Estimating frequencies of virulent isolates in field populations of a plant pathogenic fungus, Leptosphaeria maculans, using high-throughput pyrosequencing

Authors


Correspondence

Angela Van de Wouw, School of Botany, the University of Melbourne, Melbourne, Vic. 3010, Australia. E-mail: apvdw2@unimelb.edu.au

Abstract

Aim

To develop a pyrosequencing assay to monitor the frequency of alleles of an avirulence gene, AvrLm4, in populations of sexual spores of Leptosphaeria maculans, a fungal pathogen of canola (Brassica napus).

Methods and Results

The predominant mutation in AvrLm4 responsible for virulence to the corresponding resistance gene, Rlm4, is a single nucleotide polymorphism (SNP) at base 358. Pyrosequencing primers were designed to amplify a 90-bp region that included this SNP. The assay was developed and validated by analysing the frequency of AvrLm4 in isolate mixtures of different proportions. Furthermore, the frequency of avrLm4 (virulence allele) determined by pyrosequencing of populations of sexual spores was consistent with the frequency of avrLm4 determined by Sanger sequencing of the entire AvrLm4 gene from single isolates cultured from the same stubble.

Conclusion

This high-throughput assay can play an important role in predicting the risk of resistance breakdown in crops.

Significance and Impact of the Study

Similar assays can be applied to monitor frequencies of fungicide resistance in pathogens of crops and to assay diversity in microbial soil communities such as in soil samples from bat caves where white-nose syndrome has been detected.

Introduction

Blackleg disease, caused by the fungus Leptosphaeria maculans, results in major yield losses of canola (Brassica napus, oilseed rape) world-wide (Fitt et al. 2006). Wind-borne sexual spores (ascospores) released from stubble of crops grown in the previous year are the primary inoculum. Resistance to L. maculans involves both seedling and adult plant resistance. Seedling resistance is conferred by single major genes that are usually dominant, whilst adult plant resistance is conferred by multiple minor genes (Delourme et al. 2006). Very little is known about the number or types of genes conferring adult plant resistance in B. napus. Conversely, at least 16 seedling resistance genes (Rlm1, Rlm2, Rlm3, Rlm4, Rlm5, Rlm6, Rlm7, Rlm8, Rlm9, RlmS, LepR1, LepR2, LepR3, LepR4, BLMR1 and BLMR2) have been identified, and 15 of these are mapped to three different linkage groups (Yu et al. 2005, 2008; Delourme et al. 2006; Rimmer 2006; Long et al. 2011). No seedling resistance genes have been cloned.

Leptosphaeria maculans interacts with B. napus in a typical ‘gene-for-gene’ manner whereby for each seedling resistance gene in the plant, there is a corresponding avirulence gene in the fungus (Balesdent et al. 2005). The presence of an avirulence gene renders the pathogen unable to attack host plants with the corresponding resistance gene. However, when avirulence genes are mutated to virulence (virulence allele), plant hosts with the corresponding resistance gene no longer detect the pathogen, which leads to disease. Twelve avirulence genes have been identified: seven of these have been mapped and three, AvrLm1, AvrLm6 and AvrLm4-7, have been cloned in this haploid fungus (Balesdent et al. 2005; Yu et al. 2005; Gout et al. 2006; Fudal et al. 2007; Parlange et al. 2009). AvrLm1 and AvrLm6 confer avirulence towards resistance genes Rlm1 and Rlm6, respectively. Deletion of the entire avirulence locus is the major mechanism that confers virulence towards Rlm1 and Rlm6 in Australian and French L. maculans populations (Gout et al. 2007; Fudal et al. 2009; Van de Wouw et al. 2010a). In a small number of cases, repeat-induced point (RIP) mutations confer virulence towards AvrLm6 (Fudal et al. 2009; Van de Wouw et al. 2010a). RIP mutation is a process specific to ascomycetous fungi that alters the sequence of multi-copy DNA by triggering CpA to TpA and TpG to TpA nucleotide changes during meiosis. This often generates stop codons and thereby inactivates genes (Selker and Garrett 1988). AvrLm4-7 differs from AvrLm1 and AvrLm6 in that it confers avirulence towards two different resistance genes, Rlm4 and Rlm7 (Parlange et al. 2009). Furthermore, a single amino acid substitution rather than deletion is responsible for virulence towards Rlm4. The analysis of 11 isolates with the avirulence allele, AvrLm4, and 33 isolates with the virulence allele, avrLm4, from French L. maculans populations showed that a single nucleotide polymorphism (SNP) – a G to C mutation at base 358 of the AvrLm4-7 coding sequence – led to a G120 to R120 amino acid substitution that conferred virulence towards Rlm4. However, the mechanism conferring virulence towards Rlm7 remains unknown (Parlange et al. 2009). Hereafter, this fungal locus is referred to as AvrLm4.

