Genome sequencing of Rigidoporus microporus provides insights on genes important for wood decay, latex tolerance and interspecific fungal interactions

Fungal plant pathogens remain a serious threat to the sustainable agriculture and forestry, despite the extensive efforts undertaken to control their spread. White root rot disease is threatening rubber tree (Hevea brasiliensis) plantations throughout South and Southeast Asia and Western Africa, causing tree mortality and severe yield losses. Here, we report the complete genome sequence of the basidiomycete fungus Rigidoporus microporus, a causative agent of the disease. Our phylogenetic analysis confirmed the position of R. microporus among the members of Hymenochaetales, an understudied group of basidiomycetes. Our analysis further identified pathogen’s genes with a predicted role in the decay of plant cell wall polymers, in the utilization of latex components and in interspecific interactions between the pathogen and other fungi. We also detected putative horizontal gene transfer events in the genome of R. microporus. The reported first genome sequence of a tropical rubber tree pathogen R. microporus should contribute to the better understanding of how the fungus is able to facilitate wood decay and nutrient cycling as well as tolerate latex and utilize resinous extractives.

Horizontal gene transfer (HGT) is an integral part of genome evolution in fungi. The lateral transfer of genes, gene clusters or entire chromosomes can have significant effects on disease breakout and metabolic activity 22,23 . Importantly, HGT was demonstrated as an important driving force shaping the adaptation of fungi to the phytopathogenic lifestyle 24 .
We report the results of our analysis of the first genome sequence of a tropical rubber tree pathogen R. microporus. The availability of the fungus genome further complements the sequenced genome of its rubber tree host 25 . The main objectives of this study were (i) to perform the analysis of the plant cell wall-degrading machinery of the fungus, (ii) to elucidate the mechanisms of latex degradation and utilization by R. microporus, (iii) to identify putative effectors used by R. microporus during host tree colonization, (iv) to identify putative horizontal gene transfer (HGT) events and their role in the adaptation of R. microporus to its pathogenic lifestyle, and (v) to clarify the systematic position of R. microporus within the order Hymenochaetales. Whole genome and transcriptomic analysis of R. microporus grown on natural rubber latex revealed a variety of genes potentially related to pathogenicity and virulence, effector-like proteins, lignocellulose and latex degradation. This study is expected to accelerate further research on this economically important tropical wood rotting fungus.
Methods fungal isolates and culture conditions. R. microporus isolate ED310 used for the genome sequencing was isolated from a diseased H. brasiliensis tree from a rubber tree plantation at the Rubber Research Institute, Nigeria in 2012 3 . The isolate was deposited in HAMBI culture collection center with ID number HAMBI/FBCC 2356. Genomic DNA was isolated from 7 days-old R. microporus cultures harvested from 2% w/v malt extract agar (MEA) plates overlaid with cellophane according to the protocol provided in the Supplementary Note 1.
Genome sequencing and annotation. Sequencing of the Rigidoporus microporus ED310 genome was done using Pacific Biosciences RS II. Unamplified libraries were generated using the Pacific Biosciences standard template protocol to create >10 kb libraries. Five micrograms of gDNA was used to create each library, followed by shearing of the DNA using Covaris g-Tubes ™ to create sheared fragments of >10 kb. The sheared DNA fragments were then further processed using Pacific Biosciences SMRTbell template preparation kit. The DNA fragments were treated with DNA damage repair and the ends were repaired to create blunt-ended fragments which were 5´ phosphorylated. Adapters were then attached to the fragments to create the SMRTbell template for sequencing. The SMRTbell templates were then treated with exonuclease to purify them and size-selected using AMPure PB beads. PacBio primer was annealed to the SMRTbell template library and sequencing polymerase (V. P6) was bound to them. Sequencing of the SMRTbell template libraries was carried out on a Pacific Biosciences RSII sequencer using Version C4 chemistry and 1 × 240 min movie run times.
Filtered subread data was assembled with Falcon version 0.7.3 (https://github.com/PacificBiosciences/ FALCON) to generate the initial assembly. Assembly of the mitochondrial genome was done separately from the Falcon pre-assembled reads (preads) using an in-house tool (assemblemito.sh), used to filter the preads, and finally cleaned with Quiver version smrtanalysis_2.3.0.140936.p5 (https://github.com/PacificBiosciences/ GenomicConsensus). A secondary Falcon assembly was created using the filtered preads and improved with finisherSC version 2.0 26 and cleaned with Quiver. Contigs less than1000 bp were excluded.
