F. pseudograminearum Genome Sequence

The genome of F. pseudograminearum isolate CS3096 was sequenced using a combination of paired and single-end Illumina reads. De novo assembly of these reads resulted in a nuclear genome size of 36.8 Mbp assembled in 656 contigs with 50% of all nucleotides in contigs of 189 Kbp in length or greater (AFNW00000000). The average sequence coverage across these contigs was 179-fold. Compared to many other fungal genome assemblies using next generation sequencing technologies, the F. pseudograminearum genome sequence assembly has a relatively high N50 (Table 1). Gene model predictions from three programs were combined to identify 12,448 protein coding gene (see materials and methods).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Summary of the F. pseudograminearum genome sequence assembly in comparison to selected fungal genome sequences obtained by short-read sequencing and the reference genome for F. graminearum.

https://doi.org/10.1371/journal.ppat.1002952.t001

The repeated sequence content of the F. pseudograminearum genome, assessed using the RepeatMasker program, is only 1.6%, which is slightly higher than that of the F. graminearum genome (0.7%) assessed using identical parameters, albeit that the sequencing and assembly methodology differed. RepeatMasker recognizes both simple sequence repeats and transposable elements present in RepBase [27]. Although approximately four times as many base pairs were flagged as being derived from Gypsy type Long Terminal Repeat (LTR) elements in the F. pseudograminearum genome (26 Kbp) compared to F. graminearum (6 Kbp), the difference in repetitive DNA content could mostly be attributed to a higher level of simple repeats and low complexity DNA (1.5% versus 0.41% of the genome, respectively). One high coverage contig in F. pseudograminearum encodes a LTR-type retrotransposon with best match to an F. oxysporum transposon, probably present in 9–10 copies based on an average coverage of 1689 fold. This transposon matched to sequences in a number of different fungi and also to transposons in both monocot and dicot plants. The contig showed no polymorphism in the sequence read assembly across 5.5 Kbp, suggesting all copies are identical and thus could not be placed in other assembled contigs. Also not included in the overall repeat counts are the contigs that represent rRNA encoding genes. The Illumina sequencing approach used here was unable to resolve these repeats in F. pseudograminearum and these currently appear in the assembly as high-coverage contigs.

The F. pseudograminearum genome sequence assembly was also compared globally to that of F. graminearum [28] by aligning the two genomes after masking simple repeat sequences and known fungal repetitive elements. In total, 89.8% of the F. pseudograminearum genomic sequence could be aligned to the F. graminearum genome at >70% nucleotide identity (Figure 1). An alignment with increased sensitivity was performed using a six frame translations of both genomes enabling an alignment of 94% of the F. pseudograminearum genome to that of F. graminearum. Thus, at least 6% of the low copy region of the genome (approximately 2.2 Mbp) appears to be completely unique to F. pseudograminearum. Very few rearrangements between the F. pseudograminearum and F. graminearum genomes in the aligned regions were found (Figure 1A). The amount of aligned sequence between the two species decreases towards the ends of the F. graminearum chromosomes (Figure 1B) and also in regions previously reported to be undergoing higher rates of genome innovation [28].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Alignment between F. graminearum and F. pseudograminearum.

F. graminearum chromosomes are ordered in decreasing size. (A) Dot-plot representing a whole genome alignment between F. graminearum isolate Ph1 and F. pseudograminearum isolate CS3096. The alignment was generated with NUCmer, part of the MUMmer 3 comparative sequence analysis package. Sequences were pre-masked for known repetitive elements and simple repeats using RepeatMasker. The dot plot represents the best 1∶1 alignment between the two genomes. Dots closest to the diagonal represent co-linearity between the two genomes. Red represents matches in the forward direction and blue is indicative of inversions in part of the mapped contig relative to the F. graminearum genome. (B) Distribution of gaps in the alignment between F. pseudograminearum and F. graminearum, relative to the F. graminearum genome. F. graminearum chromosomes were divided into 100 Kbp non-overlapping windows and the unaligned nucleotides in each window summed and expressed as a percentage.

https://doi.org/10.1371/journal.ppat.1002952.g001

Comparative Genomics of F. pseudograminearum and Signatures of Horizontal Gene Transfer into Cereal Pathogens

Genes showing a distribution of homologues limited to cereal pathogens would be both candidates for strong selection, because of a possible involvement in pathogenicity, and for horizontal acquisition. To determine whether the F. pseudograminearum predicted gene set contained such candidates, a BLASTmatrix analysis pipeline based on the identification of gene homologues by reciprocal best BLASTp hits was developed using predicted proteomes from the genome sequence of 16 different cereal pathogens and 11 non-pathogens of cereals. The entire BLASTmatrix analysis is presented in Dataset S1 along with the filtered protein sets described below. A numerical summary of the matches is shown in Table S1. In these analyses, 156 predicted F. pseudograminearum proteins did not have reciprocal best BLAST hits of any strength in any of the 27 organisms, while 239 predicted proteins had reciprocal best BLASTp matches only in cereal pathogens, with a score of at least 200 bits. In contrast, applying the same filtering criteria to identify proteins that are found in F. pseudograminearum and only non-cereal pathogens in this BLASTmatrix analysis identified only 32 proteins, 24 of which were present in the Fusarium lineage.

