The genome sequence of the fungal pathogen Fusarium virguliforme that causes sudden death syndrome in soybean.

UNLABELLED: Fusarium virguliforme causes sudden death syndrome (SDS) of soybean, a disease of serious concern throughout most of the soybean producing regions of the world. Despite the global importance, little is known about the pathogenesis mechanisms of F. virguliforme. Thus, we applied Next-Generation DNA Sequencing to reveal the draft F. virguliforme genome sequence and identified putative pathogenicity genes to facilitate discovering the mechanisms used by the pathogen to cause this disease. METHODOLOGY/PRINCIPAL
FINDINGS: We have generated the draft genome sequence of F. virguliforme by conducting whole-genome shotgun sequencing on a 454 GS-FLX Titanium sequencer. Initially, single-end reads of a 400-bp shotgun library were assembled using the PCAP program. Paired end sequences from 3 and 20 Kb DNA fragments and approximately 100 Kb inserts of 1,400 BAC clones were used to generate the assembled genome. The assembled genome sequence was 51 Mb. The N50 scaffold number was 11 with an N50 Scaffold length of 1,263 Kb. The AUGUSTUS gene prediction program predicted 14,845 putative genes, which were annotated with Pfam and GO databases. Gene distributions were uniform in all but one of the major scaffolds. Phylogenic analyses revealed that F. virguliforme was closely related to the pea pathogen, Nectria haematococca. Of the 14,845 F. virguliforme genes, 11,043 were conserved among five Fusarium species: F. virguliforme, F. graminearum, F. verticillioides, F. oxysporum and N. haematococca; and 1,332 F. virguliforme-specific genes, which may include pathogenicity genes. Additionally, searches for candidate F. virguliforme pathogenicity genes using gene sequences of the pathogen-host interaction database identified 358 genes.
CONCLUSIONS: The F. virguliforme genome sequence and putative pathogenicity genes presented here will facilitate identification of pathogenicity mechanisms involved in SDS development. Together, these resources will expedite our efforts towards discovering pathogenicity mechanisms in F. virguliforme. This will ultimately lead to improvement of SDS resistance in soybean.

Introduction

Crop plants encounter diverse fungal pathogens that cause a wide array of diseases and severely reduce yields. The fungal genus Fusarium is comprised of highly destructive pathogens that reduce crop productivity, contaminate harvested grains with mycotoxins, and in severe instances, cause crop failures that result in famines. Thus, effective management practices to control Fusarium pathogens are urgently needed to feed the world's rapidly growing human population. Biotechnological approaches based on knowledge gained from studying plant-fungal interactions hold significant promise to provide novel disease control strategies [1]. To facilitate genetic studies of the interactions between Fusarium fungi and crop species, genome sequences of four Fusarium species are currently available, viz., Fusarium oxysporum, F. graminearum, F. verticillioides, and Nectria haematococca (F. solani) [2], [3].

F. virguliforme is a serious, yet comparatively understudied, fungal pathogen that causes sudden death syndrome (SDS) in soybean. The pathogen causes root necrosis and rot, as well as vascular discoloration of roots and stems. Root infection is often accompanied by foliar symptoms (foliar SDS), characterized initially by interveinal chlorosis followed by necrosis, and in severe cases, flower and pod abscission [4]. Interestingly, F. virguliforme has never been isolated from symptomatic foliar tissues, which strongly suggests that foliar symptoms result from translocated toxins produced in infected roots [5]. Severe yield losses are commonly associated with expression of foliar SDS symptoms. The major toxin that causes foliar SDS is a small acidic protein [6]–[8]. The toxin requires light to initiate foliar SDS symptoms [6], [9].

SDS was first detected in Arkansas in 1971, and has now spread to all soybean growing areas of the United States [6]. The disease is caused by two morphologically and phylogenetically distinct species within the Fusarium solani species complex, F. tucamaniae and F. virguliforme [10]. Populations of F. tucamaniae, which causes SDS in Argentina and Brazil, possess two mating types and thus possibly undergo sexual reproduction, whereas a sexual reproductive stage is most likely absent among populations of F. virguliforme [10], [11].