The annotation of the R. microporus genome was executed with the JGI genome Annotation pipeline, which uses tools for gene prediction, annotation and analysis 27,28 . Different gene predictors were utilized to generate multiple sets of gene models. The predicted gene models were automatically filtered based on results of similarity to proteins from other species coupled with the support from the gene expression data to produce a final non-redundant catalog of genes which represents the best gene model located at each locus. Multi-gene families prediction was carried out using the Markov clustering algorithm (MCL) 29 , a useful part of the JGI annotation pipeline, and annotation was done using PFAM domains present in cluster member sequences. Identification of secreted proteins were done using SignalP v.4.1 (sensitive mode) 30 , TargetP v.1.1 31 and TMHMM v.2.0 32 to predict the presence of signal peptide, targeted cellular localization and transmembrane domain (TM), respectively. Proteins having more than two TMs and/or a single TM not overlapping with the signal peptide were excluded from the analysis. The predicted secretome was blastp against PHI-base v.4.4, the pathogen-host interaction database 33 with the cut off E-value set to 10 -5 to identify potential virulence-related secreted proteins.

Annotation of class ii peroxidases, multicopper oxidases and GMc oxidoreductases.
A screening of the automatically-annotated genome of R. microporus was performed by BLASTing the amino acid sequences of five selected class II peroxidases (generic peroxidase, GP; short manganese and long manganese peroxidases, MnP-short and MnP-long; versatile peroxidase, VP; and lignin peroxidase, LiP) against the filtered model protein database of this fungus available at MycoCosm. Six class II gene models were identified and manually annotated based on: (i) the highest sequence identities for each protein sequence derived from the predicted gene; (ii) multiple alignment with 145 heme peroxidase protein sequences from twelve fungal species (ten species of Polyporales and two species of Hymenochaetales); and (iii) examination of theoretical molecular structures obtained by homology modeling using crystal structures of related peroxidases as templates and programs implemented by the automated protein homology modeling server "SWISS-MODEL" 34 .
The multicopper oxidase genes present in R. microporus genome were analyzed following two different strategies: (i) BLASTing with R. lignosus laccase (1V10.A) as a probe; and (ii) SEARCHing by keyword using "multicopper oxidase".
After multiple alignment with already known MCO proteins to search for conserved motifs and residues, the sequences were manually curated. Molecular models of distinctive proteins were built up to better determine the different types of MCOs.
The screening for each of the GMC oxidoreductase families, glucose oxidase (GOX), pyranose-2-oxidase (POX), pyranose dehydrogenase (PDH), aryl alcohol oxidase (AAO), methanol oxidase (MOX) and cellobiose dehydrogenase (CDH), was performed by querying in the filtered model protein database of R. microporus using previously characterized sequences of GMC oxidoreductases from 10 fungal species 35 as template, and sequences with E-values ≤ 1e-100 were selected.
Identification of putative horizontal gene transfer events. The predicted proteins from the R. microporus genome were scanned with the program DarkHorse 36 to identify candidate horizontally transferred genes. We considered as HGT candidates proteins with a normalized LPI score of less than 0.8 and that had at least 10 significant BLAST matches. For each of the 59 candidates, a BLAST search of the nr database was performed and the 50 best matches were used for phylogenetic tree construction. The protein sequences were aligned with MAFFT v.7.388 37 and phylogenetic trees were constructed with FastTree v.2 38 . BLAST searches, alignments and phylogenetic analyses were performed with Geneious v.11 39 . RnA-seq analysis of R. microporus transcriptional responses to natural latex. Growth rates of R. microporus in the presence of the rubber tree latex were estimated in vitro. Natural rubber tree latex (CAS no. 9006-04-06), Weber & Schaer GmbH & Co. KG (Hamburg, Germany) as low ammonia latex milk (Neotex LA) was utilized for the experiment. Before media preparation of the latex milk, the stabilizing ammonia was removed by centrifugation (5 min at 10,000 × g). The ammonia-free top layer was collected and utilized for further experiments. The latex was added to the modified Pachlewski P5 agar medium 40 (without carbon source) at the concentration of 0.8% (v/v). This medium was used to overlay P5 agar plates supplemented with either 0.1%, 0.5%, and 1% glucose or 1% rubber wood sawdust as a source of carbon ( Supplementary Fig. 1). Non-overlaid plates were used as a control.
Fungal hyphal growth was measured six days post inoculation ( Supplementary Fig. 2). The set-up containing 1% rubber wood sawdust had the highest hyphal growth and was thus selected for RNA extraction ( Supplementary Fig. 2). Total RNA was extracted from three biological replicates of both the treatment and control according to the protocol of 41 .
