In silico prediction of polyketide biosynthetic gene clusters in the genomes of Hypericum-borne endophytic fungi

Background The search for new bioactive natural compounds with anticancer activity is still of great importance. Even though their potential for diagnostics and treatment of cancer has already been proved, the availability is still limited. Hypericin, a naphthodianthrone isolated essentially from plant source Hypericum perforatum L. along with other related anthraquinones and bisanthraquinones belongs to this group of compounds. Although it has been proven that hypericin is synthesized by the polyketide pathway in plants, none of the candidate genes coding for key enzymes has been experimentally validated yet. Despite the rare occurrence of anthraquinones in plants, their presence in microorganisms, including endophytic fungi, is quite common. Unlike plants, several biosynthetic genes grouped into clusters (BGCs) in fungal endophytes have already been characterized. Results The aim of this work was to predict, identify and characterize the anthraquinone BGCs in de novo assembled and functionally annotated genomes of selected endophytic fungal isolates (Fusarium oxysporum, Plectosphaerella cucumerina, Scedosporium apiospermum, Diaporthe eres, Canariomyces subthermophilus) obtained from different tissues of Hypericum spp. The number of predicted type I polyketide synthase (PKS) BGCs in the studied genomes varied. The non-reducing type I PKS lacking thioesterase domain and adjacent discrete gene encoding protein with product release function were identified only in the genomes of C. subthermophilus and D. eres. A candidate bisanthraquinone BGC was predicted in C. subthermophilus genome and comprised genes coding the enzymes that catalyze formation of the basic anthraquinone skeleton (PKS, metallo-beta-lactamase, decarboxylase, anthrone oxygenase), putative dimerization enzyme (cytochrome P450 monooxygenase), other tailoring enzymes (oxidoreductase, dehydrogenase/reductase), and non-catalytic proteins (fungal transcription factor, transporter protein). Conclusions The results provide an insight into genetic background of anthraquinone biosynthesis in Hypericum-borne endophytes. The predicted bisanthraquinone gene cluster represents a basis for functional validation of the candidate biosynthetic genes in a simple eukaryotic system as a prospective biotechnological alternative for production of hypericin and related bioactive anthraquinones. Supplementary Information The online version contains supplementary material available at 10.1186/s12864-024-10475-z.

SM is a result of a synergic crosstalk between the plant and endophytic microorganisms inhabiting the host tissues.Endophytic fungi reside inside the plant tissues without causing a disease to their host plant.As many endophytes are able to produce in vitro the same compounds as their hosts, these fungi are assumed as sustainable cell-factories for heterologous production of diverse bioactive metabolites [1].
SMs that are widespread in microorganisms and rarely present in plants include anthraquinones.Natural anthraquinones are bioactive compounds synthesized predominantly via polyketide pathway.In higher plants, there are two main proposed routes leading to anthraquinones: the polyketide pathway by type III polyketide synthase and the shikimate or chorismate/O-succinyl benzoic acid pathway (reviewed by Mund and Čellárová [2]).The representatives of the genus Hypericum belong to few plant taxons that are important producers of (bis) anthraquinones, and are the only plant producers of hypericin and its derivatives.Hypericin and skyrin are promising agents in the anticancer therapy [3,4].Despite the biosynthesis of hypericin has not been fully elucidated, emodin and/or emodin anthrone are considered as possible precursors.Based on the results of LC-MS analyses, Kimáková et al. [5] proposed hypericin biosynthetic route via bisanthraquinone skyrin.Later Rizzo et al. [6] identified atrochrysone as a putative precursor and skyrin as a side product of hypericin biosynthesis.
Up to date, dozens of endophytes were isolated from Hypericum plants of different provenance [7][8][9].Most of the identified isolates are filamentous ascomycetes (e.g.Fusarium, Diaporthe, Alternaria, Septoria), less frequent basidiomycetes (Schizophyllum, Phlebia), and mucoromycetes (Mortierella) (Fig. S1).According to the metabolomic analyses, axenic cultures of several isolates were able to produce small quantities of the host plant-derived anthraquinones, especially hypericin, pseudohypericin, emodin and skyrin during the first subcultures.However, the genetic and epigenetic background of anthraquinone biosynthesis in Hypericum-borne endophytic fungi has not been studied yet.The knowledge on pathway and regulation of anthraquinone biosynthesis in fungal endophytes may significantly contribute to the understanding of hypericin biosynthesis in Hypericum spp.[2].
Fungal anthraquinone biosynthetic gene clusters (BGCs) usually consist of genes encoding enzymes catalyzing biosynthetic steps leading to basic anthraquinone skeleton, other enzymes tailoring the final product, transport and regulation genes.High-throughput genome sequencing and bioinformatics are powerful approaches that markedly help to discover novel putative cryptic BGCs, prior to further metabolomic analyses and functional validation of candidate biosynthetic genes.Moreover, most of the anthraquinone BGCs become transcriptionally inactive after several subcultures and whole-genome screening may contribute to decipher the hidden biosynthetic potential.
The objectives of the present study are: 1) In silico prediction and functional annotation of polyketide biosynthetic gene clusters in de novo assembled and annotated genomes of selected fungal endophytes isolated from Hypericum spp.2) Comparative analysis of the predicted and already identified biosynthetic gene clusters of emodinderived polyketides.

