New capabilities for Mycosphaerella graminicola research

SUMMARY

Mycosphaerella graminicola is a major pathogen of wheat worldwide, causing Septoria leaf blotch disease. Targeted gene disruption in M. graminicola, by Agrobacterium tumefaciens‐mediated transformation, has become an established functional genomics tool for M. graminicola research in recent years. However, in order to advance research into this economically important pathogen, further functional genomics tools need to be developed. Here, we report three new capabilities for M. graminicola research: (i) two selectable markers have been shown to work robustly in M. graminicola, namely G418 and the fungicide carboxin; (ii) the generation of a strain of M. graminicola in which the KU70 (MUS‐51) homologue has been disrupted; in this strain, homologous recombination efficiencies increased to more than 95%, whilst maintaining wild‐type growth in vitro and full pathogenicity on wheat leaves; (iii) the ability to efficiently target and generate precise mutations of specific genes in the genomic context in M. graminicola. In addition, the insertion of the E198A mutation into the β‐tubulin gene (MgTUB1), conferring resistance to the fungicide benomyl, suggests that this mutant allele may provide an additional selectable marker. The collective use of these tools will permit further advancements in our knowledge of the biology and pathogenicity of this important plant pathogen.

INTRODUCTION

Mycosphaerella graminicola (Fuckel) J Schrot. In Cohn causes Septoria leaf blotch disease in wheat and is the most important disease requiring control to maximize yields in the European cereals market (Mehrabi et al., 2006a). Our knowledge of the biology and pathogenicity of this economically important pathogen has increased over recent years, with the development of genomics tools for filamentous plant pathogenic fungi and their specific application to M. graminicola. Mycosphaerella graminicola was first transformed by Payne et al. (1998) by polyethylene glycol (PEG)‐mediated transformation of protoplasts, although, with this method, these researchers were not successful in disrupting specific genes through homologous recombination. This was achieved by Zwiers and De Waard (2001), who reported the first successful targeted disruption of a specific gene in M. graminicola, the ABC transporter MgAtr2, using the Agrobacterium tumefaciens‐mediated transformation protocol for fungi (de Groot et al., 1998). Indeed, this process has enabled the elucidation of the role of some of the key genes from the mitogen‐activated protein kinase pathway, and the cAMP‐dependent pathway, in the pathogenicity and development of M. graminicola (Cousin et al., 2006; Mehrabi and Kema, 2006; 2006a, 2006b). The genome of M. graminicola was sequenced by the Joint Genome Institute (JGI) in 2005 and this is now completely assembled and aligned to the 21 chromosomes possessed by the sequenced strain IPO323 (http://genome.jgi‐psf.org/Mycgr3/Mycgr3.home.html). This resource is critical in advancing functional studies of genes of interest in M. graminicola.

Mycosphaerella graminicola has a very different lifecycle and mode of pathogenicity from other economically important plant pathogenic fungi, such as Magnaporthe oryzae, in which functional genomic studies are well advanced (Betts et al., 2007; Dean et al., 2005; Kema et al., 1996; Keon et al., 2007; Talbot, 2003). It is therefore of utmost importance that specific hypotheses are tested directly in M. graminicola, rather than assumptions being made about the roles of genes from Magnaporthe oryzae, or other ‘model’ filamentous plant pathogens, such as Cladosporium fulvum which, although an exclusively apoplastic colonizer of leaves, maintains a biotrophic growth habit throughout infection (Thomma et al., 2005). The study of this unique biology requires the continuing advancement of the genomics tools already established in M. graminicola.