Library construction and sequencing were performed at the Beijing Genome Institute, Hong Kong. Messenger RNA extracted from total RNA using oligo (dT) beads was fragmented in buffer to generate short fragments of 200 bp. Random hexamers were then used to synthesize first strand cDNA, followed by addition of dNTPs, RNase and DNA polymerase I to synthesize second strand cDNA. Sequencing adaptors were attached to fragments which were then amplified by Polymerase Chain Reaction. Six cDNA libraries (3 biological replicates for latex-growing cultures and 3 for control) were created. The six cDNA libraries were sequenced separately using the Illumina HiSeqTM 2000 sequencing platform.
Afterwards, the R. microporus genome was indexed and mapped with STAR v.2.5.2b 42 . Raw read count table was generated by htseq-count script within HTSeq v.0.6.1p1 43 using the obtained uniquely mapped reads. The count table was then used to identify differentially expressed genes with DESeq v 1.38.0 44 . Gene Ontology (GO) (biological process and molecular function) enrichment analysis among differentially expressed genes were performed with topGO v.2.30.2 45 using Fisher's exact test (p < 0.05). The GO annotations of all predicted genes were retrieved from the JGI portal.
interactions of R. microporus with other fungal species. R. microporus was co-cultured with three representative species of basidiomycetes which were selected partly based on their lifestyle as saprotrophs and/ or as biocontrol agents in other pathosystems: saprotrophic white-rot fungus Phanaerochate chrysosporium, saprotrophic white-rot and saprotroph/biocontrol agent Phlebiopsis gigantea and the saprotroph Mycena sp. Dual cultures were prepared on MEA plates and cultivated at 25 °C. Three time points were selected and used for RNA extraction. For time point 1 (no hyphal contact), R. microporus was co-cultured with either the saprotrophic fungi P. chrysosporium, P. gigantea or Mycena sp. for 2 days, 4 days and 6 days, respectively. At time point 2 (initial hyphal contact), R. microporus was co-cultured with P. chrysosporium, P. gigantea and Mycena sp. for 4 days, 6 days, and 10 days, respectively. For time point 3 (a few days after hyphal contact), R. microporus was co-cultured with P. chrysosporium, P. gigantea and Mycena sp. for 8 days, 8 days, and 13 days, respectively. Expression level of the analyzed genes was compared with their expression in the pure culture of R. microporus (normalized as 1.0). RNA was extracted using TRI Reagent (Sigma-Aldrich. Inc., USA) according to manufacturer's instructions. cDNA was synthesized with Thermo Scientific reagents (DNase I, EDTA, oligo-dT, RevertAid Reverse Transcriptase) according to the protocol supplied by the manufacturer. qPCR was performed on LightCycler ® 480 Instrument II (Roche, Switzerland). Seven target genes (protein models 933844, 207751, 252098, 692493, 170617, and 238476) were selected from the predicted set of R. microporus SSPs based on their deduced amino acid sequence and transcriptomic expression profile. Additionally, two reference genes (actin and TFIIIC) 46 were used. The statistical analysis of qPCR results was performed in EasyqpcR 47 . phylogenetic analysis. Forty single-copy genes were initially screened, and ten of them were selected for the phylogenetic analysis (Supplementary Table 1) based on several criteria explained below. The single copy orthologs were obtained from the database OrthoDB v9.1 48 . The orthologs were selected based on their low evolutionary rates and presence across the class Agaricomycotina. Single copy ortholog of Fomitiporia mediterranea (Hymenachaetales) for each gene was obtained and used to perform a blastP on the filtered model sets of proteins of the respective genomes in the Joint Genome Institute (JGI) genome portal (Supplementary Table 2) using default parameters. Sequences obtained from the blast results for each gene were then analyzed further.
www.nature.com/scientificreports www.nature.com/scientificreports/ Alignments of sequences for each of the individual single copy genes were obtained using MAFFT v. 7 37 . Quality control was manually carried out to detect and remove potential paralogs. The aligned proteins were then treated with Gblocks 49 with the following settings: maximum number of contiguous non-conserved positions = 4; minimum length of blocks allowed = 10. Proteins detected as paralogs after Gblocks treatment were removed before phylogenetic analyses. The final individual 10 protein alignments after Gblocks treatment were concatenated to produce a single alignment using Geneious R6.0.6 39 . The final ten gene concatenated dataset has a length of 12,074 aa. Individual alignments for each gene and the concatenated datasets are provided as Supplementary Files 1-11. Phylogenetic trees were constructed using two different phylogenetic methods, maximum likelihood (ML) and neighbor joining (NJ). ML was performed in the online version of PhyML 3.0 50-52 , using automatic model selection by SMS 53 and Akaike information criterion (AIC) 52 . NJ analyses was conducted using MEGA 7 53 .