General characteristics of the assembled genomes
The genomic DNA of fungal endophytic isolates Fusarium oxysporum 1LF1-1 and 8RF1- The assessment of assembly and annotation completeness confirmed that all the assembled genomes are appropriate for downstream analyses.Generally, more than 95% of the predicted genes were identified as complete and single-copy and the fraction of fragmented and missing genes did not exceed 4% (Fig. 1).
The assembled genomes of Hypericum-borne endophytic isolates were compared to the assemblies obtained from NCBI database (endophytic fungus F. oxysporum GCA_014324765.1, plant pathogenic fungi D. eres GCA_022570805.1 and P. cucumerina GCA_014636675.1, soil fungus S. apiospermum GCA_002158515.1, and thermophilic fungus Thermothielavioides terrestris GCA_900343105.1).According to all-vs-all comparison of nucleotide sequences, 85 to 99.5% of the sequences were aligned to the reference genome assemblies.Occasional translocations and inversions were observed, especially in both aligned F. oxysporum genomes (Fig. 2, Fig. S2).The most remarkable deviation from diagonal line, indicating more frequent insertions or deletions, was found in P. cucumerina 2SF1-3 genome aligned to the genomic sequence of pathogenic P. cucumerina isolated from Arabidopsis thaliana (GCA_014636675.1)(Fig. S3).The genomic rearrangements were less obvious in the aligned D. eres 17SF1-1 and S. apiospermum 2RF1-5, but the overall sequence similarity was relatively low (Fig. S4, Fig. S5).The genomic alignment of two distinct species C. subthermophilus CS-A and T. terrestris revealed remarkable differences with almost 60% contigs unaligned (Fig. S6).
The fungal mitochondrial genomes were assembled de novo from genomic data.The size of circular mitogenomes ranged from 26 to 90 kb (Table 1).

Functional annotation and gene family clustering of genomic sequences
The protein-coding genes represented more than 75% of the predicted genes in all the studied genomes.Their putative functions were inferred by homology-based approach using more annotation sources.The hierarchical classification of GO terms showed that most of the predicted genes are associated with various biosynthetic processes, regulation of metabolism and stress response (Fig. 3).Majority of proteins with catalytic function showed aromatic compounds and ion binding activity.The whole-genome screening of carbohydrateactive enzymes (CAZy) revealed that glycoside hydrolases and glycosyltransferases were the most abundant in all studied genomes (Fig. 4).The lyases, esterases, and other auxiliary enzymes represented only minor part of enzymome.Besides the prediction of protein-coding genes, the functional classification of identified genes was enriched by the prediction of selected non-coding RNAs.The numbers of tRNA and rRNA genes differed among the fungal isolates (Table 1).The whole-genome screening of repetitive regions revealed that transposable elements represented up to 2% of genome size.
The number of predicted genes in mitochondrial genomes varied from 40 (S.apiospermum 2RF1-5) to 81 (D. eres 17SF1-1) and comprised protein-coding genes, rRNA genes, tRNA genes, and open reading frames with unknown functions.Functional annotation of mitogenomes confirmed all 14 genetically conserved proteincoding genes associated with electron transport chain, except for missing atp9 in C. subthermophilus CS-A.The gene encoding ATPase subunit 9 was found in the nuclear genomic sequence.Six additional genes encoding the potential mobile endonucleases were predicted in D. eres 17SF1-1 mitogenome.Several genes were duplicated, e.g.cox1 gene encoding cytochrome oxidase subunit 1 were identified in C. subthermophilus CS-A (Fig. S7).
The clustering of protein-coding genes showed that at least 84% of genes were assigned to orthogroups in all the analyzed isolates (Fig. S8).The number of speciesspecific orthogroups varied among the studied genomes.The whole genome comparison of two F. oxysporum isolates did not reveal any species-specific orthogroup, but the genomes slightly differed in the number of genes unassigned to any orthogroup.

