domains were obtained from the antiSMASH database (Blin et al., 2020). ModelFinder was used in this case to choose the best model for protein evolution.

The ability to fix atmospheric nitrogen by the bacteria isolated from Cecropia-Azteca symbiosis was evaluated by growing the strains on the semisolid nitrogen-free medium (NFb), composed of malic acid 5 g/L, K₂HPO₄ 0.5 g/L, MgSO₄.7H₂O 0.2 g/L, NaCl 0.1 g/L, CaCl₂⋅2H₂O 0.01 g/L, 2 mL micronutrient solution (CuSO₄.5H₂O 0.04 g/L, ZnSO₄.7H₂O 1.2 g/L, H₃BO₃ 1.4 g/L, Na₂MoO₄.2H₂O 1 g/L and MnSO₄.H₂O 1.175 g/L completing the volume to 1 L with distilled H₂O); 2 mL bromothymol blue (solution 0.5% in 0.2N KOH); 4 mL Fe-EDTA (solution 1.64%); 2 mL vitamin solution (0.1 g/L biotin and 0.02 g/L pyridoxine completing the volume to 1 L with distilled H₂O); 4.5 g/L KOH; 2.3 g/L agar; make up to 1 L with distilled H₂O and adjust the pH to 6.8. Glass tubes were filled with 3 mL of the NFb medium, and after solidification, the microorganisms were inoculated in the tube and incubated at 28°C in the dark for 96 hours. Reading was carried out by observing the formation of a growth film close to the surface (Döbereiner et al., 1995; Cattelan et al., 1999; Batista et al., 2018). The strains that showed positive results in this stage were re-inoculated in triplicate to confirm the result.

Pairwise interactions against human pathogens (Staphylococcus aureus and Candida albicans), a phytopathogen (Rhizopus oryzae), an entomopathogen (Metarhizium anisopliae), and Bacillus subtilis were carried out using previously described methods (Getha and Vikineswary, 2002; Visser et al., 2012). A phytopathogen fungus (Lazarotto et al., 2012; Borrero et al., 2018) identified as Pestalotiopsis clavispora FB1, isolated from Cecropia sp. leaves causing damage to the plant, was also tested. After strains inoculation, plates were maintained at 30°C and monitored for 14 days. Then, we recorded the qualitative result, in which we classified it as “inhibition” or “no inhibition” according to the presence or absence of an inhibition zone.

HPLC purifications were carried out using an HPLC system (Shimadzu, Kyoto, Japan), comprising an LC-6AD solvent pump, a CBM-20A system controller, a CTO-20A column oven, a SIL-20AF injector, a SPD-M20A diode array detector (DAD), a FCR-10A collector, and LabSolutions software for data acquisition. Purification of viscosinamide (1) was performed at 3 mL/min with a reversed phase C18 semi-preparative column (Phenomenex Gemini, 10 mm, 250 mm, 5 μm) using a gradient solvent system with aqueous acetonitrile (50% to 100% acetonitrile) for 12 min. The absolute configuration of the amino acids present in the structure was determined using advanced Marfey’s method (Harada et al., 1996). Viscosinamide (1) (1 mg) was hydrolyzed at 110°C in 500 μL of 6 N HCl for 24 h. HCl was removed under reduced pressure and the dry material was resuspended in 500 μL of H₂O and dried three times to remove residual acid. The hydrolyzate was dissolved in 100 μL of 1 N NaHCO₃ and incubated with 50 μL of 1 mg/mL L-FDLA and L,D-FDLA (FDLA - 2,4-dinitro-5-fluoro-phenyl leucineamide), in acetone, at 80°C for 3 min. The reaction was quenched by adding 50 μL of 2 N HCl. Aqueous acetonitrile (1:1) was added to dissolve the mixture. A 10 μL aliquot of the hydrolyzate derivative was analyzed by HPLC-HR-ESI-MS using an analytical C18 column (Phenomenex Gemini, 4.6 mm, 250 mm, 5 μm), with a gradient solvent system (5% to 100% ACN with 0.1% formic acid over 30 min) and flow rate of 0.7 mL/min. The oven temperature was 35°C. The ionization source parameters were: End Plate Offset voltage, 500V; capillary voltage, 3500 V; nebulizer at 5.5 bar; drying gas flow rate of 10.0 L/min; drying gas temperature, 220°C; for positive mode ionization. The spectra rate of 2.0 Hz and mass range of 300 to 1000 m/z were maintained throughout the analysis. The Data Analysis^® Software of Bruker Daltonics instrument was used to analyze the metabolic profile and UV spectrum data.