Results
Genome features. The genome of Rigidoporus microporus was assembled into 283 contigs (281 contigs >2 kb) with an average read coverage depth of 80.32× and a genome assembly size of 42.09 Mbp (Fig. 2). The genome size of R. microporus is within the range reported for other members of Hymenochaetales. Using diverse www.nature.com/scientificreports www.nature.com/scientificreports/ gene prediction methods, which included the use of previously generated RNA-seq data 10 , we predicted 10,917 protein-coding genes. The mitochondrial genome of R. microporus contains 27 predicted protein-coding genes (Supplementary Table 3) and 24 tRNA genes (Supplementary Table 4). General properties of the R. microporus genome can be found in Table 1. phylogenetic analysis. Our phylogenetic analysis based on 10 single-copy conserved genes unambiguously places R. microporus within the order Hymenochaetales. Corresponding clade received 100% bootstrap support in both ML (Fig. 3) and NJ ( Supplementary Fig. 9) analyses. R. microporus occupies a basal position within that group, being a second most basal branch after the two species of Rickenella (Fig. 3).
Wood-degrading enzyme machinery. R. microporus possesses a repertoire of carbohydrate-and lignin-degrading enzymes characteristic for white-rot fungi. A comparison of the number of predicted R. microporus CAZy genes with that of other basidiomycetes is provided in Supplementary Table 5.
Predicted lignin-degrading enzymes are represented by six class II peroxidases (PODs), five laccases and eight GMC oxidoreductases (Supplementary Note 2). Analysis of the multicopper oxidase genes present in R. microporus genome gave a total of seven multicopper oxidases (MCO). Further molecular models of the distinctive proteins were built up to better determine the different types of MCOs (Fig. 4). It was concluded that two out of the seven MCO sequences corresponded to ferroxidases (889958, 900905) similar to the fungal Fet3 proteins, while the other five sequences were laccases (921691, 942192, 942193, 942194, 406895). The identified two ferroxidases, four laccases, four manganese peroxidases and three aryl alcohol oxidases also appeared in the mentioned reduced virulence gene list from PHI-base database, further supporting their roles of promoting fungal colonization and virulence.
Our analysis showed that R. microporus has the second lowest (after Onnia scaura) number of CAZy genes among the sequenced species of the order Hymenochaetales. Within the CAZys, R. microporus possess the lowest number of glycosyltransferases (GT) and the second lowest numbers of glycoside hydrolases (GH) and enzymes with axillary activities (AA) among the Hymenochaetales. At the same time, it has the highest number of LPMO (AA9) genes and the second highest number of CBM genes among the analyzed species of this order (https://genome.jgi.doe.gov/mycocosm/annotations/browser/cazy/summary;QzCg6t?p=Rigmic1). The number of R. microporus CAZymes tentatively involved in the degradation of hemicellulose and pectin is within the range observed in other Hymenochaetales, but lower than the average values for the sequenced species of Agaricomycotina. At the same time, R. microporus has the highest total number (39) of cellulose-degrading enzymes (members of the families GH6, GH7, GH45, GH74 and AA9) among the sequenced members of Hymenochaetales. This number is also significantly higher than the average for the class Agaricomycotina.
Principal coordinates analysis based on the copy numbers of different classes of CAZymes placed R. microporus close to Botryobasidium botryosum (Cantharellales) (Fig. 5), a species with a wood decay mode that might be plesiomorphic for the Agaricomycetes as a whole 54,55 . However, unlike B. botryosum, R. microporus possesses ligninolytic class II peroxidases (PODs) and laccases, characteristic for the white-rotting fungi.
Secondary metabolism. The genome of R. microporus encodes a diverse set of genes implicated with a predicted role in secondary metabolism (Supplementary Table 6). The sole NRPS gene nrp1 in R. microporus encodes a protein with the domain structure A-T-C-T-C, which is a common architecture for production of hydroxamate siderophores involved in iron homeostasis 56 . Indeed, the protein shows 27% identity to the well-characterized SidC protein from Aspergillus fumigatus 57 . However, the domain pattern of Nrp1 suggests that it is more likely a SidD homolog (20% identity). SidD condenses anhydromevanolyl hydroxyornithine to form the siderophore www.nature.com/scientificreports www.nature.com/scientificreports/ fusarinine C, which is further acetylated to triacetylfusarinine C (TAFC), the major siderophore in Aspergillus nidulans and A. fumigatus to acquire iron from the environment 58 . The formation of hydroxamate-based siderophores requires a specific monooxygenase for hydroxylation of L-ornithine or L-lysine to form precursor molecules for the siderophore biosynthesis 55 . A SMO1-like flavin-dependent monooxygenase (mon1) is encoded  www.nature.com/scientificreports www.nature.com/scientificreports/ next to the nrp1 indicating that mon1 and nrp1 form a SM gene cluster. It is therefore very likely that R. microporus produces hydroxamate-derived siderophores, as it has already been shown for other basidiomycetes such as Laccaria sp. and Ceriporiopsis subvermispora B 59,60 .