Amongst the 239 cereal pathogen proteins, 214 had matches within the two other Fusarium cereal pathogens (F. graminearum and F. verticillioides) and not in other species. Amongst the remaining 25 F. pseudograminearum proteins, only five had no match in other fusaria while matching other cereal pathogen proteins. Since the genome sampling described herein was limited to 27 fungal genomes, the sets of 239 and 156 proteins were combined and additionally curated using BLASTp against the NCBI nr (non-redundant) database. This process identified a total of 17 proteins with strong BLASTp matches (>200 bits) to bacterial proteins, and six of these 17 proteins were present only in a small number of other fungi (Table 2). The corresponding genes for these proteins are therefore good candidates for having been ancestrally acquired from bacteria. The GC content for each of these six genes was not obviously different to that of surrounding genes, possibly with the exception of FPSE_07765 (see below). Other candidates for cereal pathogen specificity, which to varying degrees showed distributions limited to cereal pathogens, were also identified through the BLASTmatrix filtering. These are therefore also of interest with respect to putative functions in niche specialisation and virulence towards plants. Examples of predicted genes from this latter category suggesting conserved roles in virulence towards cereals, and/or horizontal gene transfer are shown in Table 3. These candidate genes were subjected to more detailed analysis of sequence relationships as described below.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Fusarium pseudograminearum genes that may be of bacterial origin and are candidates for roles in pathogenicity towards plants.

https://doi.org/10.1371/journal.ppat.1002952.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Examples of Fusarium pseudograminearum genes that may be of fungal origin and are candidates for roles in cereal pathogenicity.

https://doi.org/10.1371/journal.ppat.1002952.t003

Putative Functions and Sequence Relationships of Selected Cereal Pathogen Genes

Amongst the F. pseudograminearum genes of potential bacterial origin (Table 2), the intronless gene FPSE_07765 and its orthologues in F. graminearum and F. verticillioides encode proteins highly similar to a bacterial protein with both aminotransferase and homoserine kinase domains. FPSE_07765 is 75% identical at the amino acid level across the entire length to a protein from Microbacterium testaceum, which is a bacterial endophyte of a variety of plants including the cereals sorghum and maize [29], [30]. Other fungal hits to this sequence in GenBank only align across less than half of the protein and at much lower identities, with the best fungal match in Trichophyton verrucosum at only 31% amino acid identity and with only partial query coverage of 44%. Phylogenetic analysis of FPSE_07765 identifies a Fusarium sequence clade embedded amongst a range of related bacterial sequences and strongly supports an origin of this gene via horizontal acquisition from bacteria, with retention in the cereal infecting fusaria (Figure S1). The GC content of FPSE_07765 was 58% compared to the genome average of 51.8±3.9% (±SD) for coding sequences.

FPSE_11233, another F. pseudograminearum gene with possible horizontal acquisition, encodes a putative secreted hydrolase that has reciprocal best BLAST hits only in cereal pathogens among the 27 selected genomes as well as hits in GenBank to putative pectin- and xylan-hydrolases in bacteria (best match is 38% identical with an e-value of 1e⁻¹²⁵ to a Streptomyces hygroscopicus protein). Further manual database queries showed FPSE_11233-like sequences in the cereal endophyte Chaetomium globosum and the Brassica pathogen Leptosphaeria maculans. This gene sequence was analyzed further because, although L. maculans is not a cereal pathogen, it causes blackleg disease of canola which is a common rotation crop used in cereal farming systems [31], and L. maculans is related to other cereal pathogens in the Pleosporales family [32], [33]. Again, phylogenetic analysis supported the relatedness of these fungal genes to bacterial sequences (Figure S2). However, any potential acquisition from bacteria is clearly extremely ancient as the sequence seems to have undergone considerable vertical diversification within the fungi following acquisition from bacteria. This hypothesis is not only supported by the phylogenetic analysis but also computational predictions, suggesting that the position and number of introns in FPSE_11233-like sequences are not conserved between fungal species (data not shown). The placement of the L. maculans sequence outside of the Dothideomycete clade in the phylogenetic analysis (Figure S2) may also indicate a more complex mode of inheritance within the fungi.

A number of other enzyme-encoding genes of putative bacterial origin were also identified in the F. pseudograminearum genome, including genes encoding a hydrolase of unknown specificity (FPSE_07775) and a NAD+ dependent dehydratase/epimerase (FPSE_11221). Two putative amidohydrolase encoding genes (FPSE_00725 and FPSE_05717) that have clear bacterial homologues were also identified in this analysis. One of the amidohydrolase genes, FPSE_05717 hereafter termed FpAH1, also appeared to have a clear homologue in P. nodorum but not in any other fungus, suggesting a potential role for this gene in cereal pathogenesis and an unusual evolutionary history. The genomic organization, sequence diversity and virulence function of FpAH1 will be described in more detail later in the manuscript.

Many F. pseudograminearum genes also had homologues in other fungal cereal pathogens but no clear bacterial matches, suggesting that these genes are either laterally inherited or rapidly diversifying and therefore have been selectively retained only in a limited number of pathogenic fungi. Examples of these (Table 3) include three genes (FPSE_06956, FPSE_10646 and FPSE_02381) encoding small secreted proteins (candidate effector molecules) that were present in F. pseudograminearum and F. graminearum as well as other cereal pathogens. FPSE_06956 had orthologous matches in the Magnaporthe and Fusarium lineages but no other matches in the BLASTmatrix or GenBank. FPSE_10646 is a member of the killer protein 4 family (PFAM09044) of toxins that were shown to be extensively laterally shared between multiple fungal lineages including non-pathogens [34]. FPSE_02381 is a member of a two-gene family encoding small secreted cysteine rich proteins in F. pseudograminearum and F. graminearum and has strong, albeit non-reciprocal BLASTp matches in M. oryzae. The other member of this family in F. pseudograminearum, FPSE_05488, had reciprocal best BLASTp hits only in cereal pathogens, with the exception of a single match in the canola pathogen Colletotrichum higginsianum in the BLASTmatrix analysis (Dataset S1). Three other gene products shown in Table 3 (FPSE_05718, FPSE_05719 and FPSE_05720) were encoded in a gene cluster in the F. pseudograminearum genome. Interestingly, FpAH1, which is predicted to be of bacterial origin (see above and Table 2), also seems to be part of this cluster. The function of FpAH1 will be described in more detail later in the manuscript.

