This study identified genes that distinguish Australian Fusarium oxysporum f.sp. vasinfectum (Fov) isolates from related co-localized non-pathogenic F. oxysporum isolates and from non-Australian Fov isolates. One gene is a homologue of the F. oxysporum f.sp. lycopersici (Fol) effector gene SIX6, encoding a 215-residue cysteine-rich secreted protein. The Six6 proteins from Fol and Fov contained eight conserved cysteine residues, five of which occurred in the highly diverged 48-amino-acid region where FovSix6 differs from FolSix6 at 32 residues. Two other potential effector genes, PEP1 and PEP2, were identified in a cDNA library of Fov genes expressed during infection of cotton. The presence of FovSIX6 and other differences in DNA fingerprints clearly distinguished Australian Fov isolates from non-Australian Fov isolates and these differences further support the hypothesis based on earlier phylogenetic analysis that Australian Fov is different from Fov in other cotton-growing areas. A specific diagnostic for Fov based on FovSIX6 is described.
Fusarium oxysporum is a soilborne fungal species that includes both plant-pathogenic and non-pathogenic forms. Although this species causes wilt disease on a wide range of plant species, individual isolates have narrow host ranges and these isolates are classified at the subspecific level as formae speciales based mainly on host preference or specificity. For example, the tomato wilt pathogen is classified as F. oxysporum f.sp lycopersici (Fol) and the cotton wilt pathogen as F. oxysporum f.sp. vasinfectum (Fov). Molecular studies of the basis of pathogenicity and host range are most advanced for Fol, where large-scale investigations based on genome sequencing and directed and random insertional mutagenesis have identified genes that are required for virulence (Di Pietro et al., 2003; López-Berges et al., 2009; Michielse & Rep, 2009; Michielse et al., 2009; Ma et al., 2010). Additionally, secreted effector proteins of Fol have been identified by proteomic analysis of xylem sap from tomato plants infected with Fol and are called secreted in xylem (Six) proteins. Six1 was identified as a cysteine-rich 12-kDa protein that is processed from a larger secreted precursor protein and is the avirulence protein recognized by the tomato resistance gene I-3 and a virulence factor in the absence of I-3 (Rep et al., 2004, 2005). Three additional Fol-encoded secreted proteins: Six2, Six3 and Six4 (Houterman et al., 2007) have also been described. Six3 is the avirulence protein recognized by the tomato resistance gene I-2 (Houterman et al., 2009) and Six4 is the avirulence protein recognized by the tomato resistance genes I and I-1 and also has a virulence function in that it suppresses resistance mediated by tomato resistance genes I-2 and I-3 (Houterman et al., 2008). Additional SIX genes (SIX5 to SIX7) have been identified in Fol but not functionally characterized (Lievens et al., 2009; M. Rep, unpublished data). The full genome sequence of Fol isolate 4282 is allowing bioinformatic identification of potential effector genes (Ma et al., 2010). Fifteen chromosomes have been identified in Fol. Comparison of the Fol genome with that of a ‘sister’ species, Fusarium verticilloides, indicates high genome sequence conservation between the two species, except for Fol chromosomes 3, 6, 14 and 15, which have no counterparts in F. verticilloides. Chromosome 14 of Fol harbours all but one of the SIX effector genes, including SIX6, and consequently is referred to as a ‘pathogenicity’ chromosome (Ma et al., 2010).
Effectors may be defined as pathogen proteins and small molecules that alter host-cell structure and function. These alterations either facilitate infection (virulence factors and toxins), or trigger defence responses (avirulence factors and elicitors) or both” (Hogenhout et al., 2009)
The rapid advances in Fol research provide a model for identifying the basis of pathogenicity and host range of other F. oxysporum formae speciales, such as Fov, the cotton wilt pathogen, which occurs worldwide in cotton-growing areas, including Australia (Kochman, 1995). The present work studied Australian isolates belonging to two specific Australian vegetative compatibility groups, VCG01111 and VCG01112 (Bentley et al., 2000). A panel of Fov and non-pathogenic F. oxysporum isolates from cotton fields was set up for association genetics analysis, comparing the presence and copy number of candidate pathogenicity genes with virulence towards cotton, with a view to finding genes specifically associated with Australian Fov isolates to provide diagnostic markers for the identification of Fov in environmental samples.