NMR spectra were recorded using a BRUKER^® - DRX500 - Ultra Shield^® (¹H: 500.13 MHz, ¹³C: 125.77 MHz). Spectra were processed using MestReNova 6.0.2. High-resolution MS spectra were obtained using a micrOTOF II-ESI-TOF (Bruker Daltonics^®), positive ion mode using capillary voltage 3900 V, dry gas flow 4 L h^–1, and nitrogen as nebulizer gas. Na-TFA 10 mg/mL was used for internal calibration. Samples were prepared at 20 μg/mL, in EtOAc:MeOH:H₂O (1:2:2). The LC-MS was carried out on a Bruker Daltonics High Pressure Liquid Chromatographer coupled to a diode array UV detector, and a Mass Spectrometer with electrospray ionization source (ESI) and micrOTOF (Time of Flight) detector. MS/MS analysis was performed using the MALDI-TOF/TOF (Bruker Daltonics, UltrafleXtreme, Bremen, Germany). The matrix CHCA (α-cyano-4-hydroxycinnamic acid) was prepared using a saturated solution in aqueous acetonitrile 3:7 (0.1% TFA). The settings of the instrument were: positive ion reflector mode (RV1 e RV2 of 26.60 kV e 13.35 kV, respectively); 500 laser shots per spectrum; PIE (Pulsed ion extraction) of 110 ns, laser frequency of 1000 Hz; IS1 (ion source 1) and IS2 voltages were 25.00 kV e 23.00 kV, respectively; external calibration was carried out using a peptide mixture provided by Bruker. Samples were diluted in methanol and 1 μL of the sample was added to 1 μL of matrix and 1 μL of this mixture was applied to the MALDI plate.

Bacterial strains were isolated from Cecropia samples inhabited by Azteca ants collected at Itatiaia National Park (RJ, Brazil) in 2015 and 2018. Microbial isolations were performed from various parts of the samples gathered, ranging from ants, eggs, plant parenchyma, and fungal patches found in the hollows of Cecropia’s stalks. This process resulted in the isolation of 39 bacterial strains which had their 16S rRNA gene sequenced, allowing their identification at the genus level (Supplementary Table S1). Among the isolates, 25 (64.1%) strains belong to the phylum Proteobacteria, 12 (30.8%) to Actinobacteria, and two (5.1%) to Firmicutes. Six bacterial orders were represented including: Actinomycetales (12), Rhizobiales (10), Pseudomonadales (7), Enterobacterales (5), Xanthomonadales (3), and Bacillales (2) (Figure 1). The isolates belonged to ten different genera namely Acinetobacter, Actinomadura, Bacillus, Methylobacterium, Pantoea, Pseudomonas, Rhizobium, Raoultella, Stenotrophomonas, and Streptomyces.

Isolates were obtained mainly from Azteca ants and fungal patches. Pseudomonas spp. and Rhizobium spp. were found in three of the four sampled sites (Cecropia parenchyma, Azteca sp. ant and fungal patches); Methylobacterium spp. also in three (Azteca sp. ant, Azteca sp. egg and Cecropia parenchyma); Pantoea spp. was found in two (Azteca sp. ant and Azteca sp. egg); Bacillus spp. in two (Cecropia parenchyma, fungal patches); Stenotrophomonas spp., Streptomyces spp. and Actinomadura spp. were sampled also in two (Azteca sp. ant and fungal patches); Acinetobacter spp. and Raoultella sp. in just one site (Azteca sp. ant and fungal patches, respectively).

Proteobacteria genera Pantoea, Rhizobium, Methylobacterium, and Pseudomonas have known diazotrophic potential (Borm et al., 2008; Pinto-Tomás et al., 2009; Ruiz-González et al., 2019; Bar-Shmuel et al., 2020). Due to the significant number of strains isolated from Cecropia-Azteca that could potentially fix nitrogen (22; 56.4%), the nitrogenase activity of these strains was tested. All bacteria grew in the nitrogen-free semi-solid NFb media under the tested culture conditions. However, only those belonging to the genus Pseudomonas (strains ICBG1870, ICBG1871, ICBG1301, and ICBG1881) showed potential for nitrogen fixation as they formed a white disk near the surface after two re-inoculation rounds. Although some other strains did exhibit growth on nitrogen-free medium they did not show potential to perform biological nitrogen fixation (BNF) as indicated by the re-inoculation step, which is important for confirmation of the result (Cattelan et al., 1999), indicating that their growth may have occurred at the expense of energetic reserves of bacterial cells.