In this study, we sequenced the genome of the F. virguliforme Mont-1 strain by applying a shotgun sequencing approach. As expected, the F. virguliforme genome revealed high sequence identity to the genome sequences of previously sequenced Fusarium species [12]. To facilitate identification of candidate genes involved in SDS development, we searched the F. virguliforme genome with the pathogen-host interactions (PHI) sequence database (http://www.phi-base.org/) composed of experimentally verified pathogenicity, virulence and effector proteins from bacteria, fungi and oomycete pathogens that infect a wide range of hosts [13]. This approach identified 358 candidate pathogenicity genes in the F. virguliforme genome. In a parallel approach, we annotated 1,332 F. virguliforme-specific genes to identify candidate pathogenicity genes. The F. virguliforme genome sequence is available through the NCBI database DDBJ/EMBL/GenBank under the project ID (PID) 63281 and accession AEYB01000000 and can be viewed through a GMOD Generic Genome Browser (GBrowse) available at http://fvgbrowse.agron.iastate.edu.

The F. virguliforme genome sequence and putative pathogenicity genes presented here will facilitate the identification of pathogenicity mechanisms involved in SDS development and ultimately lead to a better management of SDS in soybean.

Materials and Methods

F. virguliforme isolate and DNA isolation

The sequenced F. virguliforme Mont-1 isolate, virulent to the soybean cultivar Essex, was produced from a single conidium. The genomic DNA was isolated from germinating conidia with a published genomic DNA isolation protocol [14]

Genome sequencing and assembly

F. virguliforme Mont-1 DNA was sequenced in a 454 GS-FLX Titanium sequencing platform by SeqWright Inc. (Houston, TX). Three types of sequencing runs were conducted: (i) shotgun sequencing of ∼400 bp DNA fragments, (ii) sequencing of ∼3 kb paired-ends, and (iii) sequencing of ∼20 kb paired-ends. In collaboration with Lucigen, Inc. (Middleton, WI), a BAC library carrying approximately 100 Kb sheared DNA fragments [15] was constructed and both ends of 1,402 BAC clones were sequenced. Considering the use of both Sanger's dideoxy- and 454-sequence data in this study, we applied PCAP assembler software that can assemble both kinds of data [16]. Transcripts of F. virguliforme germinating conidia and mycelia were sequenced in an Illumina/Solexa genome Analyzer II (GAII) at the Iowa State University DNA Facility.

Comparison of PCAP and Newbler assemblies

Newbler is a genome assembly program for 454 data. To compare the effectiveness of PCAP in assembling 454 data, we assembled libraries of 454-single reads and −3 kb paired-end reads using Newbler. The consensus sequences of each assembly derived either by PCAP or Newbler were assessed by mapping the assembled sequences of Illumina paired-end reads of the F. virguliforme Clinton-1B isolate onto the PCAP- and Newbler-derived assemblies using the Bowtie2 program [17] and calling single-nucleotide polymorphisms (SNPs) with SAMtools [18]. SNP rates for both assemblies were calculated.

Gene prediction and annotation

The genes of the assembled F. virguliforme genome were predicted with the AUGUSTUS gene prediction program with options set for (i) F. graminearum, (ii) coding sequence and (iii) GFF [19]. The predicted genes were used as a reference set for mapping RNA sequences (http://fvgbrowse.agron.iastate.edu/) using the Bowtie program [20].

The preliminary annotation of the F. virguliforme genome was conducted using Pfam domain search and information was incorporated at the http://fvgbrowse.agron.iastate.edu/ genome browser. The Pfam database models were downloaded (http://Pfam.sanger.ac.uk/). The genome was annotated by conducting hmmscan search (HMMER 3.0, http://hmmer.org/). The BLAST2GO analysis was conducted at the http://www.blast2go.org/start_Blast2GO. The predicted F. virguliforme coding sequences were searched for identical sequences by conducting BLASTX search. A cut-off, E≤10−10 was used for BLASTX and annotation.

Syntenic analysis between Nectria haematococca and Fusarium virguliforme

The homologous regions between N. haematococca chromosomes and F. virguliforme genomes were identified using MAUVE software [21]. Homologous co-ordinates were identified using NUCMER software [22] and matching F. virguliforme scaffolds with N. haematococca chromosomes was identified and visualized using MAUVE program. The chromosomes graph was visualized for locally collinear blocks (LCBs) with weight close to 5000 [21].