When crop cultivars with a particular resistance gene are grown extensively, fungal populations can undergo strong selection pressure, which results in an increased frequency of isolates virulent towards the corresponding resistance gene. This scenario occurred in Australia in 2003 with the breakdown of ‘sylvestris’ resistance conferred by resistance gene, Rlm1, in B. napus resulting in 90% yield losses, and up to $10 million AUD in losses (Sprague et al. 2006; Van de Wouw et al. 2010b). When the selection pressure is removed, the frequency of virulent isolates sometimes decreases. Thus, monitoring the frequency of avirulent and virulent isolates in fungal populations plays an important role in predicting the risk of such resistance breakdown.

An assay to monitor the frequency of AvrLm1 and AvrLm6 genotypes in L. maculans populations has been developed (Van de Wouw et al. 2010b). Ascospores released from infected stubble are trapped on tape, genomic DNA is extracted, and AvrLm1 or AvrLm6, as well as a standard gene, is amplified by quantitative PCR. The frequency of the avirulence allele can be estimated, because virulence is conferred by deletion, and thus only the avirulence allele is amplified. Because of differences in the combinations of resistance genes in different cultivars, the frequency of the virulence alleles of different avirulence genes will vary depending on the source of the stubble from which the ascospores were discharged. For instance, stubble of a cultivar that has resistance gene Rlm1, but lacks Rlm4, is likely to harbour a low frequency of the avirulence allele of AvrLm1, and higher frequency of AvrLm4. This is because only isolates with the virulence allele (avrLm1) can colonize this cultivar and cause disease and then sexually cross on the stubble.

For genes such as AvrLm4 where a SNP is responsible for virulence, an assay relying on the virulence allele being a deletion cannot be used. However, pyrosequencing may be able to be adapted for this purpose. Pyrosequencing entails sequencing of short stretches of nucleotides surrounding a known polymorphism in a pooled sample containing multiple alleles. The proportion of each allele at the SNP site is quantified based on relative signal intensities (Alderborn et al. 2000). This method is used in medical diagnostics such as quantifying allele frequencies of SNPs in pooled samples of human blood (Gruber et al. 2002; Wasson et al. 2002). This approach reduces costs of genotyping blood samples involved in association studies for complex diseases because genotyping of individual samples is expensive. Pyrosequencing has also been used to differentiate closely related strains of the food-borne bacterial pathogen, Shigella (Hayford et al. 2011). Individual strains can be rapidly identified using pyrosequencing of 24 SNPs within nine different genes, aiding in clinical and epidemiological decisions. In addition, pyrosequencing has been used to assess changes in the frequency of different haplotypes in populations of the blood stage of Plasmodium falciparum, the organism that causes malaria (Takala et al. 2006). Populations of P. falciparum are monitored in humans prior to and following exposure to malarial vaccines to determine whether any haplotypes change in frequency, which may affect immunity to the vaccine.

In this study, we describe a pyrosequencing assay that can be used to estimate the frequency of virulence alleles of the AvrLm4 locus of L. maculans. In Australia, cultivars harbouring the corresponding Rlm4 resistance gene are grown extensively (Marcroft et al. 2012b). Therefore, monitoring the frequency of isolates that are virulent towards Rlm4 is essential for predicting risk of breakdown of Rlm4 resistance in these cultivars. The combination of pyrosequencing with trapping of airborne sexual spores of L. maculans released from stubble of different B. napus cultivars provides a high-throughput method for estimating the frequency of isolates virulent towards Rlm4 resistance.