Bioassays of the 39 isolated strains were performed against human pathogens, phytopathogens and entomopathogen (Supplementary Figure 2). Sixteen isolates (41%) inhibited at least one of the microorganisms tested, and eight strains (20.5%) inhibited at least two. Bacillus spp. strains inhibited the growth of at least three pathogens, including the human pathogen C. albicans. The isolates with biological activity against phytopathogens and the entomopathogen were of particular interest since this inhibition could denote ecological relevance. Five out of the 16 strains referred above were active against P. clavispora, ten against R. oryzae, and seven against M. anisopliae. Interestingly, four strains identified as Pseudomonas spp. inhibited the growth of almost all microorganisms tested, including the phytopathogen P. clavispora FB1 (Supplementary Figure 3).

According to the biological screening, Pseudomonas spp. were the most bioactive bacterial isolates. We selected one strain (ICBG1301) to sequence its genome and analyze the respective biosynthetic gene clusters. Genome analysis of Pseudomonas sp. ICBG1301 with antiSMASH 5.0 (Blin et al., 2019) revealed 14 biosynthetic gene clusters (BGCs), encoding the production of siderophore, non-ribosomal peptides, N-acetylglutaminylglutamine amide, beta-lactone, thiopeptide, arylpolyene, and bacteriocins. We therefore decided to investigate BGCs encoding for the production of antifungal compounds. Cluster 9.1 showed 43% similarity to the antifungal non-ribosomal peptide (NRP) viscosin according to the antiSMASH output. NRPs are biosynthesized by specialized enzymes called Non-Ribosomal Peptide Synthetases (NRPS). These NRPS are megaenzymes, which have several catalytic domains, each of which is responsible for a stage of the biosynthesis of the final product. Essential domains are involved in the formation of the mature peptide, called PCP (peptide carrier protein), A (adenylation), C (condensation) and TE (thioesterase termination, usually found only once in a gene cluster) (Finking and Marahiel, 2004). Bioinformatic analyses on these domains can give us insights into the specificities of the monomers recruited and their stereochemistry. In order to evaluate whether the BGC could be responsible for the production of viscosin or not, analyses were performed with its condensation and adenylation domains.

Phylogenetic analyses of the adenylation domains of cluster 9.1 and other 20 viscosin-producing strains were done and the final prediction of the peptide was obtained based on the substrate specificity of nine modules, being: Leu1-Gln2/Glu2-Thr3-Val4-Leu5-Ser6-Leu-7-Ser8-Ile9. These analyses showed that the adenylation domains responsible for recruiting the same monomer, in the case of modules 6/8 (Ser) and 1/7 (Leu), are subdivided according to the groups observed in Figure 2 (Supplementary Figure 8). The compound encoded by cluster 9.1 belongs to a group – the viscosin group – of cyclic lipodepsipeptides. Members of this group comprise viscosin, viscosinamide, WLIP (White Line-Inducing Principle), pseudodesmin and massetolide (Supplementary Figure 16). These compounds differ from each other in the absolute configuration of one amino acid (Leu5) or in the amino acid composition of the third and/or fourth positions. Adenylation domains were grouped according to the amino acid incorporated, and each one formed two clades. The domains of Pseudomonas strains that produce pseudodesmin or WLIP were grouped together, and the domains involved in the biosynthesis of viscosin and massetolide formed another. A specific analysis of Pseudomonas sp. ICBG1301 showed that the A domains are grouped with those of P. fluorescens SBW25, P. antarctica PAMC 27494 and Pseudomonas sp. LBUM990. P. fluorescens SBW25 produces viscosin, as confirmed by NMR, raising the hypothesis that the ICBG1301 strain also produces viscosin. It could be possible to bioinformatically differentiate viscosin, viscosinamide, WLIP, pseudodesmin and massetolide producers by analyzing the specificity of the adenylation domain of the second module. Strains that show such specificity for glutamic acid could be classified as producers of viscosin, WLIP and massetolide, whereas those that show specificity for glutamine could be producers of viscosinamide and pseudodesmin. However, the structural difference between these two amino acids is not so significant and the adenylation domains are grouped together.

