Genome sequencing and comparative genomics of the broad host-range pathogen Rhizoctonia solani AG8

Abstract

Rhizoctonia solani is a soil-borne basidiomycete fungus with a necrotrophic lifestyle which is classified into fourteen reproductively incompatible anastomosis groups (AGs). One of these, AG8, is a devastating pathogen causing bare patch of cereals, brassicas and legumes. R. solani is a multinucleate heterokaryon containing significant heterozygosity within a single cell. This complexity posed significant challenges for the assembly of its genome. We present a high quality genome assembly of R. solani AG8 and a manually curated set of 13,964 genes supported by RNA-seq. The AG8 genome assembly used novel methods to produce a haploid representation of its heterokaryotic state. The whole-genomes of AG8, the rice pathogen AG1-IA and the potato pathogen AG3 were observed to be syntenic and co-linear. Genes and functions putatively relevant to pathogenicity were highlighted by comparing AG8 to known pathogenicity genes, orthology databases spanning 197 phytopathogenic taxa and AG1-IA. We also observed SNP-level "hypermutation" of CpG dinucleotides to TpG between AG8 nuclei, with similarities to repeat-induced point mutation (RIP). Interestingly, gene-coding regions were widely affected along with repetitive DNA, which has not been previously observed for RIP in mononuclear fungi of the Pezizomycotina. The rate of heterozygous SNP mutations within this single isolate of AG8 was observed to be higher than SNP mutation rates observed across populations of most fungal species compared. Comparative analyses were combined to predict biological processes relevant to AG8 and 308 proteins with effector-like characteristics, forming a valuable resource for further study of this pathosystem. Predicted effector-like proteins had elevated levels of non-synonymous point mutations relative to synonymous mutations (dN/dS), suggesting that they may be under diversifying selection pressures. In addition, the distant relationship to sequenced necrotrophs of the Ascomycota suggests the R. solani genome sequence may prove to be a useful resource in future comparative analysis of plant pathogens.

Figure 1. A novel pipeline was employed to assemble the multinucleate, heterozygous genome of Rhizoctonia solani AG8.

A) Genomic DNA from multinucleate cells with variable nucleic copy numbers was prepared for next-generation sequencing (NGS) Illumina paired-end (B) and mate-paired (C) short-read libraries. The 3′ ends of read pairs from both (B) and (C) were tested for overlapping sequence, indicating short DNA fragment sizes. Overlapping pairs in mate-paired libraries (C) were discarded as these indicated paired-end contaminants which would lead to assembly errors. D) De novo assembly was performed combining the non-overlapping and overlapping paired-end read pairs that were merged into longer single-end reads. E) Redundant haplotypes where equivalent regions of the genome from multiple nuclei were present more than once in the assembly were merged into a single haplotype sequence. F) Non-overlapping mate-paired reads were used to build assembled sequences into larger scaffold sequences. Stretches of unknown bases (polyN) in the assembly were filled where possible (G) by alignment of genomic NGS reads to the assembly and regions predicted to contain tandem-duplication errors were corrected (H). Processes F, G and H were repeated for several rounds to ensure complete assembly. I) Minor assembly errors and the presence of RIP mutation between nuclei were corrected by substitution of the most dominant or pre-RIP allele. The final RIP-depleted, haploid consensus genome assembly (J) was manually annotated using a combination of RNA-seq and protein homology supporting evidence, producing a final dataset of 13,964 protein-coding genes (K).

Figure 2. RIP-like mutation was observed across repetitive and gene-encoding regions of the R. solani AG8 assembly.

A) Fluorescence micrograph of Rhizoctonia solani AG8 hyphae (stained with SYBR green) displaying multiple nuclei within a single cell. Nuclei appear as brightly fluorescent structures. Hyphal septa are indicated with arrows and the scale bar is equivalent to 20 µm. Prior to genome analysis, sequence variation between nuclei was unknown. B) Close-up view of the genomic region corresponding to actin gene RSAG8_00181, with short genomic sequence reads used in its assembly. This is representative of most genomic regions, in which constituent short reads exhibit two dominant haplotypes differentiated by low frequency SNP mutation. C) Percentage frequency matrix of SNP mutation type at heterozygous sites in the AG8 assembly. The majority were transition mutations between cytosine and thymine (reverse complement adenine and guanine). D) Frequency logos of the base composition of the sequences flanking heterozygous C↔T transitions in gene and repeat sequences, exhibiting a moderate bias for a 3′ guanine (i.e. CpG) in both. E) Distribution of genes, repeats and cytosine hypermutations across AG8 nuclear scaffolds of at least 100 kbp in length (scaffolds 1-76 and 78-117). All plot data in concentric rings are calculated within sequential 100 kbp windows, in order from the centre outwards: (i) G:C content (green, from 40 to 60%); (ii) percentage of 100 kbp window region covered by protein-coding genes (blue, from 0 to 100%); (iii) percent coverage of repetitive sequences (red, from 0 to 100%). The presence (black) or absence (white) of gene or repeat regions are also indicated directly below rings (ii) and (iii) respectively; (iv) frequency of heterozygous C↔T (and A↔G) polymorphisms (orange, 0 to 1000); (v) ratio of heterozygous C↔T (and A↔G) sites relative to all SNPs (orange, 0 to 100%).

Figure 3. Genome assembly sequence comparisons between R. solani AG8 and isolates from alternate anastomosis groups.

Dot-plots depict nucleotide sequence matches detected via MUMmer (nucmer) between the two largest scaffolds (both Scaffold_1) of R. solani AG8 and AG1-IA, as well as other homologous scaffolds from AG8, AG1-IA and AG3. Sequence alignments exhibit a predominantly co-linear, macrosyntenic configuration, however a small number of structural rearrangements can be observed between the larger scaffolds of AG8 and AG1-IA. Due to partial assembly of these genomes and thus short length of many scaffold sequences depicted here, only longer scaffolds have been labelled with their numbers along the x- and y-axes, however full details of alignments can be found in Supporting Table S3.

Figure 4. Summary of secreted proteins predicted by 3 different methods: SignalP, WolfPsort and Phobius.

Figure 5. Summary of R. solani AG8 genes assigned with CAZyme functional annotations.

An overall summary is presented for the 5 CAZyme categories: carbohydrate-binding molecules (CBMs), carbohydrate esterases (CEs), glycosyl transferases (GTs), glucoside hydrolases (GHs) and polysaccharide lyases (PLs). Individual summaries are also presented for each category, showing their most abundant CAZyme classes.

Figure 6. Percentage of heterozygous CpG↔TpG mutations not occurring in repetitive DNA versus distance from nearest repetitive DNA region.

Percentage values were calculated based on mutations contained within incremental distance ranges of 100