Materials and methods

Leptosphaeria maculans isolates

Eighty-nine L. maculans isolates were cultured from individual ascospores discharged from infected B. napus stubble collected from canola-growing regions of Australia as described previously (Sprague et al. 2006). Six additional isolates were part of the International Blackleg Collection Network (IBCN; Purwantara et al. 2000). All isolates were grown and maintained on 10% Campbells V8 juice agar.

Brassica napus cultivars

Nine B. napus cultivars with different combinations (complements) of resistance genes were used (Table 1). The nomenclature Rlm4 is used for the presence of resistance gene Rlm4 and rlm4 for its absence; the former allele confers resistance to avirulence gene, AvrLm4, whilst the latter allele does not. Some of the cultivars contained resistance genes other than Rlm4 and Rlm1, for example Rlm9 in cultivar ATR-Cobbler and RlmS in cultivar Monola 76TT. The presence of these additional resistance genes would not impact on the frequency of virulence alleles of AvrLm1 and of AvrLm4 as they correspond to different avirulence genes.

Table 1. Brassica napus cultivars used in this study
CultivarResistance genes in cultivara
  1. a

    Seedling resistance genes identified by Marcroft et al. (2012a).

WestarNo resistance genes
Thunder TT Rlm4
CB-Telfer Rlm4
ATR-CobblerRlm4, Rlm9
ATR-BeaconRlm3, Rlm4
Monola 76TTRlm1, RlmS
CB Jardee HT Rlm2
Hyola50Unknown but lacks Rlm4 and Rlm1

Leptosphaeria maculans spore trap samples

Stubble of B. napus cultivars (termed stubble sample hereafter) was collected from different locations across Australia. Fifteen to twenty pieces of stubble (5–7 cm in length) containing mature pseudothecia (sexual fruiting bodies) of L. maculans were immersed in water for 3–5 min and then placed inside a single chamber of a Burkard spore liberator (Hirst and Stedman 1962). Air drawn from outside the chamber carried ascospores released from the stubble downwind during a period of 30–60 min, through an aperture onto wax-coated Melinex plastic tape (Burkard Manufacturing Company Ltd, Hertfordshire, UK) fixed to a microscope slide. For each stubble sample, three replicate slides with ascospores trapped on this tape were collected and stored at −20°C until further use.

Extraction of fungal DNA

Mycelia were harvested from individual L. maculans isolates grown in 10% V8 juice, freeze-dried and genomic DNA was extracted as described by Sexton and Howlett (2000). DNA was extracted from ascospores deposited on pieces of wax-coated tape (termed spore trap samples hereafter) using the CTAB extraction protocol (Rogers et al. 2009; Van de Wouw et al. 2010b).

Identification and characterization of AvrLm4 alleles from individual L. maculans isolates

Alleles of AvrLm4 in individual isolates were identified by PCR amplification and Sanger sequencing, or by the disease phenotype of B. napus cultivars with the complementary resistance gene, Rlm4, after inoculation with the isolates. A 1127-bp region encompassing the entire AvrLm4 open reading frame was amplified and Sanger-sequenced from genomic DNA of 95 individual isolates. Primers (listed in Table 2) were designed using the program Primer3. The 1127-bp region was amplified as follows: 95°C for 3 min; 35 cycles of 95°C for 30 s, 59°C for 30 s, 72°C for 40 s and 72°C for 10 min. The amplified fragment was Sanger-sequenced at the Australian Genome Research Facility, Melbourne, Australia. All sequences were aligned to the AvrLm4 sequence of L. maculans isolate v23·1·3 (Genbank accession number AM998638).

Table 2. Primers used in this study
Primer nameSequence (5′ to 3′)Amplicon length (bp)Use
  1. a

    Primer is biotin labelled.

  2. b

    Primers published previously (Van de Wouw et al. 2010b).