Also shown in Table 3 is a gene encoding a putative dienelactone hydrolase (FPSE_08136), hereafter termed FpDLH1, with very strong orthologues in F. verticillioides and C. graminicola with 98% and 88% identities, respectively at the amino acid level. The next best BLASTp match (69% identical) of FpDLH1 in GenBank, which is not a reciprocal best hit, is to FGSG_00089 from F. graminearum. The absence of strong reciprocal matches in all other fungi suggests that this gene may have been either shared between, or independently acquired from another donor species, by these Fusarium and Colletotrichum species through horizontal transfer mechanisms. The genomic organization and function of FpDLH1 will be described in more detail later in the manuscript.

FpAH1, a Horizontally Acquired Amidohydrolase Gene

FpAH1 encodes a putative amidohydrolase (Pfam domain PF07969). The best BLASTp match (e-value 2e⁻¹⁴¹, 88% identical, 94% similar at the amino acid level) of FpAH1 was to a predicted protein from the wheat glume blotch pathogen P. nodorum (encoded by SNOG_04819 hereafter termed PnAH1) [35]. The lack of FpAH1 homologues in other fusaria and presence in P. nodorum was confirmed via hybridization of a FpAH1 probe to genomic DNA of various Fusarium and a P. nodorum isolates (Figure S3). The overwhelming majority of subsequent BLASTp matches of FpAH1 in GenBank were to predicted proteins from bacteria (Table 4). The next fungal match after P. nodorum was to the F. graminearum protein FGSG_10599 (25% identical 2e⁻²⁵) but this was much weaker than the bacterial matches (Table 4). FGSG_10599 also had another much stronger, indeed nearly identical (97%) protein (FPSE_00725) encoded in the F. pseudograminearum genome. The FPSE_00725/FGSG_10599 orthologs also had next best matches in bacteria. Phylogenetic analysis of these two F. pseudograminearum amidohydrolases strongly supports the hypothesis that they are of bacterial origin (Figure 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Phylogram of fungal amidohydrolases found in F. pseudograminearum that appear to be of bacterial origin and a number of bacterial amidohydrolases.

Fungal sequences are highlighted in grey boxes. Numbers on branches indicate approximate likelihood ratio test branch support values.

https://doi.org/10.1371/journal.ppat.1002952.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Summary of BLASTp analysis of FpAH1 versus the non-redundant protein database (nr) at NCBI.

https://doi.org/10.1371/journal.ppat.1002952.t004

The F. graminearum genome contains six entries in this class of amidohydrolases (PF07969), all of which contain predicted orthologous matches with other genes in the F. pseudograminearum genome (Table S2). In other fusaria, clear orthologous relationships exist between the remaining five amidohydrolases with Pfam domain PF07969 found in F. pseudograminearum, although both F. oxysporum and F. verticillioides contain additional members in this class of amidohydrolases (Table S2). In P. nodorum, only four proteins fall into this class of amidohydrolases. Thus FPSE_00725 and FpAH1 encode two amidohydrolases with extremely restricted distribution in fungi and are most likely of bacterial origin. BLASTp analysis of the other F. pseudograminearum amidohydrolases identified a much wider distribution in fungi with FPSE_00474, FPSE_02365, FPSE_03227 and FPSE_11444 having more than fungal 30 hits of greater strength than the best bacterial hit (Table S2). FPSE_05738 was less widely distributed with seven hits of greater score than the best bacterial hit (Table S2) and may also be a candidate for acquisition from bacteria.

The predicted FpAH1 protein of 570 amino acid residues was encoded by an uninterrupted open reading frame of 1710 bp that was confirmed by RNAseq analysis of cDNA (data not shown). The P. nodorum genome annotation for PnAH1 contained a single intron, but it is likely that this prediction was not correct as the PnAH1 genomic region has a single uninterrupted open reading frame. In the coding region, FpAH1 was conserved between F. pseudograminearum and P. nodorum with 89% identity at the nucleotide level and 174 bp upstream of the predicted start codon and 85 bp downstream of the predicted stop codon could also be readily aligned.

The chromosomal complement of F. pseudograminearum has not been characterized and therefore the location of FpAH1 in the F. pseudograminearum genome is unknown. FpAH1 is present near the end of a ∼90 kb sequence contig, the first 70 kb of which aligns with the end of F. graminearum chromosome 1, as shown in Figure 3. Interestingly, the genes surrounding FpAH1 did not have clear orthologues in the F. graminearum genome. On the contig containing FpAH1, 28 genes (FPSE_05686 through to FPSE_05714, excluding FPSE_05712) had clear orthologues in F. graminearum. However, FPSE_ 5715 to FPSE_05720 did not. Furthermore, three of the gene products (FPSE_05718, FPSE_05719 and FPSE_05720) were also identified in the BLASTmatrix analysis to be cereal pathogen-specific, suggesting that parallel selection or coordinated acquisition may have affected this region (Table 3). The putative function and position of the genes in this region in F. pseudograminearum are shown in Figure 3. Of the remaining 40 genes on the end of chromosome 1 in F. graminearum, only eight encoded proteins with reciprocal best BLAST hits in F. pseudograminearum, and these were distributed across six different contigs in the F. pseudograminearum assembly (data not shown). In contrast, PnAH1 is ∼100 kb from the end of supercontig seven of the P. nodorum genome and appears to be surrounded by genes encoding proteins conserved in other fungi. In both P. nodorum and F. pseudograminearum, the GC content in the region of the PnAH1 and FpAH1 was similar to that of other gene rich regions of the respective genomes and some of the surrounding genes in the regions had introns. In summary, the genomic region containing FpAH1 has no equivalent sequence at the syntenic location in the F. graminearum genome nor the region containing PnAH1 in the P. nodorum genome.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Nucleotide alignment of the end of F. graminearum chromosome 1 (top) and the F. pseudograminearum contig containing FpAH1 (bottom).

