High-quality genome assembly-based and functional analyses reveal the pathogenesis mechanisms and evolutionary landscape of wheat sharp eyespot Rhizoctonia cerealis

Wheat (Triticum aestivum) is one of the most important staple crops. The necrotrophic binucleate fungus Rhizoctonia cerealis is the causal agent for the devastating disease wheat sharp eyespot and additional diseases of other agricultural crops and bioenergy plants. In this study, we present the rst high-quality genome assembly of R. cerealis Rc207, a highly aggressive strain isolated from wheat. The genome encodes expand and diverse sets of virulence-related proteins, especially secreted effectors, carbohydrate-active enzymes (CAZymes), metalloproteases, Cytochrome P450 (CYP450), and secondary metabolite-associated enzymes. Many of these genes, in particular those encoding secretory proteins and CYP450, showed markedly up-regulation during infection in wheat. Of 831 candidate secretory effectors, ten up-regulated secretory proteins, such as CAZymes, metalloproteases and antigens, were functionally validated as virulence factors required for the fungal infection in wheat. Further intra-species and inter-species comparative genomics analyses showed that repeat sequences, accounting for 17.87% of the genome, are the major driving force for the genome evolution, and frequently intraspecic gene duplication contributes to expansion of pathogenicity-related gene families. This is the rst genome-scale investigation elucidating the pathogenesis mechanisms and evolutionary landscape of R. cerealis. Our results provide essential tools for further development of effective disease control strategies. proles The current research rstly veried that the secretory tripeptidyl peptidase RcTP (Rc_02278.1) functions as a virulence factor of R. cerealis Rc207. Our previous research revealed that the secretory M43 domain-containing metalloprotease RcMEP1acted as virulence factor of R. cerealis 23 . In the maize anthracnose pathogen fungus Colletotrichum graminicola, the M36 domain-containing metalloprotease Cg was demonstrated to act as virulence factor through inhibiting host chitinases 31 . Here, our functional analyses demonstrated that the M36 domain-containing metalloprotease RcFL1 could induce plant cell-death and contribute to fungal infection by repressing the expression of the wheat chitinases and TaCERK, whose ortholog in Arabidopsis has been shown to be essential for chitin elicitor signaling 32 . The recent paper reported that VdPDA1, functioning as a secretory effector, contributes to virulence by preventing CERK-mediated chitin-triggered immunity during V. dahliae infection 32 . Thus, RcFL1 likely prevented the host chitin-triggering immunity, which shed light on a novel mechanism underlying the virulence roles of the kind of proteases in necrotrophic fungal pathogens.


Introduction
Wheat (Triticum aestivum L.) is the most-extensively cultivated staple crop (~ 17% of the total cultivated area) in the world 1 . It serves as a leading food for human consumption as an essential source of starch, proteins, vitamins, dietary ber, and phytochemicals 2 . Wheat crops worldwide are severely affected by sharp eyespot, a devastating culm/stem-base disease that is primarily caused by the necrotrophic and lamentous basidiomycete fungus Rhizoctonia cerealis van der Hoeven (teleomorph Ceratobasidium cereal Murray & Burpee) [3][4][5] . This disease can cause severe yield losses and a reduction in wheat quality 6,7 . In China, sharp eyespot has been a very highly economically important wheat disease for many decades 6,7 , with a total of 6.67-9.33 million hectares of wheat plants being affected by the disease every year 6-8 (http://www.agri.cn/V20/bchqb). R. cerealis has a broad range of host species and a necrotrophic lifestyle, surviving both in the soil and on infected crop residues 5,7−11 . Besides causing wheat sharp eyespot, R. cerealis is also responsible for sharp eyespot disease in other cereal crops (such as barley, oat, and rye), yellow patch in grasses, and root rot disease in cotton, potato, sugar beet and several species of legumes 4,5,8,10,11 . Hence, it is very important to control the infection of this pathogen and urgent to ensure food security and the quality of life across the globe. However, current management of wheat sharp eyespot is not effective due to a poor understanding of the pathogenic mechanisms of this pathogenic fungus.