Bioinformatic prediction of polyketide biosynthetic gene clusters
The studied fungal endophytes were isolated from hypericin-accumulating aerial tissues of several Hypericum representatives, except for the isolate 8RF1-3 from H. humifusum roots.Our previous results indicated that fungal endophytes are able to produce the same anthraquinones as the host plants [5,7,9].The genomemining of Hypericum-borne fungal endophytes led to the prediction of candidate biosynthetic gene clusters.In particular, non-ribosomal peptide synthetase (NRPS), PKS, terpene, indole, and hybrid (NRPS and PKS) clusters (Fig. 5) with diverse proportional representation in the analyzed genomes were detected.The number of BGCs in the studies fungal genomes differed.The isolate of P. cucumerina 2SF1-3 contained 24 BGCs, S. apiospermum 2RF1-5 37 and C. subthermophilus CS-A 42 BGCs.The predicted BGCs in C. subthermophilus with their core biosynthetic genes and putative products based on the homology with known clusters were summarized in Table S1.The highest overall number of BGCs in D. eres 17SF1-1 (109 BGCs) reflected the larger genome size compared to the other studied species.Forty-seven clusters found in F. oxysporum 8RF1-3 corresponded with the BGCs in 1LF1-1, with an exception of one NRPS-type cluster.
Comparative analysis of polyketide BGCs of the de novo assembled endophytic genomes and corresponding assemblies obtained from NCBI database showed that  Predicted non-reducing PKS (g5692) contained SAT domain motif GXGXG typical for atrochrysone carboxylic acid synthases.Besides these genes, the genes functionally annotated as fungal-type transcription factor Zn2-Cys6 (g5684), hypothetical protein (g5685), cytochrome P450 monooxygenase (g5686), short chain dehydrogenase (g5687), and oxidoreductase (g5688) in the cluster 94.2 were identified (Fig. 6B).All the predicted genes and their functions are summarized in Table 2. Comparative analysis revealed a similarity with the fungal BGCs of dimeric polyketides, such as rugulosin from Talaromyces rugulosus W13939 or neosartorin from Aspergillus novofumigatus IBT (Fig. 7).It is presumed that dimerization step is catalyzed by predicted gene g5686 annotated as CYP450 monooxygenase sharing the sequence similarity of 52-53% with the predicted dimerization enzymes in Talaromyces and Aspergillus (rugG and nsrP).The predicted MFS transporter, CYP450 and enzymes catalyzing the steps to emodin or phenolic compounds with basic 9, 10-anthraquinone skeleton were identified as putative orthologs of skyrin biosynthetic genes (possible rugulosin precursor from Talaromyces sp.) with sequence similarity from 44 to 60%.The remaining genes g5685, g5687, g5688, and non-catalytical g5684 are presumably pathway-specific genes of novel candidate fungal with the sequence similarity of the A. novofumigatus neosartorin isomerase nsrQ.The gene g5687 was annotated as short-chain dehydrogenase/reductase with a predicted NADP-binding active site and g5688 as a putative member of questin oxidase-like family.A conserved domain of unknown function (DUF4243) was detected in gene g5688 and functional analysis predicted oxidoreductase activity.These genes showed similarity with neosartorin reductase (nsrO, 67%) and oxidoreductase (nsrF, 59%).