Clear orthologous relationships were evident between regions joined by grey shaded boxes. FpAH1 is indicated in red. The genes surrounding FpAH1 in the F. pseudograminearum genome that do not have clear orthologous relationships in F. graminearum are blown up and the putative function and species of the closest match in GenBank is indicated.

https://doi.org/10.1371/journal.ppat.1002952.g003

AH1 Sequence Phylogenies Indicate Ancient and Independent Horizontal Transfers into Fungal Pathogens

The lack of close orthologues of FpAH1 and PnAH1 in other fungi and the presence of similar genes in bacteria (Table 4) based on BLASTp searches suggested an origin for these fungal genes via horizontal acquisition from bacterial species. fusaria belong to the order/class Hypocreales/Sordariomycetes while Phaeosphaeria is in the distantly related Pleosporeales/Dothidiomycetes. Acquisition of the gene may have occurred independently in both species or alternatively was horizontally transferred between the fungal species. To differentiate these possibilities, the sequence diversity of each gene was assessed in several globally sourced Fusarium and Phaeosphaeria isolates (Table S3). Isolates identified as F. pseudograminearum by multi-locus DNA sequencing [22] or by sequencing the elongation factor 1 alpha (EF1α) gene, were the only Fusarium sp. out of six tested that produced positive amplicons. PCR analyses of the distribution of FpAH1 in fusaria were confirmed for a limited number of isolates by hybridization analysis (Figure S3). PnAH1 sequences were PCR amplifiable from all P. nodorum isolates tested and also from two of four sister species, P. avenaria f. sp. tritici (Pat) 1 and Pat3, suggesting PnAH1 was present in a common ancestor in this lineage.

In order to test the hypothesis of horizontal transfer between F. pseudograminearum and P. nodorum, a ∼500 bp region of the AH1 gene was sequenced in isolate collections. A haplotype network showing the sequence relationships across both genera is shown in Figure 4. In F. pseudograminearum, seven haplotypes observed did not correspond to the geographic origin of isolates, consistent with global gene flow within the species as previously described [22]. Diversity in the Phaeosphaeria spp. sequences was more limited with only one haplotype observed in a global sample of P. nodorum and two haplotypes detected within each of the two Phaeosphaeria sister species containing PnAH1 orthologues. There were no shared sequence haplotypes between Phaeosphaeria spp. and F. pseudograminearum. The presence of AH1 orthologues in up to four related Phaeosphaeria spp. suggests that the acquisition of this gene occurred before the divergence of these species. Divergence of PnAH1 between the Phaeosphaeria spp. was comparable to divergence observed at neutral sequence loci (M.C. McDonald, unpublished data). Furthermore, the haplotypes observed in the Phaeosphaeria species complex and F. pseudograminearum were distinct, suggesting independent acquisitions of AH1-like sequences by each lineage.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Statistical parsimony haplotype network for FpAH1/PnAH1 in Fusarium pseudograminearum and Phaeosphaeria spp. Haplotype bubbles are proportional to sample size shown in parentheses.

Dots on line segments indicate single point mutations.

https://doi.org/10.1371/journal.ppat.1002952.g004

FpAH1 Contributes to Virulence on Wheat and Barley

The likely independent acquisition and retention of AH1 orthologues by F. pseudograminearum and P. nodorum suggests that this gene may encode a protein that is necessary for virulence, at least in some hosts. To test this hypothesis, a functional analysis of the FpAH1 and PnAH1 genes was undertaken. FpAH1 was expressed in both infected barley and wheat leaves and roots (Figure S4). The role of the FpAH1 amidohydrolase in FCR of barley was also assessed by generating gene deletion mutants in F. pseudograminearum by replacing 635 nucleotides of the FpAH1 locus with a geneticin resistance gene cassette (Figure S5A). There were no obvious defects in appearance or sporulation of the mutants. A culture time-course was used to compare the growth rates of one mutant to that of the parental strain and these were indistinguishable (Figure S5C). As shown in Figure 5, two independently-derived knock-out mutants consistently showed reduced virulence towards barley (cv. Golden Promise) seedlings across multiple independent experiments using a previously established FCR inoculation assay [36]. A second barley cultivar (cv Gairdner) also showed similar reduced virulence (data not shown). FCR disease severity caused by the mutant strains was significantly (P-value<0.01) reduced compared to those caused by the wild type strain. Genetic complementation of an FpAH1 mutant with a cassette containing FpAH1 under the control of the Aspergillus nidulans TrpC promoter restored virulence towards barley (Figure 5C and 5D), providing further evidence that FpAH1 is required for virulence against barley.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Virulence assay of the Fusarium pseudograminearum amidohydrolase 1 mutants (ΔFpAH1) compared to the parental strain towards barley cultivar Golden Promise at 21 days post inoculation (dpi) (A–B).

(A) Survival of plants in the assay 21 dpi. N = 3 with each replicate consisting of 13–15 plants divided between two paper towel rolls and maintained in separate vessels. The difference between wild type and each mutant was highly statistically significant (P-value <0.01). (B) Representative rolls of plants from the assay. (C–D). Complementation of ΔFpAH1 (strains 96T681 and 96T682) restores virulence towards barley cultivar Golden Promise. (C) Proportion of plants surviving at 13 dpi. The total number of plants in each case was 30. (D) Photograph of assays at 13 dpi.

https://doi.org/10.1371/journal.ppat.1002952.g005

F. pseudograminearum has a wide host range within cereals with no evidence for race specialization [37], [38] for FCR disease on wheat. Unlike the experiments conducted on barley, highly replicated FCR assays on wheat using the FpAH1 knock-out mutants failed to reveal reduction of virulence on two unrelated varieties of wheat (Figure S6). However, reduced virulence was detectable when inoculated directly on roots toward both wheat and barley and again complementation restored virulence (Figure 6). P. nodorum is not pathogenic on barley and therefore was not tested on this host but mutant strains of this pathogen with deletions of the PnAH1 gene were also generated (Figure S5D) and tested on wheat plants in replicated leaf infection assays. No significant differences in virulence were observed between mutant and wild type strains on wheat (Figure S7). These experiments indicate that FpAH1 is required for full virulence on both wheat and barley but the importance of PnAH1 in pathogenesis remains unknown.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 6. Fusarium root-rot virulence assay of the Fusarium pseudograminearum amidohydrolase 1 mutant (ΔFpAH1) towards wheat cultivar Kennedy (A and B) and barley cultivar Golden Promise (C and D).