One of the limiting factors in the ability to efficiently disrupt specific genes in M. graminicola is the frequency at which homologous recombination occurs at specific loci. This is highly variable, with frequencies of 0.5%–75% being obtained so far (Mehrabi, 2006; Zwiers and De Waard, 2001). The differences in the frequency of homologous recombination observed between genes is likely to be caused by target locus‐specific effects, such as chromosome position, loci chromatin structure and the transcriptional status of the gene, all of which may affect how available the DNA is for homologous recombination (Bird and Bradshaw, 1997; Michielse et al., 2005). This variability in targeted gene disruption frequencies has also been found for other filamentous fungi. Homologous recombination occurred at frequencies of 10%–30% in Neurospora crassa (Ninomiya et al., 2004) and at 0%–40% in Aspergillus nidulans (Nayak et al., 2006). In eukaryotes, there are two main pathways which repair double‐stranded breaks in DNA. The first, homologous recombination, involves the interaction between homologous sequences, whereas the second, nonhomologous end‐joining (NHEJ), involves the direct ligation of strands of DNA irrespective of DNA homology. In Saccharomyces cerevisiae, homologous recombination is the main mechanism for the repair of DNA breaks, whereas, in many other organisms, including mammals, plants, insects and filamentous fungi, NHEJ is the main repair pathway (Ninomiya et al., 2004; Villalba et al., 2008). During NHEJ, the Ku protein complex (consisting of the Ku70–Ku80 heterodimer) binds to the ends of the DNA. This complex recruits a DNA‐dependent protein kinase which phosphorylates the DNA exonuclease Artemis. The activated complex then stimulates the binding of the DNA ligase IV‐Xrcc4 to the DNA ends and the double‐strand break is repaired (Villalba et al., 2008). It was first reported in N. crassa that, when the MUS‐51 (KU70) or MUS‐52 (KU80) gene was deleted, the frequencies of homologous recombination in this strain increased to 100%, because, without the functional NHEJ pathway, homologous recombination became the major mechanism for the repair of double‐stranded DNA breaks (Ninomiya et al., 2004). Since then, it has been shown that disruption of the KU70 or KU80 gene increases homologous recombination efficiencies for many other fungi, including Magnaporthe oryzae (Villalba et al., 2008), A. nidulans (Nayak et al., 2006), A. fumigatus (Krappmann et al., 2006) and Sordaria macrospora (Poggeler and Kuck, 2006). We hypothesized that disruption of the KU70 homologue in M. graminicola would also increase the frequencies of homologous recombination in this pathogen.

The only selectable marker currently used for M. graminicola transformation is hygromycin (Payne et al., 1998; Skinner et al., 1998; Zwiers and De Waard, 2001). In order to generate a ΔMgKU70 strain, which could be used as a recipient in further transformations, an additional selectable marker was required. We therefore tested two further selectable markers which could be used for M. graminicola transformation, the widely used neomycin phosphotransferase II (nptII) gene from Escherichia coli, previously shown to confer resistance to G418 in the fungus Ashbya gossypii (Wendland et al., 2000), and the carboxin‐resistant allele of the succinate dehydrogenase subunit B MgSDHB(H267Y) gene from M. graminicola. Carboxin is a fungicide which acts by preventing the oxidation of succinate to fumarate in complex II of the respiratory electron transport chain. It binds to the iron–sulphur protein, encoded by the SDHB gene, of the succinate dehydrogenase enzyme (Skinner et al., 1998). Skinner et al. (1998) showed that the amino acid change of histidine to tyrosine at position 267 conferred resistance to carboxin, and thus had the potential to be used as a selectable marker for M. graminicola transformation.

The frequencies of homologous recombination in the ΔMgKU70 M. graminicola strain were compared with those achieved in the wild‐type strain, using a construct that replaced the wild‐type MgTUB1 gene, encoding β‐tubulin, with a resistant allele E198A. It has been reported in field isolates of several fungal species, including M. fijiensis (Canas‐Gutierrez et al., 2006), Tapesia yallundae (Oculomacula yallundae), Tapesia acumformis (Oculomacula acumformis) (Albertini et al., 1999) and M. graminicola (Fraaije et al., 2007), that the change of glutamic acid to alanine at position 198 in β‐tubulin confers resistance to the benzimidazole fungicides. We successfully generated M. graminicola transformants containing the A198 resistant allele and confirmed that these were resistant to the fungicide benomyl. We believe that this is the first report of the generation of a precisely engineered manipulation to a specific gene in M. graminicola.

RESULTS

Validation of two new selectable markers for M. graminicola

Sensitivity assays were performed to confirm that M. graminicola was sensitive to the selection agents and to determine appropriate concentrations for use. Wild‐type M. graminicola plated at concentrations of 5 × 10⁷ and 1 × 10⁷ cells/mL were controlled by 250 and 100 µg/mL G418, respectively. For carboxin, wild‐type M. graminicola plated at concentrations of 5 × 10⁷ and 1 × 10⁷ cells/mL were controlled by 40 and 20 µg/mL, respectively.

To assess the use of nptII and MgSDHB(H267Y) as selectable markers for transformation, constructs were generated in a pNOV2114‐based binary vector. Using these binary vectors, we were able to obtain transformation in the absence of acetosyringone or other phenolic virulence‐inducing compounds. Transformations with the nptII gene were selected on 250 µg/mL G418. After 14 days, the transformants were visible, with 164 putative transformants per 1 × 10⁷ cells/mL. The putative transformants were transferred individually onto G418‐containing plates for two further rounds of selection, and then verified to be positive by polymerase chain reaction (PCR). Transformations with MgSDHB(H267Y) were selected on 20, 40 and 80 µg/mL carboxin. With 20 µg/mL, there were 83 putative transformants per 1 × 10⁷ cells/mL, but the background growth was high. At 40 µg/mL, there were 13 putative transformants per 1 × 10⁷ cells/mL, with no background growth. At 80 µg/mL, there were no transformants. Putative transformants were subcultured on 40 µg/mL carboxin for two further rounds of selection, and verified to be positive by PCR. All transformed strains were highly resistant, growing on 400 µg/mL G418 or 80 µg/mL carboxin (Fig. 1A, B) after 5 days. Better suppression of background M. graminicola growth was found using the nptII gene with G418, and so this selectable marker was chosen to generate the MgKU70 knockout strain.