The Rhizoctonia genus is a heterogeneous and very complex group of lamentous fungi of basidiomycete, and includes uninucleate, binucleate, and multinucleate Rhizoctonia species based on young cell nuclear number. According to reproductively incompatible anastomosis, the binucleate Rhizoctonia species can be divided into 7 anastomosis groups (AG A to AG Q), while the multinucleate Rhizoctonia species are classi ed into 14 distinct groups (AG1 to AG13, and AGBI) 9 . Among these, the Rhizoctonia anastomosis group AG D subgroup I (AG-DI) is responsible for sharp eyespot disease in wheat, and for yellow patch on turfgrasses 4,6,7,9 . In contrast to the multinucleate R. solani, R. cerealis contains narrower hyphae and two nuclei within a single hyphal cell, and is able to grow at a relatively slowarate 3,7,12,13 . Previous studies on R. cerealis mainly focused on the identi cation of the fungal pathogen and its life cycle, the symptoms and geographical distribution of the disease, and the genetic structure of populations [3][4][5][6][7][8][9][10][11][12][13] . The binucleate R. cerealis exists primarily as vegetative mycelia and sclerotia, does not form asexual spores, and its sexual stage is probably rare in nature 3,9,12 . Hence, it is especially di cult to establish effective methods that allow for the stable transformation and study pathogenesis mechanisms in R. cerealis. The whole genome assembly, using Illumina next-generation sequencing (NGS) or long-read sequencing technologies, provides an important tool for enhancing current knowledge on the original pathogenic mechanisms and evolution in pathogens [14][15][16][17][18] . However, no study addressing whole genome and transcriptomic sequencing has ever been reported for R. cerealis during infection of host plants. Furthermore, the current knowledge regarding pathogenesis and evolution mechanisms of R. cerealis remains very scarce.
In this study, we aimed to elucidate the pathogenesis and evolutionary mechanisms of the necrotrophic fungus R. cerealis, and the molecular basis of host-pathogen interactions. In order to achieve this, we employed both Oxford nanopore and Illumina NGS technologies to generate a high-quality whole-genome assembly for R. cerealis strain Rc207, a highly aggressive isolate of the wheat sharp eyespot in China. Subsequently, we observed the process of fungal infection on and within the cells of wheat leaf sheaths through scanning electron microscopy (SEM), and pro led gene expression patterns of the fungal pathogenic determinants during wheat infection. We identi ed 831 (including 29 novel) candidate secretory effectors with diverse activity, and validated the functions of ten up-regulated candidate secreted effector proteins, including: 5 carbohydrate-active enzymes (CAZymes), 1 metalloprotease, 1 tripeptidy peptidase, 2 antigens and 1 guanyl-speci c ribonuclease, which have all been con rmed as effectors or virulence factors. Most importantly, to uncover mechanisms explaining the adaptive genomic evolution of R. cerealis, we also completed the whole genome assembly of R. cerealis Rc301 (R0301, another strain is virulent to wheat) by using Illumina NGS and assembling technologies, then performed intraspeci c and interspeci c comparative genome analyses and identi ed abundant potential mobile chromosomes or plastic regions driven by repeat sequences. Beyond uncovering the evolutionary landscape of the R. cerealis genome, this study also reveals genome-scale insights onto the pathogenicity and adaptation mechanisms, and provides effective control strategies.

Results
2.1 A high-quality whole-genome assembly and annotation of R. cerealis Rc207 We generated 178,680 clean reads with mean length of 34.1 kb and N 50 length of 44.8 kb by using Oxford Nanopore long-read sequencing. After a de novo assembly by using the ltered and polished subreads of the Nanopore sequencing, and a further polishing/correcting by using ~ 32 million Illumina NGS reads (from six libraries), we obtained the nal high-quality R. cerealis Rc207 genome assembly consisted of 55 scaffolds with a total length of 56.36 Mb, a max scaffold length of 3.52 Mb, an N 50 scaffold length of 1.68 Mb, and a GC content of 48.63% (Table 1). Benchmarking Universal Single-Copy Orthologs (BUSCO) analysis revealed that the genome assembly contained 91.03% (264/290) complete BUSCOs, of which 90.53% (239/264) were single-copy (Table S1). Mapping with the Illumina short reads showed that 90.16% of the reads were properly mapped (Table S1).  (Table S4).
Also, we predicted 165 genes that belong to 15 secondary metabolite-associated gene clusters, including one non-ribosomal peptide synthetases (NRPS) type cluster, six terpene synthases type clusters, and eight NRPS-like type clusters (Table S5). All of these proteins may play signi cant roles in pathogenesis of R. cerealis, and might also be associated with the exclusive necrotrophic lifestyle and the adaptation of this pathogen to an unique ecological niche.