N = 16 individual plants. CS3096 is the parental strain for the FpAH1 mutant (96T492), which was complemented with FpAH1 driven by the TrpC promoter (96T681). Error bars represent the standard error of the mean. Different letters on A and C indicate statistically significant differences (P<0.05) in pair-wise t-tests. B and D are the plants used to score the shoot length shown in A and C.

https://doi.org/10.1371/journal.ppat.1002952.g006

Novel Genomic Context and Virulence Function on Cereals for the FpDLH1 Gene

The F. pseudograminearum locus FpDLH1 encoding a dienelactone hydrolase was identified in the BLASTmatrix analysis as a cereal-pathogen associated gene with homologues detected only in F. verticillioides and C. graminicola. Strikingly, the gene (FPSE_08135 hereafter termed FpAMD1) immediately adjacent to FpDLH1 in the F. pseudograminearum genome encodes an amidase that is also present in the same divergently transcribed arrangement in the genomes of F. verticillioides and C. graminicola (Figure 7), indicating this is likely to be a two-gene cluster. The amidase family is much larger in ascomycetes than the dienelactone hydrolase family, making one-to-one orthologous relationships between species difficult to detect. However, for both FpAMD1 and FpDLH1, strong homology is detected between all three species at the nucleotide level (>80%) across the coding sequences of these genes. Furthermore phylogenetic analysis also supports the grouping of both of these genes in the three species in clades incongruent with expected evolutionary relationships (Figure S8). Thus, their genomic arrangement, discontinuous distribution in the fungi and strong homology to each other all suggest these genes represent a two-gene cluster that may have a common origin.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 7. The FpDLH1-FpAMD1 locus in Fusarium pseudograminearum and the arrangement of the genes in this region in F. graminearum, F. verticillioides and Colletotrichum graminicola.

The DLH1-AMD1 locus is boxed in grey in each of the species. Genes of the same color in different species are reciprocal best blast hits and likely to be orthologous. Sequence contigs for each species are truncated to the relevant regions only.

https://doi.org/10.1371/journal.ppat.1002952.g007

The FpDLH1 gene has the same intron-exon structure as another dienelactone hydrolase that is present in both F. pseudograminearum and F. graminearum (FPSE_08131 and FGSG_00089, Figure 7). This intron arrangement is not a generic feature of dienelactone hydrolase encoding genes in F. pseudograminearum, indicating a possible gene duplication event in the Fusarium lineage. Furthermore, FPSE_08131 had reciprocal best BLAST hits in all fusaria included in the analysis, with the exception of F. solani, as well as in two Magnaporthe spp., and the intron-exon structure was maintained in all species. The synteny in this region is also well conserved in the F. pseudograminearum-F. graminearum comparison (albeit split into two regions that align near the ends of different chromosomes in F. graminearum) but outside of this comparison the synteny weakens with the orthologues of the genes flanking this two-gene cluster in F. pseudograminearum appearing on multiple different contigs in F. verticillioides (Figure 7). In the F. verticillioides-F. pseudograminearum comparison, the conservation of FpDLH1/FvDLH1 (98% identical at the amino acid level) is much greater than the more widespread dienelactone hydrolase (FPSE_08131 and its F. verticillioides orthologue FVEG_12625) at 75% identity, suggestive of either an intra-fusaria transfer or strong conservation and selection. In C. graminicola the DLH1-AMD1 gene cluster is located on chromosome 2 in a region where only two of 11 genes flanking the cluster have clear orthologous relationships between C. graminicola and the closely related C. higginsianum (data not shown). However, three of the nine genes on one flank of the C. graminicola DLH1-AMD1 gene cluster have orthologues present in the flank adjacent to FpDLH1-FpAMD1 (Figure 7). These genomic associations suggest an ancient relationship in these regions.

Both FpDLH1 and FpAMD1 are expressed during infection of root by F. pseudograminearum and leaf tissue with higher expression in wheat than in barley (Figure S4). The role of FpDLH1 in fungal pathogenesis was assessed by creating knockout strains of two different F. pseudograminearum strains (CS3096 and CS3427), where the FpDLH1 gene was replaced by the geneticin resistance cassette (Figure S9). No obvious differences in sporulation were observed in the mutants nor were there differences in growth rate compared to their respective parents (Figure S9C). However, FpDLH1 mutants in both strain backgrounds showed significantly reduced virulence towards both wheat and barley in both root-rot and FCR assays (Figure 8 and Figure S10), suggesting that FpDLH1 contributes to fungal virulence against cereal plants.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 8. Virulence assay of the Fusarium pseudograminearum dienelactone hydrolase 1 mutant (ΔFpDLH1) mutant towards wheat cultivar Kennedy in crown rot (A and B) and root rot (C and D) assays.