Comparison and visualization of Pfam

The comparison of F. virguliforme (Scaffold 1) with the genome sequences of four Fusarium spp. was conducted using MUMmer genome comparison tool, and dot-pot was created using Mummer plot [22]. We used PROmer program [22], which aligns translated nucleotide sequences (six possible reading frames) with a filter of 100 nts to remove the noise. The visualization of Pfam annotation (heat map) was conducted using the multi-experiment visualization (MeV) tool [23]. We normalized the number of Pfam domain hits in an organism by dividing Pfam domain hits with total number of genes of that organism and then corrected with the respective SDs for visualization of the significant variation of a particular domain among species.

Phylogenetic analysis of F. virguliforme

F. virguliforme and four Fusarium genomes along with M. grisea, N. crassa, A. nidulans, R. oryzae, P. blakesleeanus, U. maydis and the two oomycetes, P. infestans and P. sojae, were considered for phylogenetic analysis, which was conducted using 10 highly conserved (E≤10−30) single copy proteins; viz., g242, g255, g270, g326, g330, g364, g375, g380, g422 and g426 (http://fvgbrowse.agron.iastate.edu/). A maximum-likelihood tree was generated using the PHYML tool [24] (WAG model) with 1,000 bootstraps.

Unique gene analysis

BLAST program (BLASTP) with a significant cut-off level, E≤10−9, was used to identify proteins common to all five Fusarium species or proteins that are unique to individual Fusarium species. Unique proteins (N) were calculated as N = X-Y-Z; where, X is the total number of proteins in a particular organism, Y is the proteins common to all Fusarium species, and Z is the number of proteins common to at least two Fusarium species.

Analyses of the F. virguliforme genome for pathogenicity proteins

The F. virguliforme proteins were interrogated with the pathogenicity proteins of the pathogen-host interaction (PHI) database to identify candidate F. virguliforme pathogenicity proteins. PHI protein sequences were downloaded and interrogated with the predicted F. virguliforme proteins (BLAST locally with a cutoff, E≤10−9) to identify candidate F. virguliforme pathogenicity proteins. The selected sequences were reanalyzed to eliminate any false positives.

Analyses of the F. virguliforme genome for secretory proteins

To identify the putative F. virguliforme secretory proteins, we analyzed the genome using SignalP program. The SignalP consists of two different predictors based on neural network and hidden Markov model algorithms Method. We used hidden Markov model algorithms and >0.9 probability value to identify the probable candidate secretory proteins.

Results and Discussion

The F. virguliforme Mont-1 isolate used in this study was propagated from a single conidium and was confirmed to be virulent on soybean. Whole-genome shotgun sequencing was conducted on a 454-GS-FLX Titanium platform. Initially, single-end reads of a 400-bp shotgun library were assembled using the PCAP genome assembly program [16]. The consensus genome sequence was determined from raw data with an average of 20-fold genome coverage. To facilitate assembly of the single read sequences into larger contigs, sequences of paired ends of approximately 3 and 20 Kb DNA molecules were obtained through sequencing on a 454 GS-FLX Titanium sequencer. In addition to paired-end sequences of random DNA fragments, sequences of both ends of inserts from 1,400 BAC clones with an average size 100 kb were obtained to support the assembly of shotgun sequences. The assembled genome sequence is 51 Mb (Table 1) with an N50 scaffold number of 11, an N50 scaffold length of 1,263 Kb and an N50 contig length of 73 Kb. The 1,386 scaffolds, which include 23 major scaffolds (0.5 to 5 Mb) and 1,363 (1 to 499 Kb) minor scaffolds, represent the entire 51 Mb genome sequence. The 51 Mb F. virguliforme genome sequence obtained in this study was comparable in size to the sequenced genomes of F. graminearum (36 Mb), F. verticillioides (42 Mb), Nectria haematococca (54 Mb) and F. oxysporum (60 Mb).

Table 1

General assembly statistics of the draft Fusarium virguliforme genome.

