The marine water sample was collected from four divergent spots in the sea area of the Gulf of Khambhat, India (22.1775N 72.4763 E) (Supplementary Figure S1). Sea water was collected 10 m apart from each spot and at 1 m depth in a germ-free container and transported to the laboratory under cool conditions for further isolation and purification of fungi. Fungi from collected samples were isolated using a marine agar medium (Bonugli-Santos et al., 2015) by adopting serial dilution followed by incubation for 48–96 h at 25°C. Single and pure fungal colonies were picked up and further allowed for growth on marine agar plates. A pure fungal strain, KHW-7, was identified by morphological and microscopic observation as well as by sequencing of amplified ITS region of the fungal gene.

The workflow for the WGS experiments is shown in Figure 1. Genomic DNA from the KHW-7 strain was extracted using the silica spin column DNA extraction method following the manufacturer’s manual. Subsequently, 08% agarose gel electrophoresis was performed to check the quality of isolated genomic DNA. The presence of a single intact band within the gel matrix is indicative of the superior quality of isolated genomic DNA. Additionally, a 2-μl fraction of the genomic DNA sample was subjected to spectrophotometric analysis using the BioTeK Epoch spectrophotometer to determine the A260/280 ratio. This ratio serves as a pivotal measure of DNA purity.

The libraries were prepared using the commercially available Ion Xpress^™ Plus Fragment Library Kit (Thermo Fisher Scientific, United States) as per the manufacturer’s instructions. This process involved stages such as DNA fragmentation, fragment purification, ligation of fragments, fragment amplification, and final quantification. The commercially available Ion Library TaqMan Quantitation kit was used for quantification purposes. The size of the fragmented DNA was assessed (Quality Control Step) using the Agilent^™ High Sensitivity DNA Kit on the Agilent^™ 2,100 Bioanalyzer, following the provided instructions. After library preparation, the template was prepared using the Ion Chef automated system according to the manufacturer’s instructions available with the Ion 550 Kit (Thermo Scientific, United States), and the Ion 550 Chip Kit was utilized for loading samples with the assistance of the Ion Chef, followed by sequencing on the Ion GeneStudio S5 Plus System (Ion Torrent, Thermo Scientific, United States).

The Ion Torrent single-end sequencing reads were subjected to adapter and quality trimming using cutadapt (v4.7) and Trim Galore (v0.4.1) with phred score cutoff of 20. The obtained superior quality reads were built from scratch using SPAdes v3.13.0. The initial analysis, such as base pair calling and trimming of sequences, was performed using the Ion Torrent browser. This process resulted in obtaining readings of good quality. The sequence readings were assembled de novo using the SPAdes assembler v3.1.0 (Torrent browser) with the default parameters. The scaffolds obtained were filtered based on their respective length, with a minimum threshold of 500 base pairs. Assembly statistics were generated by QUAST (v5.2.0).

A repeat library was created from scratch for the chosen assembly of the KHW-7 strain using RepeatModeler v2.0.4. This library was then utilized as a customized library for softmasking with RepeatMasker v4.1.5. The Funannotate v1.8.16 pipeline was used to structurally annotate the masked assemblies. The BUSCO database v5.7.1 was utilized to identify conserved gene models for the purpose of training the Ab initio gene predictors such as Augustus, glimmerhmm, and snap. The generation of gene models was based on evidence, achieved by matching the sequences of contigs with the unified protein sequence database (UniProtKB; https://www.uniprot.org/) using the DIAMOND program. Subsequently, the gene models were refined using Exonerate. The Funannotate process utilized the EVidenceModeler,¹ which included a weighting method, to choose the consensus models from a pool of ab initio and evidence-based gene models. Functional annotation of the consensus models was conducted following the elimination of models with insufficient lengths, gaps, and transposable elements (TEs).