AvrLm4FAGAAGGGTAAGGGGCAAGTC1127Sanger sequencing of AvrLm4 alleles
AvrLm4RGAAGAACCCTGCTAGATAGGTAAGC
Pyro F aTCCCTATAGCAGCTTTAGCTCAG90Amplification of AvrLm4 for pyrosequencing
Pyro RCTTACGTTAACAACCCATTCAGG
Pyro SeqAACCAGTCTCCTGGCN/APyrosequencing sequencing primer
AvrLm1 qF bGGGTGTTTACTTCGCCTCAC198Quantitative PCR analysis of AvrLm1 allele frequency in spore trap samples
AvrLm1 qR bACGTTGTAATGAGCGGAACC
LmacF bCTTGCCCACCAATTGGATCCCCTA330Quantitative PCR analysis of ITS region in spore trap samples
LmacR bGCAAAATGTGCTGCGCTCCAGG

A subset of 20 isolates was inoculated onto seedlings of Rlm4-containing cultivars (CB-Telfer and Thunder TT; Table 3; Marcroft et al. 2012a). All isolates were also inoculated onto a cultivar with no resistance genes (Westar) as a susceptible control. Vegetative spores (104 per droplet) of each isolate were inoculated onto wounded cotyledons (Purwantara et al. 1998). Symptoms were assessed at 17 days post-inoculation (dpi) on a scale from 0 (no darkening around wounds) to 9 (large grey–green lesions with vegetative spores) (Koch et al. 1991). Mean pathogenicity scores (determined from 32 inoculation sites) of <3·0 reflected that lesions were restricted (<3 mm in diameter) indicating that the fungus could not cause disease (avirulent phenotype). Scores >5·0 were assigned when lesions were >5 mm in diameter (virulent phenotype); these lesions often contained asexual fruiting bodies. Isolates that gave scores between 3·0 and 5·0 (intermediate phenotype) were re-tested; no isolates consistently caused lesion scores between 3·0 and 5·0.

Table 3. Characterization of the AvrLm4 gene in Australian Leptosphaeria maculans isolates
AvrLm4 alleleaAmino acid position within the AvrLm4 proteinbNumber of isolates with allele (%)Representative isolates screened for virulence towards Rlm4Disease phenotypec
68086120121
  1. a

    The AvrLm4 gene was Sanger-sequenced from 95 Australian L. maculans isolates and alleles were assigned.

  2. b

    The amino acid sequence was determined for each of the alleles. E6 to K6 amino acid substitution is conferred by a G to A mutation at base 16 of the coding sequence. I80 to T80 is conferred by a T to C mutation at base 239. D86 to N86 is conferred by a G to A mutation at base 256. G120 to R120 is conferred by a G to C mutation at base 358. Q121 to K121 is conferred by a C to A mutation at base 361. Alleles AvrLm4-5, AvrLm4-6, AvrLm4-7 and AvrLm4-8 appear to have been generated via repeat-induced point (RIP) mutations and had 40, 42, 39 and 40 nucleotide changes, respectively. These nucleotide changes generated nine stop codons within each allele.

  3. c

    Twenty of the 95 isolates were characterized as avirulent or virulent by assessment of disease symptoms 17 days post-inoculation of cultivars of Brassica napus containing the resistance gene Rlm4.

AvrLm4-0 EIDGQ20 (21)IBCN17, IBCN18, 04S005, 06S013, D13Avirulent
AvrLm4-1 ETDRQ11 (12)05S013, 10W002Virulent
AvrLm4-2 EINRQ55 (58)04P042, 05P032, 05P033, 06P039, D9, 08P011Virulent
AvrLm4-3 KINRQ3 (3)04P013Virulent
AvrLm4-4 EINRK2 (2)11P026, 11P028Virulent
AvrLm4-5 RIP-like mutations1 (1)IBCN15Virulent
AvrLm4-6 RIP-like mutations1 (1)IBCN16Virulent
AvrLm4-7 RIP-like mutations1 (1)IBCN75Virulent
AvrLm4-8 RIP-like mutations1 (1)IBCN76Virulent