Materials and methods
Fov isolate 24500 (VCG01111) was used as a reference isolate in this study for infection of cotton plants. For the non-pathogenic samples, 10 F. oxysporum isolates obtained from cotton field soils in two regions of New South Wales and Queensland were used. These isolates were chosen on the basis of AFLP-based molecular phylogenetic analysis that indicated that (i) the isolates were genetically distinct and (ii) among the non-pathogens in the collection, these 10 isolates had the closest genetic relationship with the pathogenic isolates (so called lineage A isolates, B. Wang, unpublished data). Ten F. oxysporum isolates from native Australian Gossypium spp. were recovered from a survey from across northern Australia and were also used in the panel. Among these, two were weakly pathogenic on cotton, inducing vascular discoloration, and one was mildly pathogenic, inducing leaf necrosis in infected seedlings. An isolate of Fol from Australia provided by Dr David Jones, of the Australian National University, was also included for comparison. The isolates are listed in Table 1. DNA from another 38 F. oxysporum isolates collected from various parts of Australia and representing 14 formae speciales groupings were kindly provided by Matthew Laurence, Sydney University (email@example.com) (listed in Table 2 with M. Laurence isolate numbers). DNA from six Fov isolates representative of six races and five VCG groups found in various parts of the world other than Australia and which differed from the Australian Fov isolates in various phylogenetic analyses were obtained from Dr Rebecca Bennett, USDA-ARS, Shafter, CA, USA (Table 3). These isolates are described by Kim et al. (2005). DNA of SIX4 was kindly provided by Dr Donald Gardiner, CSIRO Plant Industry, Brisbane, Queensland.
Table 1. Australian Fusarium oxysporum f.sp. vasinfectum (Fov) and other F. oxysporum (Fo) isolates used for association studies between virulence and candidate ‘pathogenicity’ genes
aWeakly pathogenic to cotton.
Fov isolates of VCG 01111
Darling Downs, QLD
Darling Downs, QLD
Darling Downs, QLD
Darling Downs, QLD
Darling Downs, QLD
Darling Downs, QLD
Fov isolates of VCG 01112
Non-pathogenic F. oxysporum isolates from cultivated cotton fields
Darling Downs, QLD
Darling Downs, QLD
Darling Downs, QLD
Darling Downs, QLD
Darling Downs, QLD
F. oxysporum isolates from soils with native Gossypium sp.
Mt Isa, QLD
Mt Isa, QLD
Mt Isa, QLD
Alice Springs, NT
Alice Springs, NT
Alice Springs, NT
Alice Springs, NT
Alice Springs, NT
Flinders Ranges, SA
Flinders Ranges, SA
Table 2. Additional Fusarium oxysporum isolatesa from which genomic DNA was analysed
Table 3. Isolatesa of Fusarium oxysporum f.sp. vasinfectum from outside Australia from which genomic DNA was analysed
aOrigins and references in Materials and methods.
bVegetative compatibility group.
Isolation of Fov from infected plant tissue
Root and stem tissues from cotton plants showing wilt symptoms were collected from cotton fields. Control plants were collected separately from cotton-growing regions where Fov does not occur (Narrabri, New South Wales). Fov was isolated from infected tissue on potato dextrose agar (PDA) plates containing 500 mg streptomycin sulphate L−1 incubated at 28°C.
Rapid DNA isolation from fungal colonies for PCR analysis
Fungal DNA was isolated from 0·5-cm-square fungal mats from PDA using the REDExtract-N-Amp plant PCR kit (Sigma Aldrich) in100 μL extraction buffer and 4 μL extract were used for PCR.