The R. microporus NPRS-like protein 1 Nlr1 is 65% identical to the L-α-aminoadipate reductase Lys2 from C. subvermispora, Dichomitus squalens and Heterobasidion irregulare. L-α-aminoadipate reductases catalyse the ATP-and NADH-dependent reduction of L-α-aminoadipate to its 6-semialdehyde, the 6 th step in fungal L-lysin biosynthesis starting from α-ketoglutaric acid. Like most Lys2-like enzymes, Nrl1 has an N-terminally extended domain (ADA domain), that is shown to be essential for its catalytic activity 61 . Therefore, we assume that R. microporus synthesizes L-lysine de novo as shown for many other fungi 62 .
In total, 16 ArmP-like terpene cyclases (TCs) encoded in 11 gene clusters were identified in the R. microporus genome (Supplementary Fig. 6). All TCs contain the aspartate (D) rich domain of the active site that coordinates a trinuclear Mg 2+ cluster that binds the diphosphate of the isoprenoid substrate 63 . However, 15 TCs have the classical consensus DDX motif where X is a hydrophobic amino acid such as Phe, Tyr, Trp, Leu or Val. In Ter12 one Asp residue is replaced by Glu at the first position (EDX) and the protein is probably inactive. The position of the active sites in the TCs is highly variable in R. microporus: The catalytic sites are positioned either at the N-terminus (9 TCs), at the C-terminus (4 TCs) or in the centre of the polypeptide chain (3 TCs). Active sites with a DDY/FX 2-3 CD consensus motif are exclusively found in N-terminal active sites. Consensus motifs of LIX 6-9 DX 2-3 DD are located in N-terminal and central active sites, whereas MDD consensus motifs are only observed in C-terminal active sites. Ter1 and Ter10 are highly identical (99.1%) and differ only in 3 amino acids (I286L, Y306H, M341V) suggesting that both TCs are redundant in their function as frequently observed in SM enzymes from basidiomycetes 64 .
Interestingly, two possible large SM gene cluster encode three TCs each (Ter3A-C and Ter6A-C, respectively). The close localization of the genes indicates a clustered co-expression, suggesting that R. microporus follows the biosynthesis gene cluster paradigm as shown for numerous species of ascomycetes and basidiomycetes 65 . Many reading frames were identified that may encode tailoring enzymes, often located in close vicinity to the key enzymes. The FAD, NAD or metal ion-dependent oxidoreductases/monooxygenases represent the largest group of possible backbone-modifying enzymes in R. microporus. Interestingly, within the cluster of ter3A-C, a possible major facilitator superfamily (MFS) transporter was found. MFS transporters are known to transport small molecules rather non-specifically into the extracellular space, which in turn causes resistance to the producer of anti-fungal metabolites 66 .
Three out of 11 terpene gene clusters encode C6 zinc finger proteins. C6-binuclear Zn (II) 2 Cys 6 transcription factors have been shown to be involved in the regulation of sexual life cycle and secondary metabolism in ascoand basidiomycetes 67 . Both identified large gene clusters encode zinc finger proteins, suggesting that the set of genes are co-expressed and need special triggers to be activated such as stress conditions. Very often, these gene clusters remain silent under standard laboratory conditions 68 . Horizontal gene transfer events. Analysis of the predicted proteins with the program DarkHorse identified 59 candidate horizontally transferred genes with normalized LPI (Lineage Probability Index) scores of less than 0.8 and at least 10 BLAST hits. Of the 59 candidates, 5 had best hits to proteins from bacteria, and 47 had best hits to proteins from the Ascomycota (Supplementary Table 7). A phylogenetic tree was constructed for each www.nature.com/scientificreports www.nature.com/scientificreports/ candidate. Nine final HGT candidates were selected based on manual inspection of the phylogenetic trees. Among the nine final candidates, 2 indicate putative HGT events from bacteria and seven indicate HGT events from the Ascomycetes (Supplementary Fig. 6). The identified genes had similarities to pectate lyase (gene models 941964 and 975179), L-amino acid oxidase (983304), Na + /H + exchanger family proteins (871220), fucose-specific lectin (1012254), peptidylprolyl isomerase (978311), aldehyde-activating protein (897619) and Metarhizium anisopliae and Pseudogymnoascus sp. hypothetical proteins (882912 and 870991, respectively). Supplementary Fig. 6. No significant differences in the expression level of the analyzed genes could be observed at the time point 1 (Supplementary Fig. 6). At the time point 2, all selected genes except for the gene 933844 had lower expression level than in the control culture ( Supplementary Fig. 7). At the time point 3, three genes encoding predicted small secreted proteins (933844, 207751, and 170617) were found to be upregulated in R. microporus co-cultured with saprotroph P. chrysosporium (Fig. 6). The gene 941140 was induced in the co-culture with Mycena sp., whereas the gene 692493 was downregulated in the co-cultures with P. gigantea and Mycena sp. (Supplementary Fig. 8).