Number of contigs	3,098
Average length of contigs	16,284 bases
Total length of contigs	50,448,805 bases
Contig size N50	72,747 bases
N50 number of contigs	200
Number of scaffolds	1,386
Average length of scaffolds	36,757 bases
Total length of scaffolds	50,994,238 bases
Scaffold size N50	1,263,176 bases
N50 number of scaffolds	11
Total coverage	20×
Predicted genes	14,845

Comparison of Newbler and PCAP assemblies

In assembling the genome sequence, we applied PCAP assembler software because this program can assemble sequences generated by both Sanger's dideoxy and pyrosequencing technologies, applied in this study. PCAP was originally developed for Sanger's dideoxy ABI 3730 reads. For assembling 454 sequence data, Newbler program was developed. Each assembly has its unique features. For example, PCAP produces support information from each type of read pairs for each region of every scaffold. The support information is useful in estimating the likelihood that a particular region of a scaffold is accurate. In addition, PCAP produces candidate SNPs with their consensus alignment columns and locations in scaffolds.

We compared the quality of the assembly obtained by PCAP with that of the assembly generated by Newbler by computing the SNP rates of these assemblies with that generated for the Illumina paired-end reads of the F. virguliforme Clinton-1B isolate. The SNP rate of the Newbler assembly with Clinton-1B reads was 1 SNP in 10,000 bp, whereas that for the PCAP assembly with Clinton-1B reads was 1.6 SNP in 10,000 bp. The low SNP rates suggest that both assemblies are equally effective in assembling 454 sequence data. The Newbler assembly is slightly more accurate in generating consensus sequences probably because it uses more trace information than PCAP, which uses only bases and their quality values.

F. virguliforme is close relative of the pea pathogen, N. haematococca

To determine the evolutionary relationship of F. virguliforme with F. graminearum, F. verticillioides, N. haematococca and F. oxysporum, the largest scaffold (scaffold 1; 5.05 Mb) of the F. virguliforme genome was aligned with the genome sequences of these four Fusarium spp. F. virguliforme scaffold 1 showed different levels of conservation with the genomes of F. verticillioides (scaffold 3.1), F. oxysporum (scaffolds 2.1), F. graminearum (scaffold 3.1) and N. haematococca (scaffold Sca.1). The highest synteny was observed between F. virguliforme and N. haematococca (Figure S1) suggesting that from the evolutionary point of view, F. virguliforme is closet to N. haematococca. We, therefore, further investigated the local collinear blocks between the F. virguliforme Mont1 genome and N. haematococca chromosomes. Since the F. virguliforme sequenced genome has not been assigned to chromosomes, we identified the N. haematococca chromosome-specific F. virguliforme scaffolds using the NUCMER program [20]. The N. haematococca chromosome-specific F. virguliforme scaffolds were visualized using the MAUVE program [19]. Large F. virguliforme scaffolds, bigger than 5 kb, were mapped to the N. haematococca chromosomes and regions showing synteny are listed in Table S1 and Figure 1.

Figure 1

Synteny of Fusarium virguliforme Mont1 sequences to the Nectria haematococca chromosomal sequences.

The colored blocks show the alignment of F. virguliforme sequences to the sequences of the N. haematococca chromosomes. Blocks below the central line indicate the regions that align in the reverse complement orientation.

The contiguously colored regions are local collinear blocks (LCBs); i.e., regions without rearrangement of homologous backbone sequence [19]. LCBs below a genome's central line are in the reverse complement orientation relative to the reference genome. The highest number of LCBs was found in the N. haematococca chromosome 3 (56 LCBs) followed by chromosome 1 (48 LCBs). We did not identify any F. virguliforme scaffolds specific to N. haematococca chromosome 16.

The lack of greater levels of synteny throughout the genome may be due to rearrangement following separation of the two species from a common progenitor species. This led to development of mosaic patterns, unique to each species with least conservation between Fusarium species that are distantly related. Additionally, the high level of genomic variation likely stemmed in part from evolution of repeat sequences, low complexity sequences, or repeat-induced point mutation (RIP; [3]). These mechanisms have been reported to be the cause of genetic variations and source of genomic instability in other fungi [23]. The proportion of repeat sequences was determined by comparing the assembled F. virguliforme genome sequence to itself using the DDS2 program [25]. It is estimated that about 18% of the F. virguliforme genome is composed of repeat sequences, most of which contain low GC contents.