Discussion
Plants and fungi are considered as the inexhaustible but underexplored source of valuable natural bioactive compounds.Various plant-colonizing endophytes are able to  produce the same bioactive natural compounds as their hosts [10], and thus provide new biotechnological possibilities for large-scale production of valuable bioactive compounds.Fungal biosynthetic genes are often grouped into gene clusters (BGCs) comprising the genes encoding enzymes responsible for the formation and subsequent modification of metabolite scaffold, transport proteins and DNA-binding proteins.Computational approach is a powerful and rapidly evolving strategy for fungal genome-mining and discovery of novel biosynthetic genes candidates prior to experimental validation.The representatives of the genus Hypericum produce several valuable phytochemicals.Hypericins are bioactive naphtodianthrones with a great biotechnological and therapeutical potential.The biosynthesis of hypericins that proceeds by the polyketide pathway through potential anthraquinone precursors emodin [11], emodin anthrone [12] or skyrin [5] has been generally accepted, but the successive steps are still unclear.Based on the knowledge that Hypericum-borne endophytic fungus, C. subthermophilus (formerly Thielavia subthermophila), is capable of emodin and hypericin synthesis in several subcultures of axenic culture [7], Henzelyová et al. [9] isolated and characterized more than 30 Hypericum-borne endophytes and some of them contained detectable amounts of emodin, emodin anthrone and bisanthraquinones.These endophytes along with C. subthermophilus were selected for genomic DNA sequencing and downstream bioinformatic analyses focused on the identification of candidate polyketide BGCs.As the genome of C. subthermophilus has not been sequenced yet, the assembly presented in this paper is the first reconstructed and characterized whole-genome sequence of this species.Taking into account GC content > 50%, large number of predicted complete genes and gene clusters together with relatively small genome size, it can be concluded that the genome of C. subthermophilus is highly compact.The annotation of C. subthermophilus mitogenome revealed atp9 gene is absent.Several studies reported that fungal atp9 gene was transferred into nuclear DNA [13].Comparative analyses of the other studied Hypericumborne endophytes with their reference genomes from NCBI database confirmed high level of synteny only in F. oxysporum, with two additional polyketide BGCs predicted in both isolates 1LF1-1 and 8RF1-3.Gottschalk et al. [14] observed similar differences of two genome assemblies of F. avenaceum.Pairwise comparisons of endophytic D. eres, P. cucumerina and S. apiospermum with the relevant reference strains revealed more frequent occurrence of indels, overall lower sequence similarity and diversity of BGCs repertoire, especially in D. eres.It is known that fungal isolates from distant geographic regions with diverse lifestyles may differ not only in the composition of BGCs but even in the chromosome sizes [15,16].
Fungal PKS and hybrid NRPS-PKS BGCs represent a major class of biosynthetic clusters.Despite many polyketide BGCs have been identified in fungal genomes, linking of sophisticated bioinformatic prediction with metabolic pathways is still challenging.However, prediction of fungal biosynthetic genes may be complicated by branching of metabolic pathways and observed crosstalks between distant BGCs [17].Most of the fungal anthraquinone derivatives are formed via emodin or chrysophanol precursor (reviewed by de Mattos-Shipley and Simpson [18]).The first steps of biosynthetic route (condensation of eight molecules of malonyl CoA and C6-C11 cyclization) are catalyzed by non-reducing iterative type I Group V PKS with the following features: i) lack of thioesterase domain ('endocrocin-type' PKS), ii) nearby discrete beta-metallo-lactamase-type thioesterase for the release step, iii) domain composition of SAT-KS-AT-PT-ACP, iv) loss of the active cystein site in SAT domain by C to G mutation in the motif GXGXG [19].Based on the previous metabolomic analyses, such PKSs were expected in all studied isolates, except for F. oxysporum 8RF1-3.However, TE-lacking PKSs were identified only in C. subthermophilus and D. eres.Detailed analyses of domain architecture and multisequence alignments revealed that the only PKS that met all the mentioned criteria is the PKS from C. subthermophilus (cluster 94.2) showing similarity with endocrocin PKS from A. fumigatus [20].The next steps of emodin biosynthesis are catalyzed by decarboxylase and anthrone oxidase [21].The predicted gene arrangement in C. subthermophilus cluster 94.2 corresponded with the proposed emodin BGC from parasitic fungus E. weberi predicted by Heine et al. [22].
The basic anthraquinone skeleton is a key structure of many natural dimerized compounds, mainly bisanthraquinones and related xanthones (reviewed by Yuan et al. [23]).Dimerization step involves coupling of two precursor molecules by formation of C-C bond, which is assumed to be catalyzed by cytochrome P450 monooxygenase.Dimerization role of CYP450 was confirmed in the biosynthesis of bisanthraquinones (cladofulvin, skyrin) and xanthones (neosartorin) [24][25][26].Predicted C. subthermophilus CYP450 showed higher sequence similarity to cytochrome P450 from T. rugulosus skyrin cluster (rugG) than CYPs from neosartorin and cladofulvin cluster.RugG is a substrate-promiscuous enzyme that accepts a broader spectrum of emodin-like substrate molecules and catalyzes C5-C5' dimerization to skyrin or bisanthraquinone intermediate air-oxidized to protohypericin [12].
The other candidate tailoring NADP-dependent short-chain dehydrogenase/reductase and predicted oxidoreductase showed similarity with reductase and ring-cleaving dioxygenase from BGC of neosartorin.However, functional classification of oxidoreductase into the questin oxidase-like protein family was ambiguous.Similarly, predicted hypothetical protein was identified as a member of very broad superfamily.Based on the sequence similarity with neosartorin isomerase, it can be presumed that this protein catalyzes interconversion of anthraquinone intermediates.
The alignment of C. subthermophilus cluster 94.2 and skyrin BGC from T. rugulosus showed that biosynthetic genes and additional transport protein are putative homologs.On the contrary, Jahn et al. [27] proposed skyrin BGC from Cyanodermella asteris composed of non-reducing PKS lacking SAT domain and containing TE domain, two short-chain dehydrogenases, scytalone dehydratase and flavin-binding monooxygenase with a different catalyzing mechanism than CYP450.Another PKS with similar domain layout KS-AT-PT-ACP-TE that catalyzes cyclic octaketide synthesis via atrochrysone, was recently identified in a macroscopic fungus Cortinarius sp.[28].
Different PKS domain layout and gene composition of BGCs indicates that anthraquinone biosynthetic routes evolved independently in various fungal taxons.The convergence of anthraquinone biosynthetic pathways is also expected between Hypericum-borne endophytes and hypericin-producing host plants.The suggested hypericin biosynthetic pathway in Hypericum spp.involves octaketide PKS type III [6,29].Compared to fungal multidomain type I PKSs, plant-type III PKSs comprise a single ketosynthase domain and thus presumably requires additional octaketide cyclase and thioesterase to form atrochrysone carboxylic acid further converted by decarboxylase and oxygenase to emodin.However, any candidate hypericin biosynthetic gene has not been validated yet.