Pyrosequencing of AvrLm4 from L. maculans spore trap samples

Primers were designed to PCR-amplify a 90-bp fragment that included the codon G120 of AvrLm4 (Fig. 1; Table 2). The forward primer (Pyro F) was biotin labelled and HPLC purified (Integrated DNA Technologies, Coraville, IA). A sequencing primer was also designed to anneal directly 3′ of base 358 that is mutated to confer virulence (G to C) in the codon G120 of AvrLm4. The reverse primer (Pyro R) and sequencing primer (Pyro seq) were standard sequencing grade. These primers were designed using pyromark assay design software V2 (Qiagen, Victoria, Australia). The 90-bp fragment was amplified as follows: 95°C for 3 min, 35 cycles of 95°C for 30 s, 59°C for 30 s, 72°C for 40 s and 72°C for 10 min. The biotinylated PCR product was bound to Streptavidin Sepharose High Performance beads (GE Healthcare Life Sciences, New South Wales, Australia) and then denatured and washed on the Pyrosequencing Vacuum Prep Tool (Qiagen) as per the manufacturer's recommendations to isolate a single-stranded template. The template-bound beads and pyrosequencing primer were transferred to a sequencing plate then heated to 80°C followed by cooling to room temperature to allow annealing of the sequencing primer. Pyrosequencing was performed at the Australian Genome Research Facility, Perth, Australia on a PyroMark 24 Pyrosequencing System (Qiagen). Data were analysed on the pyromark Q24 software to give the relative quantities (percentage G and C) of each allele in the sample.

Figure 1.

Position of the primers for pyrosequencing analysis of the mutation within the AvrLm4-7 gene of Leptosphaeria maculans. The amino acid substitution glycine (G)120 to arginine (R)120 (box) confers virulence towards Rlm4 and is encoded by a G to C mutation at base 358 of the AvrLm4-7 coding sequence (Genbank accession number AM998638). The forward biotin-labelled and reverse PCR primers amplify the region of interest (90 bp) whilst the reverse sequencing primer is used for pyrosequencing analysis.

Preparation of AvrLm4 DNA mixtures with known proportions of alleles

Using the PyroF and PyroR primers, a fragment of AvrLm4 was amplified from an isolate with the avirulence allele, AvrLm4 (isolate IBCN18) and an isolate with the virulence allele avrLm4 (isolate D9). PCR fragments representing both the avirulence and virulence alleles were mixed together in different proportions and used to assess whether pyrosequencing could discriminate different frequencies of these alleles. These DNA mixtures contained frequencies of the avirulence allele, AvrLm4 at 100, 80, 60, 40, 20 or 0%. The 100% AvrLm4 sample consisted of 250 ng μl−1 of PCR product amplified from isolate IBCN18, whilst the 0% AvrLm4 sample consisted of 250 ng μl−1 of PCR product amplified from isolate D9. The mixture containing a frequency of 80% AvrLm4 consisted of 200 ng μl−1 of PCR product amplified from isolate IBCN18 and 50 ng μl−1 of PCR product amplified from isolate D9. All DNA mixtures were made to a final concentration of 250 ng μl−1 of PCR product.

Determination of allele frequencies of AvrLm1 from spore trap samples

The frequency of the AvrLm1 allele in ascospore populations was determined as described by Van de Wouw et al. (2010b). Briefly, the internal transcribed spacer (ITS) region of ribosomal DNA of L. maculans and the AvrLm1 gene was amplified using quantitative PCR. The amount of AvrLm1 amplicon was compared to that of the ITS region to provide an estimate of the frequency of isolates harbouring the AvrLm1 gene, that is the frequency of avirulence (AvrLm1) alleles.

Statistical analysis

Significance of avirulence allele frequency data was calculated using analysis of variance (anova) tests. A 95% confidence interval was used to indicate significance.

Results

Identifying polymorphisms at the AvrLm4 locus

Sanger sequencing of the entire AvrLm4 gene in 95 Australian L. maculans isolates revealed the presence of nine alleles (Table 3), of which three (AvrLm4-0, AvrLm4-1, AvrLm4-2) were identical to those previously characterized in 44 French isolates (Parlange et al. 2009). Twenty isolates, which included all nine of the AvrLm4 alleles, were screened for virulence towards Rlm4 by inoculation of Rlm4-cultivars, CB-Telfer and ThunderTT (Table 3). Only isolates with the AvrLm4-0 allele were avirulent towards Rlm4, but all isolates were virulent towards cultivar Westar. Of the eight alleles conferring virulence, four harboured the G120 to R120 amino acid change. These four alleles accounted for 96% of all virulent isolates. The remaining four virulence alleles were generated through RIP-like mutations (Table 3) and were identified in four of the IBCN isolates that had been collected prior to 1990. Each of these putative RIP-alleles had 39 or more mutations leading to multiple amino acid substitutions and stop codons. All these alleles had nine stop codons at identical positions.