Gene content and organization of genes in the F. virguliforme genome

To predict the number of genes in the F. virguliforme genome we analyzed the genome using the AUGUSTUS gene prediction program by setting the species option to F. graminearum species [19]. It is predicted that the genome contains 14,845 genes. This number is very close to the predicted gene numbers for the genomes of F. graminearum (13,332), F. verticillioides (14,179), F. oxysporum f. sp. lycopersici (17,735) , and N. haematococca (15,707). The average G + C content of the F. virguliforme coding regions was 49%. Transcripts from germinating conidia and mycelia of F. virguliforme were sequenced on an Illumina Genome Analyzer [26] and were aligned to the CDS sequences of the predicted genes for evidence of expression by using Bowtie program [20]. Of 14,845 predicted genes, 13,375 (90%) were expressed in germinating conidia, 13,281 (89%) in mycelia; and 14,070 (95%) showed transcripts at least in either or both conidia and mycelia.

Overall, gene density was lower in the F. virguliforme genome compared to other Fusarium species (Table S2). Gene density was ∼3 genes/10 Kb throughout most of the F. virguliforme genome except in Scaffold 19, which contains three-fold fewer genes (∼1 gene/10 Kb) (Figure 2). The G + C content of the coding regions of the Scaffold 19 is only 28% (Table S2), as compared to approximately 50% in the other scaffolds. A low G + C content in Scaffold 19 is due to accumulation of repeat sequences and could be the reason for a lower gene density in this genomic region. The G + C content is generally uniform among genes within a species [27], [28] and varies slightly among the Fusarium genomes (Table S3). The uniqueness of Scaffold 19 for gene density may also suggest a possible horizontal transfer of this genomic region from another species. Further studies will be required to determine if this is the case.

Figure 2

Non-uniform gene density in the F. virguliforme genome.

Gene density of over 23 major scaffolds of the F. virguliforme genome is presented. Size of individual scaffold is presented in blue histograms covering 36.2 Mb (scale on the left) of the total 51 Mb genome. Gene density (number of genes/10 Kb) is presented with the green line (scale of the right).

Annotation of the F. virguliforme genes

Predicted F. virguliforme genes were annotated by using Pfam [29] and BLAST2GO annotation protocols [30]. Annotations based on the Pfam database (e≤10−9) assigned functions to 78% of the predicted F. virguliforme proteins. The distribution of predicted proteins among functional classes was similar to that of F. graminearum , and was dominated by the following categories: MFS 1 (Major Facilitator Superfamily) (523), Zn_clus Fungal Zn(2)-Cys(6) binuclear cluster domain (396), Fungal trans (Fungal specific transcription factor domain) (311) and Sugar tr (Sugar and other transporters) domains (266) (Figures 3, S2; Table S4). The Pfam annotations of Fusarium relatives were analyzed with the multi-experiment viewer tool; and the annotation heat-map was generated using all Pfam domains (up to 100 hits). The standard deviation of genome Pfam domain was generated and used to divide with the number of Pfam hits in each respective genome to normalize the data (Figure 3; Table S5). This analysis revealed that many abundant domains were present in comparable proportions among the Fusarium genomes; e.g., Major Facilitator Superfamily (MFS 1), Sugar (and other) transporter (Sugar_tr), KR domain (KR), NAD dependent epimerase/dehydratase family (Epimerase), and short chain dehydrogenase (adh_short). On the other hand, domains such as protein tyrosine kinase (Pkinase_Tyr), ankyrin repeat (Ank), and heterokaryon incompatibility protein (HET) domains were significantly higher in number in the F. virguliforme genome [31], [32]. These protein domains have been reported to be involved in pathogenicity and could be important in SDS development. GO annotation of the 14,845 genes were conducted using BLASTX program (e≤10−10). A functional annotation was assigned to 14,810 (99.76%) genes and 7,954 (53.58%) of these were grouped into 16 functional categories (Figure S2). The majority of the genes were classified into metabolic processes followed by cellular processes, suggesting that most of the genes are required for basal metabolism and housekeeping functions. The Pfam annotated and un-annotated genes are presented in Gbrowse (http://fvgbrowse.agron.iastate.edu). The minimum length of predicted F. virguliforme coding sequences was 201 nucleotides and the average length was 1,482 nucleotides.