The gene models were functionally predicted using InterProScan (v-5.67-99.0), which involved mapping to the Gene Ontology (GO) database² and eggNOG-mapper (v4.5.1) based on the eggNOGorthology database.³ The significantly enriched GO terms were further analyzed to find out the interactions among several biosynthetic pathways using the “ClueGO”⁴ plugin of the CytoScape software (v3.7.2.0). Signal peptides (secretome) were predicted using SignalP (v6.0) and Phobius.⁵ Biosynthetic gene clusters (BGCs) were identified in the genome using fungiSMASH,⁶ which is a specialized version of antiSMASH designed for fungal genomes. The tRNA and rRNA genes were detected using tRNAscan-SE⁷ and Barrnap,⁸ respectively.

Functional annotation using the annotate pipeline⁹ used to annotate genes by performing similarity search against databases of UniProt, Pfam, dbCAN (CAZyme), MEROPS, and BUSCO pezizomycotina gene models. Output of InterPro, EggNog, SignalP, Phobius, tRNAscan-SE, and antiSMASH was added to the final comprehensive annotation files, which can be directly submitted to the National Center for Biotechnology Information (NCBI). Further annotation was conducted using the NCBI non-redundant (NR) genome database,¹⁰ Pathogen Host Interactions (PHIs),¹¹ and the Comprehensive Antibiotic Resistance Database (RGI-CARD).¹²

The fungal isolate, KHW-7, was identified using ITS gene sequencing. This partial ITS gene sequence showed 100% homology with reference strains of C. verruculosa. Barrnap was used to predict rRNA genes within the assembled genome. Partial ITS gene sequence was searched in the predicted 9 rRNA sequences of KHW-7 using local NCBI blastn application, and Blast output produced 100% alignment against contig574 spanning 5.8S region as well as between start and end of 18S and 28S rRNA regions, respectively. Other phylogenetic marker genes, glyceraldehyde 3-phosphate dehydrogenase (gdph), found in the annotated protein sequences, show 100% similarity with other partial gdph genes of C. verruculosa.

The taxonomy of this genome is:

Cellular organisms >Eukaryota >Opisthokonta >Fungi >Dikarya >Ascomycota >saccharomyceta >Pezizomycotina >leotiomyceta >dothideomyceta >Dothideomycetes >Pleosporomycetidae >Pleosporales > Pleosporineae >Pleosporaceae > Curvularia >Curvularia verruculosa.

For genome assembly, the SPAdes assembler, version 3.13.0, was used to process and utilize a total of 6,110,868 high-quality reads. The 31.59 Mb genome of C. verruculosa KHW-7 featured a guanine–cytosine (GC) content of 50.44%. The screened reads were ordered into 1,323 contigs (≥500 bp), making 31,589,880 bp large genome with 81,050 N50 value. Moreover, the integrity was 97%, indicating excellent quality of genome assembly. The circos plot of the annotated genome of C. verruculosa KHW-7 is depicted in Figure 2. Another reported Curvularia spp. was also reported to have a genome size of 33–36 Mb and an average 50% G + C content (Quach et al., 2022). Comparative genome features of reported Curvularia spp. are depicted in Table 1. At the species level, C. verruculosa KHW-7 is the first sequenced whole genome. The presence of repetitive elements, such as interspersed repeats and low-complexity DNA sequences, in the genome assembly is significant due to their recognized involvement in genome length expansion and evolution (Sun et al., 2012). Annotation of the draft genome using funannotate pipeline and other RNA prediction tools predicted a total of 9,877 genes, of which 9,745 mRNA (CDSs), 123 tRNA, and 9 rRNA were predicted. As per the funannotate pipeline, complete CDSs were 9,489, and partial CDSs were 256. Predicted multiple exon transcripts were 7,541 and single exon transcripts were 2,204, with an average protein length of 502.68. TEs are mobile genetic elements that have a role in the occurrence of mutations, regulation of gene expression, and rearrangement of chromosomes, enabling populations to adapt efficiently to environmental changes (Lorrain et al., 2021). A total of 9,877 genes were functionally annotated by performing sequence similarity searches against the Pfam, InterPro, BUSCO, EggNOG, MEROPS, and CAZyme databases and utilizing the SignalPsecretome prediction program. These searches resulted in a total of 25,199 annotations. Further, the KEGG annotation predicts a total of 3,724 genes, the COG annotation predicts a total of 6,979 genes, and the GO terms annotation predicts a total of 6,318 genes. (Table 2). Supplementary Table S1 shows the comparison of genome features between C. verruculosa KHW-7 and other Curvularia species.