Conclusions
This work provides an insight into the genomes of Hypericum-borne endophytic fungi able to produce anthraquinones.The genomes of selected endophytes were sequenced and exploited by in silico analyses.The C. subthermophilus genome was completely sequenced and characterized for the first time.A potential anthraquinone BGC was discovered solely in the genomic sequence of C. subthermophilus using bioinformatic prediction and subsequent precise inspection of identified polyketide gene clusters.The cluster includes putative biosynthetic genes of emodin-type anthraquinone, cytochrome P450 monooxygenase with assumed dimerization role, other enzymes, fungal transcription factor and transport protein.Predicted gene cluster might be associated with the biosynthesis of bisanthraquinone further tailored by pathway-specific enzymes.The obtained computationally-guided results will be subjected to the experimental validation.The knowledge on genomic background of bisanthraquinone biosynthesis in endophytes is a key for understanding the anthraquinone biosynthesis in plants and for considering the simple eukaryotic system for prospective biotechnological production of hypericins and other potential host-plant derived anthraquinones with anticancer properties.

Fungal material and cultivation
Endophytic fungi isolated from Hypericum spp.were selected based on their ability to produce anthraquinones and bisanthraquinones innate to the host plants.The experimental group comprised two Fusarium oxysporum isolates varying in metabolite production, Scedosporium apiospermum, Plectosphaerella cucumerina, and Diaporthe eres which were collected from plants acclimated to outdoor conditions in the Botanical Garden of Pavol Jozef Šafárik University in Košice, Slovakia [9] (Table 3).The cultures were identified by molecular markers ITS rDNA and protein-coding gene region of tef1α.Moreover, fungi identified as F. oxysporum were classified to F. oxysporum species complex 33 based on the intronrich portion of tef1α and maintained at the Institute of Biology and Ecology of Pavol Jozef Šafárik University in Košice, Slovakia, while Canariomyces subthermophilus (formerly Thielavia subthermophila) isolated from Hypericum perforatum of Himalayan provenance [7] was purchased from Leibnitz Institute DSMZ -German Collection of Microorganisms and Cell Cultures (accession number DSM 21024) and its authenticity was confirmed by ITS rDNA sequence.Fungal cultures stored at -80 °C in 20% (v/v) glycerol were transferred to potato dextrose agar (PDA, Difco) and cultivated in the dark at 26 °C with 7-day subculture interval.Before DNA extraction, fungal mycelia were transferred to potato dextrose broth (PDB, Difco) or malt extract broth (MEB) and cultivated for 7 days at 130 rpm in the dark at 26 °C.