SEM and transcriptomic analysis during R. cerealis Rc207 infection in wheat
We used scanning electron microscopy to track the hyphae infection of R. cerealis Rc207 strain (on and) inside the leaf sheaths of the susceptible wheat cultivar Wenmai 6 at the tillering stage. At 18 hai, the hyphae started to pierce plant cell walls (Fig. 1A, Fig. S1). At 36 hai, the hyphae penetrated into the plant cell walls and spread into the cytoplasm of the infected wheat sheaths (Fig. 1A, Fig. S1). At 72 hai, the hyphae markedly proliferated on the surface of plant cells, and colonized and grew inside the invaded wheat cells (Fig. 1A). At the same time, small brown lesions were rst visible at the surface of the infected leaf sheaths. At 96 hai, the thriving fungal hyphae occupied the whole plant cells and destroyed the cell walls ( Fig. 1A), while the dark-brown lesions on the wheat leaf sheaths had expanded and continued to develop (Fig. S2). At 240 hai, the fungal hyphae massively proliferated inside the colonizing plant cells and continued to destroy the host cells, while the sharp eyespot symptoms on the inoculated wheat sheaths and stems became more severe (Fig. S2).  Fig. 1B). The corresponding proteins of the 912 genes that were signi cantly up-regulated during infection belonged to a total of 529 OrthoMCL-annotated ortholog groups and 40 paralog groups, but 58 proteins could not be grouped. When compared to the whole genome, we found that 13 ortholog groups were signi cantly over-represented among the 912 genes (Table S8). Nine of these groups contained the CAZymes of Glycoside Hydrolase (GH) families 3 (OG6_100201), 6 (OG6_109747), 10 (OG6_102896), 32 (OG6_101843), 43 (OG6_110274), 51 (OG6_109774), and 61 (now is Auxiliary Activity family 9 or AA9; OG6_118077), carbohydrate-binding module (CBM) family 5 (OG6_100121), and AA7 (OG6_113017). Moreover, three of the groups contained aromatic peroxygenases (OG6_118111), malate dehydrogenase (OG6_123068), and sterol 14demethylase (Cytochrome P450 family 51; OG6_103408). One of the groups contained stress responserelated proteins (OG6_111600). Almost the entire set of protein groups belonged to the predicted secretome (Table S8), indicating that signi cantly up-regulated secreted proteins may play important roles during the interactions between R. cerealis and wheat, and that these secreted proteins may be major pathogenesis determinants .

Secretory effectors in R. cerealis Rc207 contribute to the infection
Although previous research annotated the candidate effectors with a signal peptide-containing protein pipeline 17 , experimentally-validated effectors and virulence factors tend to be cysteine-rich (≥ 4) or upregulated secreted proteins 15,19−21 . The R. cerealis Rc207 secretome is comprised by 1,080 secreted proteins (Table S13). Transcriptome analysis showed that 300 genes coding for secreted proteins were signi cantly up-regulated during fungal infection in wheat (Table S14, Fig. 3A). Of these, we found that 177 and 105 genes coding even and odd cysteine-containing secreted proteins, respectively, were signi cantly up-regulated. Furthermore, the R. cerealis Rc207 secretome includes 755 cysteine-rich (the number of cysteine ≥ 4) secretory proteins, which likely function as candidate effectors (Fig. S8A). Taken together, a total of 831 candidate effectors, including 755 cysteine-rich secretory proteins and 282 upregulated secretory proteins, were identi ed from the R. cerealis Rc207 ssecretome and classi ed according to their functional annotation (Table S14). Among them, those containing even cysteines outnumbered those containing odd cysteine numbers.