Developing and validating the pyrosequencing assay

A 90-bp fragment that included the G120 codon of AvrLm4 was amplified from isolates IBCN18 (AvrLm4 allele) and D9 (avrLm4 allele), and mixtures representing known proportions of the AvrLm4 allele were generated. These samples were pyrosequenced to estimate the percentage of G and C nucleotide incorporation at base 358. For each sample, the estimated frequency of the virulence allele based on pyrosequencing was plotted against the known frequency of the virulence allele in the mixture. There was a linear relationship between estimated and known allele frequencies with an R2 value of 0·999 (Fig. 2). Samples comprising entirely of the AvrLm4 fragment or of the avrLm4 fragment were tested in triplicate, and standard deviations of 0·21 and 0·32, respectively, were found, indicating no significant differences between pyrosequencing runs.

Figure 2.

Relationship between known frequencies of the virulence allele (avrLm4) of Leptosphaeria maculans AvrLm4 and allele frequencies estimated by pyrosequencing. PCR fragments representing both the avirulence and virulence alleles of AvrLm4 were mixed in different proportions to give a known frequency of avrLm4, the virulence allele. The avrLm4 allele frequency was estimated by pyrosequencing these mixtures with different proportions of the avrLm4 allele. Error bars represent the standard deviation of the mean (= 3).

Single ascospores (isolates) were cultured from stubble of cultivars with genotypes Rlm4 rlm1 (six from ThunderTT and five from ATR-Cobbler), and the entire AvrLm4 gene was Sanger-sequenced. All eleven isolates were virulent at the AvrLm4 locus; that is, they had the avrLm4 allele. Genomic DNA was extracted from spore trap samples collected from the same stubble used to culture the single isolates and analysed using pyrosequencing. The frequency of the virulence allele (avrLm4) in this spore trap sample was 97%.

Spore trap samples were collected from stubble of cultivars with different combinations of resistance genes. Pyrosequencing was used to determine the frequencies of the avrLm4 allele, and frequencies of the avrLm1 allele were determined as described by Van de Wouw et al. (2010b). The frequency of the avrLm4 allele was greater than 95% in spore trap samples derived from both Rlm4 rlm1 cultivars, ATR-Cobbler and ATR-Beacon (Fig. 3). The frequency of the avrLm1 allele was significantly lower (67 and 43%) than the frequency of the avrLm4 allele in these spore trap samples. In spore trap samples derived from the rlm4 Rlm1 cultivar, Monola 76TT, the frequency of the avrLm1 allele was 99% whilst the frequency of the avrLm4 allele was 54%. The frequencies of both the avrLm4 and avrLm1 alleles were <80% in spore trap samples derived from cultivars CB Jardee HT and Hyola50, which lack both Rlm4 and Rlm1. The frequency of avrLm4 was significantly lower than the frequency of avrLm1 in the spore trap sample derived from Hyola50; however, there were no significant differences in allele frequencies of the avrLm4 and avrLm1 alleles in spore trap samples derived from cultivar CB Jardee HT. The frequency of the avrLm4 allele was highest when Rlm4 was present in the cultivar from which the spore trap sample was derived, that is cultivars ATR-Cobbler and ATR-Beacon, whilst the frequency of avrLm1 was highest when Rlm1 was present (cultivar Monola 76 TT) .

Figure 3.

Frequency of the virulence alleles, avrLm4 and avrLm1, of Leptosphaeria maculans in ascospore populations derived from stubble of Brassica napus. Genomic DNA was extracted from ascospores trapped on tape (spore trap samples) collected from stubble of cultivars with different resistance gene complements and analysed by pyrosequencing (for avrLm4) or quantitative PCR (for avrLm1). The significance of differences in frequencies of the virulence alleles of the two genes in spore trap samples collected from the same stubble was determined by analysis of variance. Virulence frequencies from the same spore trap sample that have the same letter are not significantly different at the 95% confidence interval. Error bars represent the standard error of mean (= 3).