Expression of a sub-set of SSP or putative effector-like genes in interspecific fungal interactions. The development of fungal co-cultures is illustrated in
transcriptomic analysis of R. microporus response to natural latex. Our RNA-seq analysis identified 51 genes with the significantly higher transcript abundance (FC ≥ 4, FDR ≤ 0.05) (Supplementary Table 8) and 199 genes with the significantly lower transcript abundance (FC ≤ 0.25, FDR ≤ 0.05) (Supplementary Table 9) during the growth on latex-containing medium. A list of the most highly up-regulated and down-regulated transcripts that have functional annotation are shown in Tables 2 and 3. The set of the genes with the higher expression level included three genes encoding predicted tripeptidyl peptidases, two aspartic protease genes, two genes for cytochrome P450 as well as genes for a putative GH16 glycosyl hydrolase, an ammonium transporter, a MFS transporter, a manganese peroxidase, a NADPH oxidase, a catalase and four hydrophobin-encoding genes. Among the genes with the lower transcript abundance, we identified two LPMO-encoding genes, genes for GH5 and GH13 glycosyl hydrolases, CE16 carbohydrate esterase and two genes encoding putative carbohydrate-binding module proteins CBM1 and CBM13. Additionally, Gene Ontology (GO) terms enrichment with respect to lower and higher transcript abundance is shown in Supplementary Table 10.
Genes putatively involved in pathogen/host interactions. The predicted set of R. microporus proteins included 813 putative secreted proteins. Among the secreted proteins that have hits in the database PHI-base 33 , 229 proteins have hit annotations of "reduced virulence", "loss of pathogenicity" or "effector (plant avirulence determinant)". The hits of 38 proteins were annotated as "effectors" based on experimental evidence collected in this database. A total of 70 and 201 proteins were assigned to the annotation of "reduced virulence" and "loss of pathogenicity", respectively (Supplementary Table 11). Potential function of effectors based on PHI blast is listed in Supplementary Table 12.

Discussion
The availability of the genome sequence of R. microporus provides excellent opportunities for the genome mining and comparative analysis of this important tree pathogen. Our multigene phylogenetic analysis clearly showed that R. microporus is nested within the representatives of the order Hymenochaetales. This observation agrees with   Table 2. Twenty most up-regulated genes with functional annotation during growth of R. microporus on latex. a corresponds to assembled and annotated genes from the transcriptome. b Binary logarithm of fold change calculated from the fragments per kilobase per million reads (FPKM). c is the best hit of InterPro database.  Table 3. Twenty most down-regulated genes with functional annotation during growth of R. microporus on latex. a corresponds to assembled and annotated genes from the transcriptome. b Binary logarithm of fold change calculated from the fragments per kilobase per million reads (FPKM). c is the best hit of InterPro database. www.nature.com/scientificreports www.nature.com/scientificreports/ the results of the previous phylogenetic reconstructions 3,69 . However, the placement of R. microporus within the order of Hymenochaetales might require some nomenclatural rearrangements, as the type species of the genus Rigidoporus belongs to the order Polyporales 8 . The nomenclatural issues concerning Rigidoporus and related species will be fully addressed in a separate forthcoming publication (Miettinen et al., in prep.).
The size of the R. microporus genome is similar to the genome size of such members of Hymenochaetales as Trichaptum abietinum, Schizopora paradoxa and Onnia scaura. However, the number of predicted protein-coding genes in R. microporus is somewhat lower than in the mentioned species.