Large-scale DNA isolation
For DNA isolation, fungal isolates were cultured in potato dextrose broth (PDB) for 4–5 days at 28°C with constant shaking at 125 r.p.m. Fungal mycelia were harvested by passing through three layers of Miracloth and blotting onto Whatman filter paper no 1. Total DNA was isolated using a CTAB method (Saghai-Maroof et al., 1984) with minor modifications. Following isopropanol precipitation and suspension in 500 μL TE (10 mm Tris, 1 mm EDTA; pH 8·0) the DNA was treated with RNase at 100 μg mL−1 at 37°C for 3 h followed by phenol:chloroform:isoamyl alcohol (25:24:1) and chloroform extractions and ethanol precipitation, and dissolved in TE buffer. Cotton DNA was isolated from young leaf tissue of cotton cv. Siokra 1–4 plants using the QIAGEN Plant DNA extraction kit.
Unless otherwise stated all PCRs were performed using 25 ng genomic DNA (50 ng for cotton) as template, primers at 0·5 μm each, dNTPs at 100 μm each and 1 U Taq polymerase (MBI, Fermentas). The amplification parameters were: 2 min at 94°C; then 30 cycles at 94°C for 30 s, 55°C for 30 s, 72°C for 1 min; then 5 min at 72°C. PCR products were analysed by agarose gel electrophoresis on 1% agarose gel run at 5 V cm−1 with 1× TAE buffer.
Sequencing of PCR products
Appropriately sized bands were excised from gels and eluted using a Qiagen Gel Purification kit according to the manufacturer’s instructions. Of the PCR product, 50–200 ng were used for sequencing with appropriate primers using the Big Dye Terminator sequencing kit. Sequences of genes described in this work were deposited in GenBank (NCBI) with the accession numbers indicated in brackets: FovSIX6 (HM467128), FonSIX6 (HM467129), SIX6 (HM467130), PEP1 (HM467131) and PEP2 (HM467132).
DNA gel botting and hybridization
For DNA gel blot and hybridization experiments, ∼10 μg fungal genomic DNA was digested with 10 U HindIII (MBI Fermentas) for 5–6 h. The digested DNA was separated on 1% agarose gel using TAE buffer and blotted onto nylon membranes by alkaline transfer. PCR products (directly sequenced for verification) were used as probes and radioactively labelled using α-P32-dCTP according to standard procedure (Sambrook et al., 1989). Pre-hybridization and hybridization was carried out in hybridization buffer (0·5 m sodium phosphate pH 7, 7% SDS, 1% BSA) and washing was performed at 65° using 2× SSC (Sambrook et al., 1989), 0·1% SDS and 1× SSC, 0·1% SDS.
Fov protoplast preparation
For pulse-field gel electrophoresis, protoplasts were prepared from freshly germinated conidia according to Mes et al. (1999) with some modifications. PDB (100 mL quarter-strength) was inoculated with spores, shaken (5 days, 125 r.p.m., 25°C) and conidia were collected by filtering through one layer of Miracloth followed by centrifugation (3500 g, 15 min, room temperature). Washed conidia were resuspended in water (0°C) and counted with a haemocytometer, then 5 × 108 conidia were germinated in 40 mL PDB for 20–24 h and harvested by centrifugation (3500 g, 15 min), washed with 1 m NH4Cl and incubated in 10 mL Driselase solution (2–5% w/v in 1 m NH4Cl) at 30°C for 18 h with constant shaking. Protoplasts were filtered through two layers of Miracloth, chilled on ice and four volumes of cold 1 m sorbitol added. Protoplasts were harvested by centrifugation (850 g, 15 min, 4°C), washed and resuspended in 1 mL resuspension solution (1 m sorbitol, 50 mm CaCl2, 10 mm Tris HCl, pH 7·4) at 4°C.