Secretory metalloprotease RcFL1 is required for fungal infection in wheat
Compared to the closely-related fungus R. solani AG1 IA (307 protease-coding genes), the R. cerealis Rc207 genome contained more genes coding for proteases (461 genes), particularly richer in aspartic proteases, including pepsin A1As (44 vs. 23), cysteine proteases (99 vs. 64), metalloproteases (116 vs. 100), serine proteases such as prolyl aminopeptidase S33s (178 vs. 103), and threonine proteases (24 vs. 17) (Fig. S11, Table S16). Among the 116 metalloproteases (Fig. S12) (Fig. 4A). Our previous functional analyses showed that the up-regulated secretory RcMEP1 acted as virulence factor 23 . Hence, we further investigated functional role of the RcFL1 during fungal infection to wheat. Agrobacterium tumefaciens mediated transient expression assays in Nicotiana benthamiana leaves showed that the RcFL1 was able to secrete into the apoplasts and to trigger plant cell death, but that the signal peptide-deleting mutant lost both of these activities in planta (Fig. 4B). Furthermore, compared to His-TF (CK), the heterologously-expressed proteins His-TF-RcFL1 (expressing the full RcFL1 protein) and His-TF-RcFL1-M36 (expressing the M36 domain of RcFL1) were able to trigger necrosis and plant cell death on the in ltrated leaves of susceptible wheat cv. Wenmai 6 or of N. benthamiana (Fig. 4C, Fig. S14A, Fig. S14B). Notably, both the RcFL1 and RcFL1-M36 proteins were able to increase the fungal infection size after R. cerealis liquid mycelia inoculation on the wheat leaves ( Fig. 4D-4E). Additionally, the RcFL1 protein signi cantly repressed the expression of wheat defense genes encoding chitinases (TaChit3 and TaChitIV) and the receptor-like kinase CERK1 (TaCERK1) in the in ltrated leaves (Fig. 4F). These results demonstrated that the secretory metalloprotease RcFL1, acting as a virulence factor, is required for the fungal infection into wheat, and that both the signal peptide and the M36 domain are necessary to ensure the virulence role of RcFL1.
2.6 Repeat sequences and frequently-intraspecies gene duplication contribute to the genome evolution of R. cerealis and expansion of pathogenicity-related gene families A phylogenetic tree including R. cerealis Rc207, ten previously sequenced basidiomycotina fungi (Coprinopsis cinerea, Cryptococcus gattii, Laccaria bicolor, M. larici-populina, Phanerochaete carnosa, Postia placenta, Puccinia graminis, R. solani AG1 IA, R. solani AG8, and U. maydis), and the ascomycete F. graminearum as an outgroup, was constructed based on their core orthologs. The tree showed that R. cerealis Rc207 is more closely related to R. solani AG1 IA (and R. solani AG8) than to the other fungi in the basidiomycotina family. The divergence between R. cerealis Rc207 and R. solani AG1 IA/R. solani AG8 occurred approximately 140 million years ago (MYA; Fig. S15).
To explore the major mechanisms underlying the genomic evolution of R. cerealis, we completed the genome assembled of R. cerealis Rc301 (R0301, another strain is also virulent to wheat isolated from Nanjing, China) by using Illumina reads (Table S18), then performed intra-species gene collinearity analysis between R. cerealis Rc207 and R. cerealis Rc301, and inter-species gene collinearity analyses between R. cerealis and R. solani AG1 IA/R. solani AG8. These analyses showed there were more collinear gene clusters identi ed from the longer scaffolds of R. cerealis Rc207 genome assembly (nanopore sequencing), indicating that these scaffolds include conserved regions. In contrast, more R. cerealisspeci c the genes were present in the short scaffolds ( Fig. 5A-5B). In addition, we identi ed 3,010 intraspecies syntenic (IS) genes (20.86%) in 53 syntenic gene clusters within the R. cerealis Rc207 genome (Table S19). These IS genes exhibited a high frequency among the short scaffolds and a low frequency among the long scaffolds (Fig. 6), and the similar pattern was also observed in R. cerealis Rc301 genome (Fig. 5B). Interestingly, the distribution pattern of IS genes was similar to that of repeat sequences that usually accompany the genomic structures for a rapid evolution 16,24,25 . These ndings suggest that there may be abundant duplicated genes and gene clusters within the R. cerealis genome, and that these genes are frequently located at repeat sequences-rich genomic regions that may be undergoing rapid evolution.
According to the lengths of the scaffolds, from long to short, and to the frequency of IS genes, with 4.6%, 21.6% and 40.3%, we divided the scaffolds into three groups to perform preliminary comparisons. Speci cally, group1, group2 and group3 included the top ve longer scaffolds, the 13 medium-length scaffolds, and the remaining 37 short scaffolds, respectively ( Fig. 5B; Fig. 6, Table S21). In a manner consistent with the observations above on the repeat sequences, the IC genes, the secretome-and CAZymes-encoding genes and the species-speci c genes, the unexpressed and low-expressed genes also exhibited an increasing frequency trend from group1 to group3. In contrast, an opposite trend was observed for gene density, GC content, the frequency of conserved genes and that of genes with collinearity to Rc301 (Fig. 5B, Fig. 6; Table S20). In addition, the IC genes, the species-speci c genes, the unexpressed and low-expressed genes, and the repeat sequences, concentrated on the edge of scaffolds; while the conserved genes, the highly expressed genes, the majority of the secretome, and the genes with collinearity to RC301 concentrated on the central regions of scaffolds (Fig. 5B, Fig. 6A-C; Table S21). These analyses further revealed that repeat sequences likely drive a rapid evolution in speci c regions of the R. cerealis genome, which may be in the forms of mobile chromosomes or plastic regions, such as the edge of the chromosomes.