For A and B N = 3 with each biological replicate consisting of 4 paper towel rolls consisting of 7–8 plants per roll. For C and D N = 15–16 plants. Error bars represent the standard error of the mean. Letters indicate statistically significant differences at P<0.05. CS3096 and CS3427 are the parental strains for the FpDLH1 mutants, 96T926 and 27T892 respectively. Mock treatments are inoculations performed with agar plugs that have not been colonized by Fusarium.

https://doi.org/10.1371/journal.ppat.1002952.g008

Fungal Isolates and Culture

The F. pseudograminearum isolate (CS3096) chosen for genome sequencing was originally isolated from a wheat crown collected in 2001 near Moree in Northern New South Wales, Australia [26]. Isolates for phylogenetic analysis of the FpAH1 gene were selected from a collection housed at CSIRO Plant Industry, Brisbane, Australia, consisting of isolates from Australia, New Zealand, Canada, United States of America and Turkey (Table S3). Phaeosphaeria isolates were selected from a collection housed at ETH Zurich and the Australian isolate SN15.

Sequencing and Assembly

Two lanes of single-end 75 bp and one lane of 100 bp paired-end sequencing was performed on an Illumina GAII_x genome sequencer by the Australian Genome Research Facility, Melbourne Australia. Reads were imported into CLCBio Genomics Workbench 3 and quality trimmed (default parameters) prior to assembly. A total of 60 million paired reads and 33 million single end reads were used in de novo assembly plug in (version 3.03), again using default parameters with a minimum contig size of 500 bp. Low coverage (<40×) contigs (compared to the genome average of 179×) were excluded from further analyses. BLASTn comparison to the F. graminearum mitochondrial sequence and coverage information (3,500–6,000×) was used to separate nuclear and mitochondrial sequence. 29 contigs with total length 107 Kbp were identified in this process and excluded from the genome annotation but are included in the F. pseudograminearum genome submission to GenBank. This Whole Genome Shotgun project has been deposited at DDBJ/EMBL/GenBank under the accession AFNW00000000. The version described in this paper is the first version, AFNW01000000.

Comparison to the F. graminearum Genome

Repeat masking was performed using RepeatMasker version open-3.2.8 [50], run in sensitive mode with cross_match version 0.990329 as the search engine and RepBase version 20090604 [27]. The species descriptor “fungi” was used.

Whole genome alignments were performed with the NUCmer and PROmer algorithm in the MUMmer package [51]. The minimum cluster length was set to 100 bp. The total length of the genome alignment between F. graminearum and F. pseudograminearum was calculated by summing the length of all non-redundant alignments extracted by using the show-coords program (part of the MUMmer package).

Gene Prediction

Protein coding genes were ab initio predicted in the F. pseudograminearum genome using FGENESH [52] based on the F. graminearum gene models as part of the MolQuest2 package from Softberry, AUGUSTUS [53] and GeneMark-ES version 2 [54]. AUGUSTUS was run with a training set of F. graminearum genes that had 100% nucleotide alignment and identity to the F. pseudograminearum genome. When the three programs disagreed in the splicing for the same gene, the following strategy was used to prioritize the predictions. When two of the three programs agreed, this prediction was taken. When all three programs differed for what appeared to be the same gene, priority was given in the order FGENESH, GeneMark and AUGUSTUS. Predicted transcripts encoding <20 amino acids were removed manually from the main prediction set. In total 17,503 unique transcripts were predicted, which after removing 4833 alternative predictions for the same gene left a putative set of 12448 unigenes.

BLASTmatrix Pipelines to Identify Genes Specific to F. pseudograminearum or Discontinuously Distributed in Fungi

The proteins coded by 27 different fungal genomes were downloaded from the Broad Institute (www.broadinstitute.org), Joint Genome Institute [55] or NCBI for the F. oxysporum strain 5176 genome [56]. Reciprocal best BLASTp hits were used to determine the orthologous protein relationships between F. pseudograminearum and each of the 27 fungal protein sets. This was performed using a pair-wise all versus all BLASTp analysis was conducted on the National Computational Infrastructure Specialized Facility in Bioinformatics Cluster located at the University of Queensland, Australia. Perl scripts obtained from http://sysbio.harvard.edu/csb/resources/computational/scriptome/unix/Protocols/Sequences.html were used to extract best reciprocal hits and the associated blast score from each species comparison BLASTp outputs. The F. pseudograminearum–F. graminearum relationship was treated as a special case and used to curate the dataset. In total, 11481 genes had clear one-to-one relationships between the two species, but differences in gene prediction (eg splice sites or predictions split/fused with respect to each other in each of the genomes) resulted in false identification of genes without orthologous matches in this analysis. These were accounted for by performing BLASTn analysis using F. pseudograminearum gene sequences containing introns as the query against the F. graminearum genomic sequence. Transcripts showing strong matches to the F. graminearum genome (match length greater than 50%) were classified as having different annotation between F. graminearum and F. pseudograminearum. 456 genes were removed from the analysis using this process.

Logical formulae and filtering functions in Microsoft Excel were used to identify potential cereal pathogen specific proteins via the presence of reciprocal best BLAST hits in one or more of the 16 cereal pathogens at a BLAST bit score of greater than 200 and in the complete absence of a reciprocal best BLAST hit in any of the 11 non-cereal pathogens. Manual curation of putative cereal pathogen specific genes was conducted to account for the possibility that genes were predicted in F. pseudograminearum that were missed in the annotation of other genomes. This was performed by querying the intergenic sequences of the nearest non-cereal pathogen, F. oxysporum f. sp. lycopersici in a tBLASTn search using the putative cereal pathogen specific proteins. Through this process, 112 proteins were removed due to potential missed predictions in the F. oxysporum f. sp. lycopersici genome.

Phylogenetic Analyses

Phylogenetic analyses were implemented using phylogeny.fr [57]. Briefly, multiple alignments were generated using MUSCLE [58] with default parameters, and curated using Gblocks [59]. Phylogeny was performed using PhyML [60] with the WAG amino acid substitution matrix [61] using an approximate likelihood-ratio test for branch support [62]. Trees were drawn using TreeDyn [63]. Trees were exported to Adobe Illustrator to allow shading of fungal branches.