Contour-clamped homogeneous electric field (CHEF) gel analysis
For CHEF gel analysis protoplast plugs were as described by Mes et al. (1999). Chromosomes were separated by electrophoresis on a 1% Seakem Gold agarose gel in 0·5× TBE buffer using a BioRad Chef Mapper XA. The run was performed at 10°C for 12 days with field strength of 1·5 V cm−1, included angle of 120°, and initial and final switch times of 20 min and 80 min, respectively, in a linear ramping pattern. The buffer was continuously circulated and changed every 2 days. Saccharomyces pombe and Hansenula wingei chromosomes (BioRad) were used as size markers during the run. After the run the gel was stained with ethidium bromide for 1 h and destained for 24 h before being photographed. The DNA was transferred to Hybond N+ membranes (Amersham) following the alkaline blotting method (Sambrook et al., 1989) with an additional 15 min of depurination and blotting for 48 h. Hybridization of the membranes was carried out as described before.
Presence of a SIX6 homologue in Fov isolates from both VCG01111 and VCG01112
Primers (Table 4) designed to PCR amplify coding regions of secreted proteins from Fol [seven SIX genes and ORX1, which encodes a secreted oxidoreductase enzyme with an unknown substrate (Houterman et al., 2007)] were used to screen DNA from a panel of Australian F. oxysporum isolates (Table 1) for homologues of these genes. The panel included 11 Fov isolates in VCG01111, nine isolates in VCG01112, 10 non-pathogenic F. oxysporum isolates collected from cultivated cotton fields (Fo-CC) and 10 isolates from Australian native Gossypium species (Fo-WC). Whilst all of the SIX genes (except SIX4, which was absent from the Australian Fol isolate used here) were amplified from Fol, only SIX6 was amplified from Fov (Fig. 1a). The presence of SIX6 in Fov alone was also confirmed by DNA gel blot analysis (Fig. 1a). As in Fol, the gene appeared to occur as a single copy in Fov. SIX6 was not detected in any non-pathogens or isolates from wild Gossypium species (Fig. 1a). The absence of SIX1–SIX4 in the isolates on the panel was confirmed by DNA gel blot analysis using the Fol genes as probes.
Table 4. Primer sequences used for PCR in this study
To gauge how widespread SIX6 is among F. oxysporum, amplification of SIX6 was carried out on a set of DNA samples from a range of Australian isolates of F. oxysporum, including 14 different formae speciales (Table 2). In addition to the Fov and Fol isolates in the collection, SIX6 was amplified from three F. oxysporum f.sp. passiflorae (Fop; 695, 710, 711) and two of the four F. oxysporum f.sp. niveum (Fon) isolates (546, 704) (Fig. 1b). PCR products from Fov, Fop and Fon were sequenced and predicted protein products compared to FolSix6. There was no amino acid variation within different forma specialis groups (including the two VCG groups in Fov), but considerable variation between the groups (Fig. 1c). Henceforth, the variants are referred to as FolSIX6 (GenBank FJ755835), FovSIX6, FopSIX6 and FonSIX6. FovSIX6 encodes a 214-amino-acid protein compared to the other 215 residue proteins. All eight cysteine residues, including one in the predicted signal peptide, are conserved. Five of these conserved cysteines occur in the most divergent 48-amino-acid region between residues 62 and 110, where FovSix6 differs from FolSix6 at 32 sites (boxed in Fig. 1c). FonSix6 is identical to Six6 from F. oxysporum f.sp. melonis (Fom) and F. oysporum f.sp. radicis-cucumerinum (Lievens et al., 2009).
The region of divergence between FovSIX6 and the other SIX6 genes allowed the design of a specific PCR primer FovSIX6-F2 that, together with SIX6-R1 (Table 4), amplified a SIX6 gene fragment only from Fov isolates and not from other formae speciales that carry SIX6 homologues (Fig. 1d). The PCR test was applied to DNA isolated from cotton plants from uninfected fields and plants with symptoms from infected fields, but no FovSIX6 product was amplified, probably because of low Fov biomass in infected samples. However, FovSIX6 products were detected in diseased plants after seminested PCR; the first reaction of 35 cycles using general SIX6 primers SIX6-F1/SIX6-R1 followed by 35 cycles of PCR using the product of these reactions with the specific primer pair FovSIX6-F2/SIX6-R1 (Fig. 1d).