R. microporus is an efficient wood decomposer, capable of simultaneous degradation of lignin and cellulose, characteristic for white-rot fungi 2 . Our analysis of PCW-degrading capabilities of R. microporus revealed that this species has relatively low numbers of GT-and GH-encoding genes. At the same time, the repertoire of CAZys encoded by R. microporus reflects its specialization towards an efficient cellulose degradation. In particular, a high number of LPMO-encoding genes in R. microporus genome might indicate the important role of this class of enzymes in cellulose degradation. The number of genes encoding lignin-degrading enzymes in R. microporus genome is lower than in other representatives of Hymenochaetales, nevertheless, their repertoire is sufficient for lignin degradation, as indicated by the in vitro experiments on wood decomposition 2 . The presence of lignin-degrading peroxidases clearly separates R. microporus from Botryobasidium botryosum, even if these two species grouped together in our PCoA analysis based on the copy number of CAZyme-encoding genes.
A role of R. microporus secondary metabolites (SM) in the infection process or in the metabolic switch from the necrotrophic to saprotrophic lifestyle is not established. The majority of encoded key enzymes involved in SM production are terpene cyclases (TC) (Supplementary Table 6). Hence, soluble and volatile terpenes seem to be the major metabolites produced by R. microporus, as already demonstrated for other higher fungi 70 . In addition, there are a few characteristic multidomain enzymes such as one non-ribosomal peptide synthetase (NRPS) protein and a NRPS-like protein encoded within the genome. However, putative polyketide synthases (PKS) or PKS-NRPS hybrid genes seems to be absent in the genome. In contrast, a wide variety of SM tailoring enzymes, such as monooxygenases and methyltransferases are encoded to modify the terpenoid backbones. There are no Arm1-like flavin-dependent halogenases encoded in the genome, suggesting that no halogenated terpenes are formed. Furthermore, many of the SM genes are clustered and contain regulatory elements (transcription factors) and/or putative transporters for efficient metabolite export. An in-depth analysis of the fungus´ metabolome and volatome may not only contribute to the discovery of novel biosynthetic routes and compounds but would also unravel the ecological impact of the produced metabolites on the fungus itself, its host tree, forest ecosystem or its microbial competitors.
The majority of the candidate horizontally transferred genes appear to have originated from ancestral members of the Ascomycota. This was unexpected since, in the case of Colletotrichum spp., bacteria appeared to be the most frequent donors of horizontally transferred genes 71 . In fact, most horizontally transferred genes in fungi are of bacterial origin 22 and the presence of a relatively large number of ascomycete to R. microporus transfers suggests that there may be a unique aspect to the R. microporus ecology or biology that enables such transfers. An alternative explanation to HGT is that the genes were vertically inherited from a common ascomycete/basidiomycete ancestor. This would require massive numbers of gene losses in many basidiomycete lineages, an explanation that is less parsimonious as it requires a larger number of evolutionary changes. Also, if HGT candidates were the product of vertical inheritance followed by gene loss, then one would expect to find basidiomycete/ascomycete HGT candidates in the ascomycetes, which is not the case.
The use of artificial agar media to assess fungal growth and interactions have always been a major concern particularly on whether such results could be extrapolated to natural conditions 72,73 . Previous studies have used artificial agar media to assess the dynamics of fungal growth and interactions 74,75 . Other authors 76 , noted that growth in artificial medium provides best possible ways of analyzing fungal interactions and growth. Other studies 77 reported that interactions of some fungi on woody substrate and artificial (agar) media did not significantly vary. The analysis of interspecific interactions between R. microporus and other fungi indicated a potential role of SSP in these interactions. The changes in the expression of SSP-encoding genes were particularly pronounced in the case of interaction of R. microporus with the saprotroph P. chrysosporium. The latter species was characterized by faster growth rate, and the higher expression of SSP by R. microporus could be a defense reaction against a strong competitor. Previous studies have shown saprotrophic basidiomycetes SSPs as putative effectors 78 . The interactions of R. microporus with saprotrophic biocontrol agent P. gigantea and saprotrophic Mycena sp. had less pronounced effect on the expression of SSP-encoding genes.