EggNOG-mapper annotated 75% of the predicted proteins in the genome, assigning them to 6,979 eggNOGorthogroups, which represent over 24 functional categories. The most abundant functional categories were S, G, O, U, and E. These results suggest that C. verruculosa KHW-7 is well-adapted to carbohydrate metabolism, protein turnover, and intracellular trafficking. The annotation comprised 991 putative proteins and 6,979 proteins with confirmed functions. The proteins were categorized as follows: 1,771 proteins were assigned Enzyme Commission (EC) numbers, 3,374 proteins were assigned GO assignments, and 3,803 proteins were linked to the KEGG pathways. EggNOG annotated 6,964 proteins using Pfam, a curated protein domain family (Figure 3). While EggNOG provides a reliable and precise genome annotation, the approach and execution of the annotation differ conceptually from that of BlastKoala (KEGG). Thus, the functional annotation was conducted utilizing the anticipated protein sequences of the genome. BlastKoala annotated 3,724 entries (38.2%) from a total of 9,868 entries (protein sequences), and 403 pathways were classified into 22 functional categories as per Figure 4 and Table 3.

InterProScan consolidates protein signatures from many databases into a unified and searchable resource, leveraging their unique capabilities to create a robust integrated database and diagnostic tool for classifying protein sequences. InterProScan classifies proteins into families and identifies important domains and sites, which is invaluable for identifying distantly related proteins and predicting their functions. Interproscan has annotated a total of 9,489 genes out of a total of 9,745 genes predicted, of which 6,318 are with GO annotation. A total of 3,056 genes were assigned the Enzyme Commission (EC) codes (Supplementary Figure S2; Table 4). A more precise identification of the interaction among different biosynthetic pathways was performed by the CytoScape network analysis of the various significantly enriched GO terms (biological process, molecular function, and cellular component), as shown in Figures 5, 6 and Supplementary Figures S3, S4.

The antiSMASH, specifically the fungal version (fungiSMASH), analysis of C. verruculosa genome revealed 25 BGCs for secondary metabolites, of which 9 regions show sequence similarity from 13 to 100%. C. verruculosa is found to be rich in T1PKS (Polyketides type 1) (6 hits, 13 to 100%) and more so than with other secondary metabolites’ signals, such as non-ribosomal peptide synthetase (NRPS) (2 hits, 46 and 50%) and terpene (40%) (Supplementary Figure S5). The gene located at position 73.1 exhibited a significant similarity with the choline biosynthesis gene cluster (GenBank: CH236925.1) from Aspergillus nidulans FGSC A4. Prior reports indicate that administering choline and alpha-lipoic acid to Balb/c mice resulted in a significant reduction in the levels of isoprostanes and reactive oxygen species (ROS) produced in bronchoalveolar lavage (BAL) fluid. This, in turn, effectively regulated oxidative stress. The administration of either choline or alpha-lipoic acid resulted in a decrease in lipid peroxidation levels and NFkappaB activity, as demonstrated by Mehta et al. (2009). Therefore, these compounds can be regarded as major antioxidants. In fungi, metabolites play a crucial role in the proliferation of filamentous fungi (Markham et al., 1993). The genome BGC region 88.1 of C. verruculosa exhibited significant resemblances to the Glarea lozoyensis 1,3,6,8-tetrahydroxynaphthalene BGC (GenBank: AF549411.1). The analysis suggests that the T1PKS gene cluster present in the genome of C. verruculosa could be accountable for the synthesis of 1,3,6,8-tetrahydroxynaphthalene (T4HN). The study conducted by Mosunova et al. (2022) showed that melanin-forming fungus actively synthesizes T4HN using the acetogenic pathway. T4HN was identified as a result of the pentaketide synthase PKS1 in the black fungus Colletotrichum lagenarium. Several instances of BGCs have been documented in Aspergillus section Nigri, which is mostly linked to the production of bioactive secondary metabolites (Wang et al., 2023). Region 393.1 was discovered to bear a strong resemblance to the peramine BGC observed in Epichloe festucae (GenBank: AB205145.1). Epichloë synthesizes peramine, a compound that exhibits antibacterial, fungicidal, and insecticidal properties. This measure protects crops post-harvest by effectively countering phytopathogenic organisms (Song et al., 2021).