Putative effector proteins PEP1 and PEP2 in Fov
To identify additional Fov effector genes, a bioinformatic approach was used to screen a sequenced cDNA library of 2100 clones isolated from cotton seedlings infected with Fov under tissue-culture conditions (McFadden et al., 2006). Genes were sought which, like the SIX genes of Fol, encoded secreted proteins with two or more cysteine residues, had no database representatives outside of the genus Fusarium (Houterman et al., 2007) and, on the basis of presence and absence, could distinguish Fov from the non-pathogens on the panel of isolates. Among the 2100 clones, 334 unigenes had homology to Fusarium genes (http://www.broadinstitute.org/annotation/genome/fusarium_group/). Of these 334, 70 encoded secreted proteins (predicted using the signalp program, Emanuelsson et al., 2007). Thirty of these that had no predicted function and two or more cysteine residues were chosen for further analysis. An additional 31 genes were selected from the library that encoded secreted proteins with two or more cysteines but with no match to any current database entry (i.e. potentially Fov or cotton proteins). The 61 genes were tested by PCR using Fov and cotton DNA as templates. PCR products amplified from Fov only were then tested for amplification against the Fusarium DNA panel. Among the 61 candidates two putative effector protein genes, named PEP1 (GenBank accession CD485943) and PEP2 (GenBank accession CD486033) were amplified from Fov, but not from non-pathogens (Fig. 2a).
PEP1 encodes a predicted secreted protein of 270 amino acids (eight cysteines). A homologue of PEP1 is present in the Fol genome sequence as a single copy gene on chromosome 14, interrupted by a retrotransposon. PEP1 was detected by DNA gel blot analysis (Fig. 2b) in all Fov isolates and a single isolate from native Australian Gossypium species. In Fov PEP1 appears to be an intact expressed gene and member of a small gene family (Fig. 2b). DNA gel blot hybridization with the PEP1 probe distinguished VCG01111 from VCG01112 isolates. Although the number and size of fragments detected in the two groups was identical, one fragment detected in VCG01111 consistently hybridized more strongly than the corresponding fragment in VCG01112 (Fig. 2b). PEP2, which encodes a secreted protein of 86 amino acids (10 cysteines) and is not present in the Fol genome, was detected in all Fov isolates and in four isolates from wild Gossypium species.
SCD1 and FTF1 gene families in Fov
McFadden et al. (2006) identified a Fov gene highly expressed during infection (GenBank accession CD485533). This gene, originally named AtsC on the basis of similarity to a gene in Agrobacterium tumefaciens, was renamed here as SCD1 because of the higher similarity of the deduced protein to a large and widespread group of short chain dehydrogenase enzymes also known as the reductase, epimerase, dehydratase (RED) or short-chain dehydrogenase/reductase (SDR) family. DNA gel blot analysis detected a single copy of the gene in the non-pathogenic isolates and Fol (Fig. 3). Database searches also detected one copy of SCD1 in Fol and F. verticilloides. In contrast, DNA gel blot analysis showed that SCD1 is part of small gene family of six to eight members in Australian Fov isolates (McFadden et al., 2006; Fig. 3). The expanded family size of this gene in Fov provides a diagnostic DNA fingerprint for Australian Fov isolates. Sequence analysis of cloned cDNA copies of SCD1 indicated that at least six different members of the family are expressed. These sequences, plus earlier ones (McFadden et al., 2006) were assembled into six separate contigs, four of which contain complete open reading frames and one lacks the first 91 codons. The deduced proteins are 96–99% identical and the full-length SCD proteins are 265 amino acids long, similar to closely related proteins from other organisms. The catalytic activity of these enzymes is unknown and interrogating the GenBank database gives few hints as to their function. The top 250 protein matches in a blast search are all annotated by homology with generic terms such as putative, hypothetical and short chain dehydrogenase. When confining the search to metazoan entries the closest match to a verified activity was to a 20 β-hydroxysteroid dehydrogenase from trout (Oncorynchus mykiss NP_001118068·1) with non-specific carbonyl reductase activity. The trout enzyme is 26% identical to the Fusarium proteins over the 265-amino-acid length of the SCD protein. Comparing the predicted structure of the SCD proteins with the nearest structures in the Protein Database (1VL8 and 3BHI) indicates that the amino acid differences between the Fov SCDs are on or close to the surface of the proteins and unlikely to result in changes in substrate specificity. There are single copies of putative SCD1 orthologues in Fol and F. verticilloides with 83% and 74% amino acid identity, respectively, to SCD1 from Fov, whilst F. graminearum seems to lack an orthologue. The next closest homologues present in all three Fusarium genome sequences are 42–48% identical.