Figure 3

Heat-map depicting the Pfam domains in all Fusarium species.

F. virguliforme genome is rich in Pkinase_Tyr (Protein tyrosine kinase), Ank (Ankyrin repeat), and HET (Heterokaryon incompatibility proteins) as compared to the other Fusarium genomes.

F. virguliforme proteins conserved across species

To determine the number of F. virguliforme genes that are broadly conserved across diverse taxa, we conducted BLASTP analyses across the genomes of 25 organisms (Table S6) [33] (e≤10−9) using Standalone BLAST (version 2.2.15) search tools (http://BLAST.ncbi.nlm.nih.gov/) and results are presented in Figure S3 and Table S7. The most closely related organisms included N. haematococca, F. oxysporum, F. graminearum, F. verticillioides, Neurospora crassa, Aspergillus nidulans, Ustilago maydis, Phycomyces blakesleeanus, Rhizopus oryzae, and Saccharomyces cerevisiae. More distantly related organisms included Danio rerio, Glycine max, Arabidopsis thaliana, Homo sapiens, Phytophthora sojae, Phytophthora infestans, Dictyostelium discoideum, Zea mays, Oryza sativa ssp. japonica, Drosophila melanogaster, Caenorhabditis elegans, Rhizobium leguminosarum, Pseudomonas syringae, Agrobacterium tumefaciens and Escherichia coli. More than 80% of the F. virguliforme proteins shared high identity with proteins of the closely related Fusarium species viz. N. haematococca, F. oxysporum, F. graminearum and F. verticillioides. Among the organisms investigated, 2,076 of the F. virguliforme proteins showed identity to corresponding proteins of the distant species. For example, a large proportion of E. coli (42%) proteins are conserved in F. virguliforme (Table S7).

We searched for conserved genes (E≤10−9) via BLASTP in a step-by-step fashion as described below. First, we identified 13,068 N. haematococca proteins that showed identity (E≤10−9) to F. virguliforme proteins. Of these 13,068 N. haematococca proteins, 11,878 showed identity to F. oxysporum proteins. We used these 11,878 proteins in the next step of the search and so on (Figure S4; Table S8). We identified 762 F. virguliforme proteins (5.13%) that are conserved in all organisms including E. coli. The set of 762 F. virguliforme proteins (Table S9) conserved in all 26 organisms was classified by molecular function into 18 groups (Figure S5). As expected, many of the conserved proteins regulate housekeeping functions such as metabolic, cellular and developmental processes in all organisms.

Identification of genes unique to F. virguliforme

In order to identify the genes unique to F. virguliforme, we compared the genome sequences of F. virguliforme and four closely related Fusarium species (Table S8) by conducting BLASTP analyses (E≤10−9) [33]. We identified 11,043 genes that were common to all five Fusarium species (Figure S6). Of the 14,845 F. virguliforme genes, 1,332 were unique to F. virguliforme. Further investigation of these 1,332 genes revealed that most were novel; only 98 of the 1,332 unique F. virguliforme genes showed similarities to known genes (Table S10). Based on GO annotations (Blast2GO; biological process), the 98 unique F. virguliforme genes were classified into 19 groups (Figure S7). Potential pathogenesis-related genes in this group included a polyketide synthase, protein serine threonine kinase and carbonic anhydrase, all of which could play important roles in initiating SDS in soybean (Table S10) [34]–[36].

Phylogenic analysis of F. virguliforme

A phylogenic analysis was conducted to determine the relatedness of F. virguliforme to other Fusarium species. Ten orthologous, single copy genes were selected arbitrarily to construct a phylogenetic tree of five Fusarium species, M. grisea [37], N. crassa [38], A. nidulans [39], R. oryzae [40], P. blakesleeanus (http://genome.jgi-psf.org/Phybl1/Phybl1.home.html) and U. maydis (http://www.broadinstitute.org/). The oomycete pathogens, P. sojae (soybean pathogen) [41] and P. infestans (potato pathogen) [42], were included as taxonomic out-groups [24]. The five Fusarium species grouped together in one clade suggesting their origin from a single progenitor species. As expected, the oomycete pathogens grouped into a separate, more distant clade. The Fusarium clade was closest to the rice blast pathogen, M. grisea (Figure 4). Within this clade, F. virguliforme formed a sub-clade with the pea pathogen N. haematococca, which suggests that N. haematococca is the closest relative of F. virguliforme among the sequenced Fusarium species. This result supports our observations from the synteny study presented in Figure 1.