The fungal genome possesses CAZymes gene families, which are widely responsible for many biological events, including the degradation of lignocellulose materials (Garron and Henrissat, 2019). Functional annotation of the genes of C. verruculosa was conducted using the CAZy database. CAZy is a specialized database for data annotation that focuses on carbohydrate enzymes (Brandi et al., 2009). There were 509 genes identified as CAZymes, and they were classified into six different kinds in the database. The genes were ranked in descending order based on their abundance, with glycoside hydrolases (GHs) having the highest count of 240, followed by auxiliary activities (AAs) with 133, glycosyl transfers (GTs) with 71, carbohydrate esterases (CEs) with 47, carbohydrate-binding modules (CBMs) with 20, and polysaccharide lyases (PLs) with 19 (Figure 7). Moreover, 47.15% of total genes of CAZymes were occupied by GHs, followed by auxiliary active enzymes (AAs) (26.12%), establishing these fungias potent strain for the breakdown of biomaterials mainly composed of lignin, cellulose, and hemicelluloses (Chylenski et al., 2019). The presence of CEs, CBMs, and PLs also confirms various plant biomass degradation capacities of C. verruculosa (Geng et al., 2021). The comparative analysis of CAZymes from all studied genomes of Curvularia spp. showed similar patterns of CAZymes found in C. verruculosa KHW-7 (Quach et al., 2022). Other than GHs and AAs, the genome of C. verruculosa KHW-7 also possesses 71 GTs family members largely involved in the biosynthesis of various polysaccharides and oligosaccharides (Lairson et al., 2008).

The MEROPS database discovered a total of 326 proteases, which may be classified into different groups, including aspartic (A), cysteine (C), metallo (M), serine (S), mixed (P), and threonine (T). Additionally, the database also includes a class for protease inhibitors (I), as depicted in Figure 8. The two highest ranking families among the transcription factor (TF) families were the fungal Zn(2)-Cys(6) binuclear cluster domain (IPR001138) and the fungal-specific TF domain (IPR007219) (Figure 9).

The Comprehensive Antibiotic Resistance Database (CARD) is designed as an antibiotic resistance ontology (ARO) that links antibiotic modules with their targets, resistance mechanisms, gene variants, and other relevant information. The Resistance Gene Identifier has given no hits against the CARD database. Hence, no drug resistance genes are present in the C. verruculosa KHW-7 genome.

Genes associated with pathogenicity factors were examined and identified utilizing the Pathogen Host Interaction (PHI) database. Sequences from the PHI database were downloaded from the Virulence Factor Database (VFDB). The total number of protein sequences in the full database was 8,216. A BLAST homology search between the DNA sequences of C. verruculosa strain KHW-7 and PHI database proteins revealed a total of 136 hits (Supplementary Table S2).

The analysis has revealed that critical pathogenic genes exhibit a variety of interacting behaviors, including an increase in virulence (4 genes), lethal (5 genes), unaffected pathogenicity (35 genes), and key non-pathogenic/low virulence genes, including the loss of pathogenicity (17 genes) and reduced virulence (69 genes), are detailed in Supplementary Table S2.