Fusarium transcription factor 1 (FTF1) was identified as a member of a gene family highly expressed during infection of common bean by F. oxysporum f.sp. phaseoli but not by non-pathogens (Ramos et al., 2007; DQ280313). The gene is a member of a family of related genes in Fol with three of the 11 members occurring on ‘pathogenicity’ chromosome 14 (Ma et al., 2010). DNA gel blot analysis (Fig. 3b) shows that this gene family has at least 12 members in Fov and three to five members in the non-pathogens from cotton fields. Although in this case the complex DNA fingerprint did not distinguish the two VCG groups, it did distinguish Australian Fov isolates from all other isolates and the fingerprint was highly polymorphic across all other tested F. oxysporum isolates (Figs 3 and 4).
Comparison of Australian Fov isolates with Fov isolates from outside Australia
The generality of FovSIX6 and other diagnostic markers for Fov was tested using DNA samples from six Fov isolates of different VCG groups isolated from cotton-growing regions outside of Australia. Using PCR primers SIX6-F1 and SIX6-R1 (Table 4) no SIX6 homologues were amplified from the non-Australian Fov isolates. DNA gel blot analysis with the Fov homologue of SIX6 detected weak hybridization to one isolate (R1, Fig. 4a). Given the high DNA loading in this lane, it is unlikely that this fragment is closely related to SIX6. The Australian and non-Australian isolates were also distinguished in gel blot analysis by the other markers developed in this study; PEP2 did not hybridize to the non-Australian isolates, SCD1 detected a single fragment and FTF1 detected multiple fragments that were highly polymorphic (Fig. 4b). The PEP1 probe hybridized to four of the six non-Australian isolates with different fragment patterns than the Australian isolates.
Chromosomal locations of putative effector and pathogenicity associated genes
In Fol, chromosome 14 harbours all but one of the SIX genes (including SIX6) and three of the 11 copies of FTF1 (M. Rep, unpublished data). CHEF gels were used to separate Fov chromosomes and DNA gel blot analysis was carried out on three representative Fov isolates from each of VCG01111 and VCG01112 along with five non-pathogenic F. oxysporum isolates from cotton field soils (Fig. 5). The SIX6 probe hybridized to a ∼2·8-Mb chromosome in the Fov isolates and a smaller ∼1·6-Mb chromosome in the Fol isolate. In Fov isolate 24500, SIX6 was also detected on a ∼3·6-Mb chromosome, although only one fragment was observed in gel blot analysis of restriction enzyme-digested DNA. The same chromosome distribution was found for PEP2 in Fov isolate 24500. The presence of SIX6 and PEP2 on both a ∼2·8-Mb chromosome and ∼3·6-Mb chromosome may suggest duplication of a chromosomal region containing both SIX6 and PEP2 in this particular isolate. SCD1 also hybridized solely to a ∼2·8-Mb chromosome, which indicates that most and perhaps all copies of this gene occur on the same chromosome. In one VCG01111 isolate, a further SCD1 family member(s) was detected on a chromosome of ∼0·8 Mb, suggesting a possible translocation or duplication event in this isolate. Co-migration of the 2·8-Mb chromosome (or chromosomes) carrying SIX6, PEP2 and SCD1 was confirmed by overlaying the three separate autoradiographs. FTF1 and PEP1 hybridized to multiple chromosomes, but at least one homologue co-migrated with SIX6, PEP2 and SCD1. In Fol PEP1 and SIX6 hybridized to the same chromosome, which is consistent with the chromosome 14 location of single copies of these genes in the Fol genome sequence. Surprisingly, and in contrast with its location in Fol, the PEP1 probe hybridized to many and perhaps all the chromosomes of Fov, which on first consideration is inconsistent with the three fragments detected in restricted DNA in Fig. 