Discussion
In this study, using an Illumina NGS platform, we generated a draft genome assembly for the R. when compared to the Illumina NGS assembly. As the only sequenced plant fungal pathogen in the binucleate Rhizoctonia, the complete genome assembly of R. cerealis will provide the essential genomicsbased resource for further comparative genomics, including genomic features and evolution, functional genomics, and for uncovering the mechanisms behind pathogenicity, virulence variation and adaptation to this important pathogenic-fungus.
Based on the high-quality genome assembly, interspeci c and intraspeci c comparative genomic analyses revealed the core and plastic genomic scaffolds and regions in R. cerealis. In particular, the repeat sequence-rich scaffolds/regions contained a relatively higher number of secreted proteins-and CAZymes-encoding genes (validated or candidate virulence factors), more species-speci c genes and more unexpressed and low-expressed genes. These ndings suggest that these genes may undergo rapid evolution, maintain the pathogen unique genome and niche adaption, and are responsible for the variation in virulence in R. cerealis. Many repeat sequence-rich scaffolds/regions are part of the short scaffolds or distributed on the edge of scaffolds in the R. cerealis genome, which is consistent with the existence of mobile chromosomes and plastic chromosome regions previously reported in the genomes of some other fungal plant pathogens 16,24,26 . Repeat sequences likely drive a rapid evolution in R. cerealis speci c genomic regions and virulence variation between R. cerealis and R. solani or between the R. cerealis isolates. Furthermore, these comparative genomic analyses identi ed that 20.86% of R. cerealis genes formed 53 syntenic gene clusters, which included abundant virulence-related genes. Hence, frequently intraspeci c gene duplication may play a major mechanistic role explaining the expansion of several pathogenicity-related gene families in R. cerealis. Even though R. cerealis is a binucleate organism, we obtained a very low estimation of genome-wide heterozygosity (1.06%) based on k-mer analysis using the Illumina NGS reads. Thus, the majority of the short scaffolds, or candidate duplicated genomic regions, are not likely the assemblies of the second copies of different allelic fragments.
The high-quality genome assembly presented here reveals that the number of CAZymes, secreted proteins, PHI factors, proteases including metalloproteases, CYP450 and secondary metabolism enzymes present in the wheat sharp eyespot pathogen R. cerealis Rc207 markedly exceeds those observed in the rice sheath blight pathogen R. solani AG1 IA 14  Compared to R. solani AG1 IA, R. cerealis Rc207 appears to require more proteases. The current research rstly veri ed that the secretory tripeptidyl peptidase RcTP (Rc_02278.1) functions as a virulence factor of R. cerealis Rc207. Our previous research revealed that the secretory M43 domain-containing metalloprotease RcMEP1acted as virulence factor of R. cerealis 23 . In the maize anthracnose pathogen fungus Colletotrichum graminicola, the M36 domain-containing metalloprotease Cg was demonstrated to act as virulence factor through inhibiting host chitinases 31 . Here, our functional analyses demonstrated that the M36 domain-containing metalloprotease RcFL1 could induce plant cell-death and contribute to fungal infection by repressing the expression of the wheat chitinases and TaCERK, whose ortholog in Arabidopsis has been shown to be essential for chitin elicitor signaling 32 . The recent paper reported that VdPDA1, functioning as a secretory effector, contributes to virulence by preventing CERKmediated chitin-triggered immunity during V. dahliae infection 32 . Thus, RcFL1 likely prevented the host chitin-triggering immunity, which shed light on a novel mechanism underlying the virulence roles of the kind of proteases in necrotrophic fungal pathogens.