Hybridization Analysis

A slot blot membrane was prepared using a Hoefer PR 648 apparatus using Amersham Hybond-XL membrane (GE Healthcare) prewetted with 0.2 M NaOH. Approximately 500 ng of DNA was prepared in 0.2 M NaOH to a final volume of 60 µL and incubated at 37°C for 15 min prior to application to the membrane. Following application of the DNA the membrane was cross-linked using a GS GeneLinker UV cross-linker (BioRad) with 150 mJ of energy. Two identical membranes were prepared and probed with either a FpAH1 or rRNA probe. Probes were generated using the PCR DIG Probe synthesis kit (Roche) according to the manufacturer's instructions using primers FpAHDiversityF and FpAHDiversityR for the FpAH1 probe and ITS1-F and ITS4 primers for the rRNA probe (Table S4). Hybridization and washing was performed with the DIG Easy Hyb solution and DIG wash and block buffer set (Roche). Detection was performed with the DIG Luminescent detection kit (Roche).

FpAH1 and PnAH1 Diversity Analysis

For Phaeosphaeria spp. PCR amplification was performed in 20 µL reactions containing 0.05 µM of PnAHDiversityF and FpAHDiversityR (supplied by Microsynth), 1× Dream Taq Buffer (Fermentas), 0.4 µM dNTPs (Fermentas) and 0.5 units of Dream Taq DNA polymerase (Fermentas). For F. pseudograminearum isolates, the forward primer was replaced with FpAHDiversityF and PCR was conducted using Invitrogen Taq DNA polymerase as per the manufacturer's instructions. Primer sequences are listed in Table S4. The PCR cycle parameters were: 2 min initial denaturation at 96°C followed by 35 cycles of 96°C for 30 s, 58°C for 45 s and 72°C for 1 min and a final 5 min extension at 72°C. The same cycling conditions were used for F. pseudograminearum isolates except the denaturation was performed at 94°C. Fusarium-derived PCR products were purified using the QIAgen PCR purification kit and sequencing was carried out by the Australian Genome Research Facility using both the forward and reverse primers. Sequencing reactions for Phaeosphaeria-derived products were conducted in 10 µL volume using the BigDye Terminator v3.1 Sequencing Standard Kit (Applied Biosystems, Foster City, CA) with both the forward and the reverse primer. The cycling parameters were 96°C for 2 min followed by 55 or 99 cycles of 96°C for 10 s, 50°C for 5 s and 60°C for 4 min. The products were cleaned with the illustra Sephadex G-50 fine DNA Grade column (GE Healthcare) according to the manufacturer's recommendations and then sequenced with a 3730×/Genetic Analyzer (Applied Biosystems). Alignment of forward and reverse sequences for each isolate was performed in SeqScape software V2.5 (Applied Biosystems). Translation and identification of protein haplotypes was also performed using this software. The program TCS v1.2 was used to visualize the most-parsimonious haplotype network [64].

Construction of FpAH1, SnAH1 and FpDLH1 Mutants

The vectors for deletion of the FpAH1 and FpDLH1 genes were constructed using lambda phage mediated recombination as described previously [65] using primers listed in Table S4. For FpAH1 the targeting construct consisted of 1858 bp of sequence immediately upstream of the start codon and 1132 bp covering the 3′ end of the gene and downstream sequence. 635 bp were deleted. For FpDLH1 the targeting construct consisted of 1523 and 1898 bp flanks, deleting 891 bp of the gene, leaving 125 bp of the 5′ coding sequence and 22 bp of the 3′ coding sequence. F. pseudograminearum protoplasting and transformation was performed as previously described for F. graminearum [66] except the STC (sorbitol, Tris and calcium chloride solution) was made with 0.8 M sorbitol instead of sucrose and the 40% PEG₈₀₀₀ solution also contained 0.8 M sorbitol.

Transformants were selected on 50 mg L⁻¹ geneticin (Sigma, St. Louis, MO, USA), subcultured onto geneticin and DNA was prepared using the REDextract and Amp kit as per the manufacturer's instructions (Sigma). Transformants were screened for gene deletion using AHKOscr1, AHKOscr2 and gpdAr primers. In total, 42 transformants were screened and six were identified as carrying a deletion of FpAH1. Mutant strains were single-spored and then stored as water cultures from ½ PDA plates.

The PnAH1 deletion construct was made using overlap PCR as previously described [67]. PnAH1KO5'F and PnAH1KO5'R were used to amplify a 1456 bp region upstream of PnAH1 whilst PnAH1KO3'F and PnAHKO3'R amplified a 1519 bp region downstream. These flanks were fused using overlap PCR with a hygromycin resistance cassette amplified from pAN7.1, resulting in a deletion construct of 5.9 kb. The deletion cassette was transformed into P. nodorum isolate SN15 protoplasts as previously described [68] and screened by using primers designed outside the flanking DNA (PnAH1KOscr-F and PnAH1KOscr-R). Copy number of the transformed construct was determined as previously described [69].

Complementation of FpAH1 Mutant

The ΔFpAH1 mutant was complemented by co-transformation of ΔFpAH1 mutant 96T492 with pAN7.1 [70] and a construct designed to expresses FpAH1 under the control of the Aspergillus nidulans TrpC promoter. This construct was created by PCR amplification of the coding sequence and 341 bp of terminator of FpAH1 from genomic DNA of isolate CS3096 with forward and reverse primers (AH-OXf and AH-OXr) containing ClaI and EcoRI sites respectively and replacement of the hygromycin cassette in pUChph [71] using the same restriction sites. The TrpC-FpAH1 section of the construct was fully sequenced prior to transformation into ΔFpAH1 mutant 492. Selection was with hygromycin at 200 mg L⁻¹. To verify that transformants were over-expressing the FpAH1 gene after successive rounds of sub-culture and single-sporing, they were grown in 6-well tissue culture plates in liquid basal media with 10 mM (NH₄)₂HPO₄ as the nitrogen source [72]. Mycelia were harvested, freeze-dried and RNA extracted using TRIzol (Invitrogen) according to the manufacturer's instructions after four days of growth at 28°C in the dark. Relative FpAH1 gene expression was measured using quantitative reverse transcriptase PCR as previously described [73]. β-tubulin and FpAH1 (FpAH1f and FpAH1r) primers are shown in Table S4. Expression of FpAH1 in the two transformants used was ∼3–5-fold higher than the wild type strain during in vitro culture and undetectable in the mutant as expected.