2. One explanation consistent with the data in Figures 2 and 5 is multiplication of a region carrying PEP1 in Fov and translocation of copies to multiple chromosomes, perhaps subtelomeric regions as described for the SIX8 gene in Fol (Rep & Kistler, 2010). PEP1 duplications could carry all six base restriction enzyme recognition sites surrounding the PEP1 gene copy(ies) and thus give lower numbers of bands in these gels than in the chromosomal karyotype. Hence, variation in intensity of bands between gel blots of restricted DNA of VCG01111 and VCG01112 may reflect differences in copy number of genes associated with that particular size of DNA restriction fragment.
As a species, F. oxysporum has a wide host range, but individual isolates are restricted in host range. The understanding of the molecular basis of virulence and host range of F. oxysporum is being accelerated by the increasing availability of genome sequence information and data from insertional mutagenesis. A picture is emerging of two broad classes of genes being critical. The first group includes genes mainly encoding recognized enzyme functions, such as protein kinases and host cell-wall-degrading enzymes, that are necessary for the basic fungal machinery of virulence towards plants. These genes are probably present in all F. oxysporum pathogens (Di Pietro et al., 2003; Michielse & Rep, 2009; Michielse et al., 2009). The second group includes effector genes that encode small secreted proteins, commonly with no obvious enzymatic function, or related orthologues in current protein databases. At least 10 of these genes occur in the genome of the tomato pathogen Fol (Houterman et al., 2007; Lievens et al., 2009; M. Rep, unpublished data). Many of these effector genes have a narrow distribution within the species and may be involved in the determination of host-range differences between formae speciales. For example, Lievens et al. (2009) reported an extensive PCR-based survey of 270 F. oxysporum isolates for the presence of seven Fol effector genes (SIX1–SIX7). With the exception of SIX4 (Avr1), which showed a presence-absence polymorphism most likely driven by selection imposed by the corresponding tomato resistance gene I, these genes were present in all Fol isolates. In contrast, with the exception of SIX6 and SIX7, the remainder of the Fol SIX genes were not detected in 13 other formae speciales. Sequence variants of SIX6 were found in some isolates of Fom and F. oxysporum f.sp. radicis-cucumerinum and SIX7 in F. oxysporum f.sp. lilii. Two orthologues of SIX6 were also found in Fon and Fop. The gene sequences of SIX6 are identical within the isolates examined of each of these formae speciales (this study; Lievens et al., 2009). The present analysis of Australian isolates of Fov using the same PCR primers detected an orthologue of SIX6, but not of other SIX genes. However, in contrast to Fol, SIX6 was not conserved in all Fov isolates, being present only in Australian Fov isolates. Among the different SIX6 orthologues, the sequence in Fov (Fig. 1c) differed most substantially from the others, particularly in the region between amino acids 62 and 100. Interestingly, although 32 of the 48 residues differed between Fov and Fol, the five cysteine residues in this region were conserved. A similar level of diversity of residues occurring between conserved cysteine residues was observed for homologues of the cysteine-rich effector protein P4 in different species in the rust genus Melampsora (Van der Merwe et al., 2009). It is likely that the cysteine residues participate in disulphide bonds and thus play a specific structural role and that the intervening variable residues are involved in effector function and have evolved either to engage specific host targets or avoid recognition by resistance proteins.