Moreover, a total of 831 diverse candidate effectors (up-regulated, cysteine-rich secretory proteins) were identi ed from the R. cerealis Rc207 secretome. Interestingly we found 29 novel effectors, such as tripeptidyl peptidase, antigen, guanyl-speci c ribonuclease, cytochrome b2, and one signal peptidecontaining isochorismatase. However, in the R. cerealis Rc207 genome, there are no RXLR-and Crinklertype effectors widely present in the oomycete species 33 , suggesting a signi cant difference in virulence mechanisms between R. cerealis Rc207 and oomycete species. Previous studies reported that genes coding effectors were often up-regulated during host infection and required for plant cell-death and pathogenicity in pathogens 14,21,31,34 . Here, our functional experiments veri ed that in the necroptrophic R. cerealis Rc207, 10 up-regulated, or cysteine-rich secreted candidate effectors, including RcGH5, RcGH6-1, RcGH6-2, RcGH28, RcAA9, RcFL1, RcTP, RcOV16-1, RcOV16-2 and RcRNase act as virulence factors through the induction of plant cell death and the promotion of infection in wheat. To our knowledge, guanyl-speci c ribonuclease, tripeptidyl peptidase and antigen (carboxypeptidase Y inhibitor) have been, for the rst time, con rmed as virulence factors involved in pathogen-host plant interaction. The antigen RcOV16 belongs to the phosphatidyl ethanolamine-binding protein (PEBP) family likely through the inhibition of serine proteinases 35 , which may be widely useful in crop protection and biotechnology 36 .
Even though some resistant quantitative trait loci were identi ed in wheat, there are no crop cultivars with high resistance to R. cerealis. The important virulence factors validated here, including RcMEP6, RcGH28, RcGH6-1, RcOV16, and RcAA9, may be knocked-out/down in host plants by host-induced gene silencing in order to improve R. cerealis-resistance in wheat and other economically important cereal crops and bioenergy plants.
In conclusion, this study reports a high-quality whole-genome assembly of the soil-borne basidiomycete pathogen fungi R. cerealis and lls gaps in our knowledge of plant pathogen biology. In addition to providing the rst binucleate Rhizoctonia genome assembly, our study focused on a genome-wide pathogenic mechanisms and functional assay, along with comparative genomic and genome evolution analyses. This study undoubtedly deepens our understanding on the necrotrophic-plant interactions and provides valuable information and resources about the genomic features, adaptation strategies, fungal taxa, evolution, and pathogenicity mechanisms of this fungal pathogen. The novel effectors or virulence factors identi ed enrich the pathways associated with pathogenesis and fungal pathogen-plant interactions. Moreover, they constitute a vital tool for the study of PHI mechanisms and for molecular breeding. The strategy for impairing the important virulence factors validated, i.e. metalloproteases, cytochrome C oxidase, antigen and peptidase inhibitors, may be applied to control sharp eyespot of cereals including wheat and even diverse diseases affecting host plants.

Materials And Methods
R. cerealis strain, culture conditions and wheat infection. The binucleate fungus R. cerealis strain Rc207 is a highly aggressive strain with strong virulence in wheat crops from Northern-China, and was collected from the wheat eld of Taian, Shandong. R. cerealis Rc301 (R0301), another isolate is also virulent to wheat isolated from Nanjing, Jiangsu,China, was kindly provided by Prof. Huigu Chen. The strains were grown in a potato dextrose broth medium at 25 ℃ for 10 days, in the dark and with vigorous shaking (100 r.p.m.), and was subsequently washed with sterile H 2 O, frozen in liquid N 2 and freeze dried for genome sequencing.
The wheat cultivar Wenmai6 is highly susceptible to R. cerealis infection. Wheat plants were grown in a greenhouse under a cycle of 13-h light (~ 25 • C) and 11-h dark (~ 15 • C). The R. cerealis isolate Rc207 mycelia were inoculated between second base sheaths and stems of wheat plants at their tillering stage using the toothpick inoculation method. Wheat sheaths at ve different R. cerealis Rc207 infection time points (18,36,72,96, and 240 hai) were sampled and used for observation by scanning electron microscope, and for RNA-seq analyses.
NGS genome sequencing and assembly. Genomic DNA was extracted with a modi ed CTAB method. The harvested DNA was detected by agarose gel electrophoresis and quanti ed using Qubit. Six Illumina NGS libraries, including one 500 bp, one 2,193 bp, two 6,111 bp, and two 10,612 bp, were constructed and sequenced using the Illumina Hiseq 2500 technology with a PE125 strategy. Whole-genome sequencing and assembly were performed at the Beijing Novogene Bioinformatics Technology Co., Ltd.