R. microporus is a successful pathogen of rubber trees, able to efficiently colonize and kill living trees. During the host colonization process, the fungus likely interacts with the latex, abundantly produced by the host tree. However, the role of latex in the antimicrobial defense reactions of rubber tree is not well understood. Due to practical and technical limitations, conducting in-vivo study of the growth of the fungus on latex on natural wood substrate was not feasible. In this study, the alternate choice of using natural latex in artificial media for the bioassay was reinforced based on results from earlier published papers. We could not observe any fungistatic or fungicidic effect of latex in our in vitro experiments. On the contrary, fungal cultures developed faster on media supplemented with latex, indicating that some of its components might be metabolized by the fungus. The main component of natural latex is poly (cis-1,4-isoprene), a polymer highly resistant to microbial degradation. However, latex also contains a fraction of proteins, preventing latex from coagulation in planta. Cis-1,4-polyisoprene, a main constituent of latex rubber, is indeed hydrophobic 79 . In planta latex particles are surrounded by hydrophilic layer, which prevents them from the aggregation within latex-carrying vessel elements. It is however possible that the ability to degrade latex might be the most relevant at the initial stages of pathogen establishment, when it has to cross host tissues rich in latex during post-penetration into sapwood. It is also possible that Rubber trees might use latex to seal wounds in a similar way as conifer trees use oleoresin. However, once the fungus reaches the sapwood, the ability to degrade latex might become less relevant, as there is very little latex www.nature.com/scientificreports www.nature.com/scientificreports/ (if any) in the sapwood. Furthermore, in our transcriptome data, several gene transcripts encoding hydrophobin proteins were differentially expressed during growth on rubber wood and latex. It is possible as it has been documented in other studies that the hydrophobin facilitates the growth of fungal hyphae into the air from moist environments thereby facilitating host interactions as well as infectivity and invasive growth of pathogenic fungi 80 . We are however unable to speculate on the precise impact of the hydrophobin genes on the Rigidoporus cell wall and the consequent interaction, further studies will be required to prove it. We were not able to identify homologs of bacterial proteins involved in latex degradation (rubber oxygenase roxA 81 and latex-clearing protein lcp 82 ) in the genome of R. microporus. The analysis of RNA-seq data suggests that R. microporus might utilize proteinaceous components of the latex, as higher transcript abundance was found for several genes encoding proteases and transporter proteins. A number of genes encoding oxidative enzymes also were up-regulated, however, it remains unclear whether any of those can participate in rubber degradation, as available data indicate that dioxygenase activity is required for the cleavage of polyisoprene backbone 83 .
R. microporus is a fungal pathogen of considerable economic importance, and the availability of its genome sequence provides better opportunities for the understanding of its pathogenicity determinants and for the development of advanced control strategies. Our analysis contributed to the establishment of phylogenetic relationships of this species and confirmed its position within the order of Hymenochaetales. The analysis of the repertoire of PCW-degrading enzymes encoded in the genome of R. microporus indicated that, despite the lower numbers of lignin-degrading peroxidases, GT and GH-encoding genes compared with other white-rot fungi, R. microporus has the capacity for efficient wood degradation. It possesses a diverse set of SM genes, and the biological role of their corresponding products deserves further investigations. Being a white-rot fungus, members of the genus Rigidoporus are known to play major roles in nutrient and carbon cycling in tropical forest. It is therefore not surprising that R. microporus harbor a repertoire of a wide range of useful enzymes important for lignocellulose degradation with potential applications in bioenergy processing and utilization.
The availability of both the pathogen and host genome would help to facilitate the long-needed detailed studies on resistance research and host-related interactions as well as on the ecology and physiology of the pathogen. The genome has also provided much needed insight and clarification on systematics of the Hymenochaetales. Furthermore, the control and management of white rot disease of rubber in most tropical countries have been done without knowledge of the population genetics of the different isolates obtainable from each country and continent. Most management practices have been based on clearing and burning of infected roots and stumps. Such measures can reduce the disease incidence. Knowledge of population genetics of plant pathogens can facilitate breeding efforts to control plant diseases. There are only a few population genetic studies reported for R. microporus pathogen on rubber tree. These include studies in Asia on genetic variability and characterization 84,85 and somatic incompatibility 86 of the pathogen in different geographic areas in Thailand and Indonesia respectively. They are limited in scope, and there is therefore a need for a detailed study of the genomics, population biology, population genetics and host pathogen interaction of this economical important tropical tree pathogen. This can now be greatly facilitated by availability of the first genome sequence of this rubber tree pathogen. We expect that the availability of the genomic data will further stimulate studies on the biology of this fungal species and its role in wood decay and carbon cycling in nature.

Data availability
Genome assembly and annotations used in this study are available at the JGI fungal genome portal MycoCosm (http://jgi.doe.gov/fungi). The genome of R. microporus can be accessed at https://genome.jgi.doe.gov/Rigmic1/ Rigmic1.home.html. The genome assembly has been deposited at DDBJ/EMBL/GenBank under the following accession number PRJNA345706. The raw data from the transcriptome have been submitted to the National Center for Biotechnology Information under the following accession number PRJNA497786.