The detection of most SIX genes in all Fol isolates but not other formae speciales could be the result of either variation between the PCR primer sequences and target sequences or absence of orthologues outside of Fol. The absence of SIX1–SIX4-related sequences in Australian Fov isolates was confirmed by DNA gel blot analysis. It is possible that each forma specialis harbours specific sets of unrelated effector genes involved in adaptation to specific host species, but further DNA gel blot analysis or genome sequencing is required to confirm this. The existence of large between-isolate variation in effector gene complements was recently highlighted by genome sequencing of two isolates of the blast fungal pathogen Magnaporthe grisea and the authors of that work proposed that the differences might be associated with the two different grass hosts of the two sequenced pathogen isolates (Yoshida et al., 2009). This underlines the importance of genome sequence analysis of more F. oxysporum pathogen genomes representing different formae speciales to determine the extent to which determinants of host specificity will be unique to each forma specialis.
The use of PCR markers based on the SIX genes is beginning to provide a means of robust molecular identification of Fol (van der Does et al., 2008) and even specific races of Fol (Lievens et al., 2009). Using sequence differences between the SIX6 homologues from Fov and Fol, a PCR marker was developed to specifically detect Fov and distinguish it from non-pathogenic F. oxysporum isolates in Australian cotton fields. Further, PCR and DNA gel blot analysis did not detect a close homologue of the Australian Fov SIX6 gene in Fov isolates from other parts of the world. These isolates were also distinguished from Australian Fov isolates by the SCD1, FTF1 and PEP1 markers (Fig. 5) and by PEP2 that did not hybridize to the non-Australian Fov isolates. These observations are consistent with proposals based on phylogenetic studies that Australian Fov isolates have evolved in Australia rather than having entered the country during or after the establishment of the local cotton industry in the 1960s (Fernandez et al., 1994; Davis et al., 1996; Bentley et al., 2000; Skovgaard et al., 2001; Wang et al., 2004, 2008; Abo et al., 2005; Kim et al., 2005; Gulino, 2007). Separate evolution of clades within different host-range groupings of F. oxysporum was also supported by molecular phylogenetic analysis for the banana pathogen, F. oxysporum f.sp. cubense (O’Donnell et al., 1998). It has been postulated that the Australian Fov may have evolved from F. oxysporum associated with the diversity of wild Gossypium species in the Australian flora (Wang et al., 2008). However, none of the ‘wild’ Australian isolates examined in the present study carried SIX6, or showed the amplification of the SCD1 gene family associated with Australian Fov isolates, and all had different DNA fingerprints with the FTF1 probe. Some did, however, carry sequences related to two putative Fov effector genes, PEP1 and PEP2.
The differences between Fov isolates and between formae speciales in general raises the intriguing question of how these asexual pathogens evolve. Recent data emerging from the genome sequence of Fol indicates that most of the SIX genes occur on a single pathogenicity chromosome and that this chromosome is capable of transmission between different isolates (Ma et al., 2010). This raises the possibility that adaptation of formae speciales of F. oxysporum to different host species, often polyphyletic within the host-range groupings, may evolve by transfer of virulence chromosomes between isolates followed by recombination, selection and diversification of suites of effector genes, optimizing adaptation to host species and environment. This is consistent with the presence of FovSIX6, PEP1 and the amplified SCD1 family on the same chromosome in Australian Fov isolates, including both VCG groups, and suggests a common origin of this chromosome in both VCGs, possibly by horizontal transfer from an as yet unidentified source.
This work was funded by the CottTech collaboration between CSIRO, Cotton Seed Distributors and the Cotton Research and Development Corporation. We thank Robyn East for providing excellent technical assistance and Matthew Laurence (University of Sydney), Dr Rebecca Bennett, USDA-ARS Shafter, CA, USA and Dr Mike Davis, Department of Plant Pathology, University of California, Davis, CA, USA for various Fusarium DNA samples used in this analysis. We also thank Dr Kathy Ophel-Keller and Dr Alan McKay (SARDI) for preparation and supply of environmental DNA samples from cotton tissue and soil.