Furthermore, phylogenetic analysis of the condensation domains revealed that, except for modules 1 and 8 (C_starter and ^LC_L, respectively), all modules contain a C_dual domain (Supplementary Figures 5–7). C_dual domains are responsible for catalyzing the condensation reaction and epimerization of the upstream amino acid. Intriguingly, domains belonging to the sixth module were split into two different groups (Supplementary Figure 7). Except for one strain (Pseudomonas sp. LBUM920), strains that belong to group 2 have a Histidine → Tyrosine mutation in the N-terminal catalytic triad (Figure 2). According to the literature, all viscosin group members show an L-amino acid in the first position. Thus, during the evolution of this pathway, a mutation in the C-domain of the second module has possibly occurred, and this domain has lost its function. Similarly, another mutation might have occurred in the region coding for the sixth module C-domain. This mutation could be responsible for the loss of function of their respective C_dual domains. Consequently, it is possible to infer that group 1 strains have the functional domain; thus, they may produce pseudodesmin, showing a D-Leu5, while group 2 strains produce viscosin, showing a L-Leu5. In general, the gene clusters related to viscosin, viscosinamide and massetolide biosynthesis are split into two loci, while the pseudodesmin and WLIP clusters are in contiguous clusters. This observation is consistent with our analyses, since group 1 strains show a contiguous gene cluster and group 2 strains show clusters split into two loci. BGC 9.1 of Pseudomonas sp. ICBG1301 is split into two loci, therefore, potentially related to the production of viscosin or viscosinamide.

After analyzing the gene clusters involved with viscosin biosynthesis, we decided to investigate the evolutionary relationship between Pseudomonas strains. We built a Maximum likelihood tree using the 16S rRNA (Supplementary Figure 4) and a multilocus phylogeny based on the 16S rRNA, gyrB, rpoB, and rpoD gene sequences (Figure 3).

It is noteworthy that the strains that showed the mutation in the C6-domain also have the cluster split into two loci and share a common ancestor, forming a monophyletic group. Viscosinamide (P. synxantha 2-79, Pseudomonas sp. J380, Pseudomonas sp. MYb193, Pseudomonas sp. A2W4.9 and Pseudomonas sp. U2W1.5) and massetolide (P. synxantha 30B and P. lactis SS101) producers formed a group, apart from the viscosin clade. Viscosinamide production appears either as an ancestor in the group or appearing three times in the evolution of the group. Massetolide appears twice in the evolution of the group and viscosin only once. The most reasonable hypothesis in this clade is the ancestral production of viscosinamide, with later change to massetolide (two evolutionary events) and viscosin (one evolutionary event). Strain ICBG1301 is related to viscosin-producing strains (P. fluorescens SBW25, P. antarctica PAMC 27494 and Pseudomonas sp. LBUM990). This raises the hypothesis that ICBG1301 could either produce viscosin or viscosinamide.

Cultures of Pseudomonas spp. ICBG1870 and ICBG1881, isolated from different ant colonies, were scaled up and extracted with ethyl acetate. Rounds of bioguided fractionation led to the isolation of the active compound, with an m/z 1147.6976 [M + Na]⁺ (error 2.4 ppm) (Supplementary Figure 9). Both samples showed identical ¹H NMR profiles, corresponding to the same compound. NMR and MALDI-MS/MS data confirmed the structure of the CLP (Supplementary Table S2, Supplementary Figures 10–14). Signals in the region of δ_H 3.44–4.57 were observed in the ¹H NMR spectrum, which correspond to the α-protons of the amino acids. These α-protons are bonded to carbons at δ_C 53.4–65.8, as deduced from gHSQC spectrum. The gCOSY spectrum showed α-protons are coupled with the side chain hydrogens (δ_H 1.87, 2.02, 5.36, 2.19, 1.77, 3.79, 1.60, 3.65, and 1.98). The α-protons also correlate with the carbonyl carbons of the adjacent amino acid (δ_C 170.1–176.6), as seen in the gHMBC spectrum. Hydrogens at δ_H 4.04 and 5.11 have no correlations with carbons in gHSQC and are directly linked to oxygens in the side chains of Ser6 and Ser8. Other hydrogen signals that do not show correlations in the gHSQC spectrum are those linked to the nitrogens of the peptide bonds (δ_H 9.23, 8.74, 8.47, 8.25, 8.02, 7.95, 7.47, 7.08, and 6.61). Resonances of CH₂ hydrogens are observed in the most shielded region of the spectrum signals (δ_H 0.7 - 2.6) and correspond to the alkyl chain of the fatty acid.

Hydrolysis and advanced Marfey’s derivatization, followed by LC-MS, confirmed the bioinformatic predictions of the stereocenters. The chromatogram shows the presence of D-Glu instead of D-Gln, causing a divergence of one mass unit with the HRESI-MS analysis, due to the conversion of Gln to Glu during the hydrolysis process (Shih, 1985; Li et al., 2010). In addition, the chromatogram showed the presence of only L-Leu in the structure (Supplementary Figure 15), corresponding to the antifungal viscosinamide (1) (Figure 4; Nielsen et al., 1999). This compound was also detected by LC-MS in the extract of Pseudomonas sp. ICBG1301. This result agrees with the antifungal activity of the producer strains against M. anisopliae, P. clavispora, and R. oryzae.