Figure 4

Phylogenetic tree showing the relationship of F. virguliforme with other Fusarium species.

The tree was constructed using 10 randomly selected single copy orthologous genes using PHYML program (WAG model of evolution) with 1,000 bootstraps.

Identification of candidate pathogenicity proteins

Two approaches were applied to identify candidate pathogenicity proteins. Firstly, the F. virguliforme genome sequence was interrogated with sequences of the pathogen-host interaction database (PHI database; http://www.phi-base.org/), consisted of experimentally verified pathogenicity, virulence, and effector proteins from bacteria, fungi and oomycetes that infect plants, humans, animals, insects, fishes and fungi. Of the 1,100 proteins in the PHI database, 786 pathogenicity genes are from fungi, 27 from oomycetes, 137 from bacteria and the rest are effector proteins [13]. The F. virguliforme genome was searched with the PHI protein database (E≤10−9) to identify possible pathogenicity genes. The 358 F. virguliforme proteins showing high sequence identity to members of the PHI protein database (Table S11) were classified into 21 groups based on GO annotation (biological process/level 2) (Figure S8). A substantial percent of the 358 genes were involved in metabolic processes; many were predicted to be involved in the biosynthesis of secondary metabolites. We identified five polyketide synthase genes that may be involved in the biosynthesis of non-proteinacious toxin (Table S11). We identified three pectate lyases, which could be involved in root necrosis or root rot. Another candidate pathogenicity protein (Fv2806) showed identity to the Pseudomonas syringae type III effector HopI1 protein (AAL84247.1). This list of pathogenicity genes laid the foundation for dissecting the pathogenicity mechanisms through functional analyses of these genes.

Some of the pathogenicity proteins are excreted to the extracellular space. These proteins carry signal sequences for excretion. In the second approach, we applied the SignalP program [43] to identify proteins containing signal peptides. Among the 14,845 predicted F. virguliforme proteins, use of the hmm model with the cut off 0.9 hmm score identified 1,155 putative secretory proteins (Table S12). These proteins were annotated using Blast2go and classified into eight groups (Figure S9). A large number of these proteins contain catalytic and binding activity sites. Some of these proteins could be important pathogenicity factors for SDS development in soybean.

Conclusions

Here, we present the genome sequence of F. virguliforme, an important soybean pathogen that causes losses estimated to be over $0.1 billion annually in the United States [44]. Although SDS is characterized by distinctive foliar symptoms, the pathogen is exclusively found in roots of diseased plants. One or more fungal toxins have long been suspected to induce foliar SDS symptoms, although the current understanding of symptom development is fragmentary. In order to identify candidate pathogenicity genes, we interrogated the F. virguliforme genome sequence with the pathogen-host interactions sequence database and identified 358 candidate pathogenicity genes (Table S11). These include five polyketide synthases, which may be involved in the synthesis of polyketide toxins [45]. We also identified three pectate lyases that may be involved in cell wall degradation in root tissues to cause root necrosis or rotting. Additionally, we identified a candidate pathogenicity protein that showed high similarity to a bacterial effector protein (AAL84247.1). Among the identified 1,332 unique F. virguliforme genes, only 98 showed similarity to previously isolated genes. One of the 98 genes encodes a polyketide synthase, which may be involved in toxin biosynthesis [35].

Comparisons of the F. virguliforme genome with that of four Fusarium pathogens revealed new information about the relatedness of the five species and fundamental genomic similarities shared by these pathogenic species. Among the Fusarium species studied, the pea pathogen N. haematococca (F. solani) is the closest relative of F. virguliforme. The genome size of F. virguliforme is comparable to that of the previously sequenced Fusarium species. We observed that the G + C content in the gene-poor regions of the F. virguliforme genome was reduced approximately to half of the average G + C content of the genome. Furthermore, we identified a set of 762 F. virguliforme proteins (Table S9) that are conserved across a set of 26 organisms including F. virguliforme. The 762 conserved proteins, as expected, primarily regulate metabolic and cellular functions (Figure S5).