Evolution analysis. A phylogenetic tree, including Rc207 and eleven other fungal species (C. cinerea, C. gattii, F. graminearum, L. bicolor, M. larici-populina, P. carnosa, P. placenta, P. graminis, R. solani AG1 IA, R. solani AG8, and U. maydis), was constructed based on conserved/core orthologs using a maximum likelihood model in TreeBest at the Beijing Novogene Bioinformatics Technology Co., Ltd. The divergence times were estimated based on the average time observed in basidiomycetes 37 .
Oxford nanopore sequencing and assembly. The genomic DNA sample of R. cerealis Rc207 was puri ed by Oxford nanopore tech. (ONT) and sequenced and assembled at Beijing Biomarker Technologies. The Pilon 38 software was used to correct the assembled genome with second-generation data in order to obtain the nal genome with higher accuracy. The nal R. cerealis Rc207 assembly was generated using two pieces of evidence: 1) the evaluation of Illumina NGS data returns ratio and 2) assessment of the integrity of the fungal genome assembly using the BUSCO v2.0 software. De novo prediction, homologous protein prediction and transcriptomic prediction was used to predict gene structure, after which the three predictions were integrated. The predicted genes were blasted to different databases, including KOG, KEGG, Swiss-PROt, TrEMBL and NCBI-nr, in order to obtain the gene function annotation.
PHI and FVF were identi ed with an E-value less than 1e − 50 and a minimal alignment length percentage larger than 40%. SignalP v5.0 39 and TMHMM v2.0 40 were used to detect protein sequences with signal peptides and transmembrane helices, respectively. Proteins with signal peptide and without transmembrane domains were identi ed as secreted proteins. Secondary metabolite-associated gene clusters were predicted using the antiSMA0 software 41 . Proteases were predicted using the MEROPS database (release 12.3) 42 .
Comparative genomic analysis. A pairwise gene synteny comparison between genome assemblies was identi ed using LAST and the MCScanX software (Python version) 43 using predicted proteins as input.
The gene synteny regions were visualized as dot plots and a macro-synteny plot using the JCVI graphics functions. Genome scaffolds, gene density and other genomic features using the synteny regions as input were plotted using Circos v0.69-8 44 .
RNA-sequencing and transcriptomic analyses. Fungal complementary DNA libraries were constructed from the R. cerealis Rc207 strain by infecting healthy wheat leaf sheaths for 18, 36, 72, 96 and 240 hours, and from in vitro mycelia stage. The libraries were sequenced on an Illumina Hiseq 2500 platform and 125 bp paired-end reads were generated in the Beijing Novogene Bioinformatics Technology Co., Ltd. The RNA expression analysis was based on the predicted genes of R. cerealis Rc207. The index was built and clean reads were aligned to the reference genome using HISAT2 v2.2.1 45 . The StringTie v1.3.5 46 software was used to construct and identify both known and novel transcripts from the HISAT2 alignment results.
The DESeq2 v1.30.0 47 software was used to perform read count normalization and differential gene expression (DGE) analysis using fold-change greater than or equal to 2.0 and a FDR P-value < 0.05.
Quantitative RT-PCR assay. qRT-PCR reactions were performed using a SYBR Premix Ex Taq kit (TaKaRa, Japan) in an ABI 7500 real time PCR system/instruction (Applied Biosystems, USA). The qRT-PCR data were analyzed using the comparative 2 −ΔΔCT method. The R. cerealis actin gene RcActin was used as an internal reference for testing the fungal genes. Each treatment included three independent technical replicates. The primer pairs used in this section are listed in Table S21.
Functional validation of candidate effectors. Among the 831 candidate effectors, we selected 10 candidate virulence genes showing up-regulation for functional veri cation. To test the cell deathinduction of the candidate effectors or virulence factors, these proteins were heterologously expressed in E. coli Transetta (DE3) and puri ed, and then in ltrated into leaves of wheat or of N. benthamiana 14,30 . To test their infection-promotion activity, Rc207 liquid mycelia were inoculated onto the wheat leaves in ltrated with the heterologously expressed proteins for 6 h and observed/photographed for 3 d 48 . A. tumefaciens mediated transient expression of RcFL1 and RcRNase in N. benthamiana leaves, and plant cell death-induction assays were performed as described by Ma et al. 48 and Yang et al. 34 . The GFP localization signals were assayed and photographed for 2 d post agro-in ltration under a confocal microscope (Zeiss LSM 700, Heidenheim, Germany).