Growth Rate Determination

A microtitre plate assay to monitor growth was performed based on the method described by Schmelz [74]. Briefly, each well contained 200 µL of basal media [72] with 5 mM glutamine as the nitrogen source with a final spore concentration of 1×10⁴ spores mL⁻¹. Absorbance at 405 nm was measured using an iEMS microplate reader. For the FpAH1 mutant test the plate remained in the instrument for the duration of the assay at ambient temperature. In the case of the FpDLH1 mutant assay after an initial reading the plate was incubated at room temperature and readings commenced 22 hours post inoculation.

Fusarium Crown Rot Virulence Assays

Assessment of fungal virulence during FCR was carried out using the method described previously [36]. For virulence testing of FpAH1 mutant and complemented strains, isolates were inoculated onto 14 cm Synthetischer Nährstoffarmer Agar plates from potato dextrose agar plugs stored as water cultures and allowed to grow for 14 days under 12/12 hour day/night cycle under white and black fluorescent light at room temperature (∼22°C). Spores were harvested by flooding the plates with water and adjusted to the same concentration. To test the role of FpDLH1 in virulence, spores were produced in shaking 1 L flasks containing 100 mL of 25% Campbell's V8 juice inoculated with a single plug taken from a 20% V8 juice plate. Flasks were incubated at room temperature for 8 days and spores harvested by filtration through miracloth followed by centrifugation (5,500×g) to remove media. Spores were resuspended in sterile water, counted and adjusted to the same concentration. Seeds were plated out two days prior to inoculation on wet paper towel in a 14 cm Petri dish and allowed to germinate on the laboratory bench. Germinated seeds were transferred to a 50 mL falcon tube containing 2 mL of the spore suspension and rolled gently to coat the seeds. Six to ten seeds were placed in a single paper towel, rolled up and tapped closed and placed in a jar containing water. Paper towel rolls were kept moist throughout the experiment. For the analysis of FpAH1 mutants a concentration of 1×10⁵ spores mL⁻¹ was used and plants were maintained on a laboratory bench without supplementary lighting. To test complemented FpAH1 mutants and FpDLH1 mutants, 4×10⁵ spores mL⁻¹ were used and plants were maintained in a closed bench top biohazard cabinet with a cool white fluorescent light suspended from the roof of the cabinet (approximately 15 cm from the plants) providing 16 hours of light per day. Assays were scored by recording the number of plants that were alive as a proportion of the total plants.

Fusarium Root-Rot Virulence Assays

Fungal strains were plated from PDA plug water storage onto 20% Campbell's V8 juice 1.2% agar plate and incubated for seven days at room temperature under white and black fluorescent light on a 12/12 hour day/night cycle. Seed for the assay were surface sterilized by soaking for 5 minutes in a 0.64% sodium hypochlorite-10% ethanol solution, followed by several rinses in sterile distilled water. Seed were plated onto three sheets of pre-wetted Whatmann no 3 12.5 cm filter disks in a 14 cm Petri dish. Seeds were incubated in the dark for 5 days at 4°C prior to germination at 20°C also in the dark. Germinated seedlings were distributed to Petri plates assembled with three filter papers wetted with 20 mL of sterile water prior to inoculation. Each plate contained 15–16 seedlings and was used for inoculation with one isolate. Inoculum consisted of agar plugs taken using either an inverted 1-mL pipette tip or a number 4 (6 mm) cork borer from the edge of the growing V8 agar plate. Plugs were placed on a single root per seedling about 1 cm below the seed with the mycelia in direct contact with the root. Plates were sealed with sealing film (PhytoTechnology Laboratories), incubated in a Thermoline illuminated incubator for 6 days at 20°C with 12 hours of light provided by fluorescent lamps. Assays were scored by measuring the length of the shoot.

Phaeosphaeria nodorum Virulence Assays

Parental and mutant strains of P. nodorum were tested on the susceptible wheat cultivar Amery as previously described [75]. Latent period sporulation was assessed using detached leaves as described previously [76].

In planta Expression Assays

Gene expression during infection of wheat and barley by F. pseudograminearum was performed using both a detached leaf infection assay and a root infection time course. For the latter, plants were inoculated as described for the root-rot virulence assay and root segments were harvested into liquid nitrogen at 24, 48 and 96 hours post inoculation. For expression in detached leaves, fourth leaf segments (7–8 cm long) of both hosts were taken from glasshouse-grown plants and each end of the cut leaf was sandwiched between water agar in a 14-cm Petri plate. Leaf segments were pierced at two points in a central region using a 200 µL pipette tip. A spore suspension of the wild type isolate (1×10⁶ sp mL⁻¹) was placed on the wound sites and the plate was sealed with sealing film. Plates were maintained in a fluorescent bulb lit growth chamber (Thermoline) at 20°C with 12 hours of light. The whole leaf segment was harvested for RNA extraction. RNA extraction was performed using TRIzol reagent (Invitrogen) according to the manufacturer's instructions. Relative expression was compared to β-tubulin. Primers used for gene expression analysis are shown in Table S4. qRT-PCR was performed as previously described [73].