Discussion

Trapping spores and analysing them by PCR provides a rapid high-throughput method for estimating frequencies of alleles in fungal populations. Previously, estimating allele frequencies of avirulence genes of L. maculans relied on culturing single sexual spores from single pieces of diseased stubble and either screening them on cultivars with the corresponding resistance genes or screening them with molecular markers (Dilmaghani et al. 2009; Van de Wouw et al. 2010a). This process is time consuming and usually only small numbers of isolates can be analysed. The use of pyrosequencing for estimating AvrLm4 allele frequencies relies on the fact that virulence is conferred by a SNP. If alternative mechanisms of virulence arise in the fungal population, then the assay would be unsuitable. Therefore, each year a subset of individual isolates should be assessed to confirm that the SNP at base 358 is still the major mechanism for virulence. Alternative methods for detecting SNPs include methods such as amplification refractory mutations system (ARMS) PCR (Newton et al. 1989; Ye et al. 2001). This method involves design of primers specific to each allele of interest and analysis of the amplification products using systems such as microplate array diagonal gel electrophoresis (MADGE). However, pyrosequencing has multiple benefits over methods such as ARMS-PCR; pyrosequencing provides accurate quantification of allele frequencies, detects any base change at the SNP site of interest and does not require primers specific to each allele.

The AvrLm4 pyrosequencing assay was validated by comparing avrLm4 allele frequencies estimated using pyrosequencing to known frequencies of the avrLm4 allele and also by comparing pyrosequencing of spore trap samples to traditional Sanger sequencing of the entire AvrLm4 locus of individual isolates. The finding that the frequency of the virulence alleles was higher in populations of ascospores derived from stubble of cultivars that have the corresponding resistance gene indicates that selection pressure is acting on fungal populations by the resistance genes in the canola crops sown. This is consistent with our previous findings that the frequency of both the avrLm4 and avrLm1 alleles decreased to approximately 70% in individual isolates derived from stubble of cultivars lacking Rlm4 or Rlm1, respectively (Marcroft et al. 2012b). However, both virulence alleles avrLm4 and avrLm1 have a fitness cost compared with the avirulence allele at these loci (Huang et al. 2006, 2010). This suggests that if B. napus cultivars lacking Rlm4 or Rlm1 were sown over a sustained period of time, the frequency of the virulence alleles in ascospore populations derived from these stubbles could reduce to very low levels. Whether this would actually happen is unknown, as no long-term studies have been carried out.

Pyrosequencing can be applied to other fungal pathogens where avirulence genes have been identified and the polymorphisms conferring virulence are known. For example, Cladosporium fulvum is a tomato pathogen where numerous avirulence genes have been characterized and the underlying SNPs identified (Stergiopoulos et al. 2007). Spore trap sampling and pyrosequencing could be applied to monitor the frequency of avirulence alleles in the field. Furthermore, pyrosequencing could be used to monitor the frequency of alleles associated with fungicide resistance in plant pathogens. A number of polymorphisms have been identified in the CYP51 gene that is involved in resistance to azole fungicides in Mycosphaerella graminicola, the causal agent of Septoria leaf blotch of wheat (Cools and Fraaije 2008). A pyrosequencing assay has been used to assess frequencies of CYP51 alleles in individual isolates of M. graminicola cultured from leaf lesions before and after exposure to fungicides (Stammler et al. 2008). This assay could be combined with trapping of sexual spores to monitor the frequency of alleles conferring fungicide resistance.

Pyrosequencing assays could also be exploited to assess microbial communities in soil sampling. For example, assays are being used to assess whether Geomyces destructans, the fungus associated with white-nose syndrome in bats, is present in soil samples in bat caves (Lindner et al. 2011). This assay involves a presence/absence of PCR test but is limited in that it detects many Geomyces spp. A pyrosequencing assay based on a taxon-specific sequence recently identified in G. destructans (Lindner et al. 2011) could be developed to detect this pathogen in bat caves. Pyrosequencing is a currently underutilized tool with potential applications across many disciplines including disease epidemics and medical diagnostics to provide high-throughput, accurate and low-cost assays.

Acknowledgements

We thank the Australian Grains Research and Development Corporation for funding and David Chandler at the Australian Genome Research Facility, Perth for useful discussions.

Ancillary