In summary, through this investigation, we have assembled the F. virguliforme genome sequence by conducting shotgun 454-sequencing and identified a set of candidate pathogenicity genes for discovering the pathogenicity mechanisms used by this serious soybean pathogen to cause SDS. Genome sequence reported here would become important public resource to a broad community of researchers engaged in developing tools to manage SDS, one of the most devastating diseases affecting global soybean production.

Dot-plot analyses of with four spp. The alignments are between Fusarium virguliforme Scaffold 1 (5.05 Mb) and genome sequences of the Fusarium spp. A) F. virguliforme with F. graminearum (8.93 Mb); B) F. virguliforme with F. verticilliodes (4.62 Mb); C) F. virguliforme with F. oxysporum (4.35 Mb); D) F. virguliforme with N. haematococca (4.93 Mb).

(PPT)

Click here for additional data file.

Classification of the genes based on GO annotation. Biological process was considered in classifying the genes.

(PPT)

Click here for additional data file.

Extent of similarity of genes with that of the selected organisms. The blue line indicates the percentage F. virguliforme genes that are similar (E≤9) to selected organisms. Brown line represents the proportion of genes in an individual that showed similarity (E≤9) to F. virguliforme genes.

(PPT)

Click here for additional data file.

Conserved proteins across a wide range of species. Organisms investigated were F. virguliforme (Fv), N. haematococca (Nh), F. oxysporum (Fo), F. graminearum (Fg), F. verticillioides (Fvt), N. crassa (Nc), A. nidulans (An), U. maydis (Um), P. blakesleeanus (Pb), R. oryzae (Ro), S. cerevisiae (Sc), D. rerio (Dr), D. melanogaster (Dm), G. max (Gm), A. thaliana (At), H. sapiens (Hs), P. sojae (Ps), P. infestans (Pi), R. leguminosarum (Rl), O. sativa ssp. japonica (Osj), D. discoideum (Dd), Z. mays (Zm), P. syringae (Ps), A. tumefaciens (Atu), E. coli (Ec), C. elegans (Ce). Conserved F. virguliforme gene numbers in a species are shown in parenthesis.

(PPT)

Click here for additional data file.

GO annotation of 762 proteins that are conserved across 25 diverse organisms. A large number of the conserved proteins control metabolic, cellular and developmental processes.

(PPT)

Click here for additional data file.

Unique . There are 11,403 genes that are common to all five Fusarium species and 1,332 genes unique to F. virguliforme at E≤10−9. The numbers in the light blue circle are the number of genes of a species that showed similarity (E≤10−9) to genes of at least another species.

(PPT)

Click here for additional data file.

GO annotation of 98 unique genes based on biological process.

(PPT)

Click here for additional data file.

GO annotation of 358 candidate pathogenicity proteins based on biological process.

(PPT)

Click here for additional data file.

GO annotation of 1,155 putative secretory proteins based on biological process.

(PPT)

Click here for additional data file.

Alignments of coordinates between and Mont1 genomes.

(DOC)

Click here for additional data file.

Gene density and GC contents across

(DOC)

Click here for additional data file.

Variation in gene density and G + C contents of coding sequences observed among 23 major scaffolds of the genome.

(DOC)

Click here for additional data file.

Pfam functional domain groups

(DOC)

Click here for additional data file.

List of the normalized species.

(DOC)

Click here for additional data file.

Organisms and data sources used in genome analyses.

(DOC)

Click here for additional data file.

( ) genes thaxft showed similarity to genes of other organisms.

(DOC)

Click here for additional data file.

Conserved genes among selected organisms.

(DOC)

Click here for additional data file.

The 762 genes that are conserved among 25 diverse organisms.

(DOC)

Click here for additional data file.

GO annotation of 98 of the 1,332 unique genes.

(DOC)

Click here for additional data file.

Identification of candidate 358 pathogenicity genes through interrogation of the genome with the PHI database.

(DOC)

Click here for additional data file.

Candidate secretory proteins.

(DOC)

Click here for additional data file.