Genomic DNA isolation and libraries preparation
Total DNA was extracted by NucleoSpin Soil Kit (MACHEREY-NAGEL) according to manufacturer's instructions.Quality and concentration of total DNA was checked by spectrophotometer (Biotek Synergy HT) and fluorometer (Qubit 4.0) using dsDNA HS Assay kit.Swift 2S Turbo Flexible v2 DNA Library Kit with KAPA Unique Dual-Indexed Adapters was used to prepare libraries using 500 ng of genomic DNA per sample.Libraries were quantified by KAPA SYBR ® FAST qPCR Kit on qTOWER 3 Series real-time thermocycler (AnalytikJena) and fragments of prepared DNA libraries were examined by 1.8% TBE agarose gel electrophoresis.Final quality control was performed by High Sensitivity DNA Assay kit (Agilent Technologies) on Agilent 2100 Bioanalyzer.

Whole-genome sequencing and assembly
The libraries were sequenced on Illumina platform NextSeq500 (CEITEC Masaryk University, Brno, Czech Republic) and MiSeq (Pavol Jozef Šafárik University in Košice, Slovakia) in paired-end mode with 2 × 150 bp read lengths and 100X coverage.After quality control by FastQC [30], the quality and adapter trimming was performed using Cutadapt V4.3 [31].The trimmed reads were assembled de novo by SPAdes with different k-mer values (21,33,55,77) [32].The genes were predicted by stand-alone version of Augustus [33], a program based on the generalized Hidden Markov models.The prediction was performed using pre-trained model for Fusarium graminearum.The completeness of assembled genomes was evaluated by locally installed BUSCO (Benchmarking Universal Single-Copy Orthologs) software V5 based on the fungal database (fungi_odb10) [34], interactive webbased framework GenomeQC [35], and the assembly statistics were extracted using AGAT V1.2 [36].The assembled genomes were aligned and compared to the reference genomes downloaded from NCBI database by MUMmer V4 [37] and D-GENIES [38].The mitochondrial genomes were assembled from genomic data by GetOrganelle V1.7.7 [39] and annotated by MITOS V2.1.7 [40].The circular map of C. subthermophilus mitogenome was drawn using Proksee [41].

Identification of orthologous gene families
The groups of orthologous genes (orthogroups) were inferred using OrthoFinder [53].The species tree was constructed by STAG (Species Tree from All Genes) algorithm based on the sets of one-by-one orthologs in all species.

Prediction of biosynthetic gene clusters
Prediction of biosynthetic gene clusters was performed via fungal version of antiSMASH V6.0 pipeline [54], using both genome assembly and feature annotation files with relaxed detection strictness.The pipeline comprised these successive steps: i. analysis of PKS/NRPS (polyketide synthase/nonribosomal peptide synthetase) domain architecture, ii.prediction of chemical structures, iii.mapping on the database of gene clusters, iv.analysis of the genomic sequences by the secondary metabolism Clusters of Orthologous Groups (smCOG).The identified clusters were manually inspected and ambiguous gene functions were additionally verified by BLASTp or BLASTx against non-redundant NCBI protein database [55].The multiple cluster alignments were performed by CAGECAT clinker [56].The BGC GenBank files were obtained from MIBIG database [57] and NCBI.The PKS domain structure was visualized by DOG V2.0 [58].

Fig. 2
Fig. 2 The representative alignment of F. oxysporum GCA_014324765.1 reference genome and F. oxysporum 1LF1-1.The observed translocation and inversion are marked by circles.The alignments are colored according to the sequence identity

Fig. 3
Fig. 3 Gene ontology classification of C. subthermophilus genes.The major category Biological process is colored by green, Cellular component by tawny and Molecular function by Chetwode blue

Fig. 4 Fig. 5 Fig. 6 A
Fig. 4 Distribution of carbohydrate-active enzymatic classes in the genomes of Hypericum-borne endophytic fungi

Table 1
Characterictics of de novo assembled genomes of Hypericum-borne endophytes a Assembly

Table 2
The gene composition of candidate anthraquinone BGC identified in C. subthermophilus genome.The gene functions were predicted by homology-based search against NCBI non-redundant protein database giving the identity of aligned sequences Fig. 7 Comparison of candidate anthraquinone cluster from C. subthermophilus and identified BGCs of emodin type compounds in other fungi.Sequence similarity of predicted genes is depicted by grayscale