Figure 1.

Wild‐type and transformed Mycosphaerella graminicola following 5 days of selection with G418 and carboxin. (A) Wild‐type (wt) and M. graminicola transformed with the neomycin phosphotransferase II (nptII) gene (G418^R), conferring resistance to G418, growing on increasing concentrations of G418. (B) Wild‐type (wt) and M. graminicola transformed with the MgSDHB(H267Y) gene (carboxin^R), conferring resistance to carboxin, growing on increasing concentrations of carboxin.

Identification of the KU70 homologue in M. graminicola

Using a blastp search, a single gene, MgKU70, was identified in the M. graminicola genome as being homologous to the N. crassa MUS‐51 (KU70) gene (Ninomiya et al., 2004) with an E‐value of 1 × 10⁻¹⁵⁴. The MgKU70 open reading frame contains two introns, is predicted to code for a 644‐amino‐acid protein and shares 61% identity with N. crassa MUS‐51 (KU70). The protein also shares the Ku‐type ATP‐dependent DNA helicase motif within the α/β Ku N‐terminal domain and the C‐terminal Ku domain present in N. crassa, which are characteristic of these proteins.

Deletion of the M. graminicola KU70 gene

The gene replacement cassette (pNOVMgKU70) was introduced into M. graminicola by Agrobacterium‐mediated transformation of single yeast‐like cells as described. In total, 197 putative transformants were generated, 149 of which were subcultured onto G418‐selective medium (250 µg/mL) for two further rounds of selection. PCR verification was performed on 93 of these, with four of 93 (4.3%) showing the correct products for the ΔMgKU70 gene replacement event (Fig. 2A–D). Southern blot analysis was performed on two of the ΔMgKU70 strains and, in both cases, there was only a single hybridization product of 2.496 kb, the size expected when the MgKU70 gene has been correctly targeted (Fig. 2B, C, E), thus indicating that the MgKU70 gene has been disrupted and there are no ectopic insertion events.

Figure 2.

Disruption of the MgKU70 locus. (A) Diagram showing the construct used to disrupt the MgKU70 gene. (B) Diagram showing the native MgKU70 locus. (C) Diagram showing the disrupted MgKU70 locus after transformation in the correctly targeted strains. (D) Agarose gel showing the results of the verification polymerase chain reaction (PCR) carried out on 13 of the putative MgKU70 disruptant transformants (sample 14 is wild‐type untransformed DNA control, and sample 15 is a no‐template control for PCR). A product from the primer pair Ku70verF1/Ku70verR1 indicates the presence of the disruption cassette within a strain. A product from the primer pair Ku70verF2/Ku70verR2 indicates the presence of the wild‐type MgKU70 locus. As expected, the untransformed control only contains a product from Ku70verF2/Ku70verR2. Ectopic transformants result in products from both sets of primer pairs: samples 2–7 and 11–13. In MgKU70 gene disruptants, a product is obtained only with primers Ku70verF1/Ku70verR1: samples 1, 8, 9 and 10. (E) Southern blot analysis of three of the MgKU70 gene disruptant strains K4901 (T1), K4902 (T2) and K4903 (T3) and the wild‐type (wt) control K4418. The genomic DNA was digested with HindIII, and probed with a fragment of the selectable marker gene neomycin phosphotransferase II (nptII) (nptII region amplified as probe as indicated by a full black line in C). K4902 and K4903 contain single‐copy insertion events as a single band of 2.496 kb has hybridized with the probe. K4901 is an uncut control.

Phenotypic characterization of the M. graminicolaΔMgKU70 strain

The growth of the wild‐type and the two ΔMgKU70 strains, which had been confirmed to contain a single copy of the MgKU70 deletion cassette, was compared in Vogel's semi‐solid minimal medium over a 7‐day time course (Fig. 3A). Analysis of variance was performed on the three strains at each time point, and the resultant F‐test probability indicated that the differences observed between the wild‐type and ΔMgKU70 strains were not statistically significant (Table 1).

Figure 3.

In vitro and in planta growth comparisons of the wild‐type and MgΔKU70 strains. (A) Growth of the wild‐type Mycosphaerella graminicola strain K4418 (wt) and two ΔMgKU70 strains K4902 and K4903 in Vogel's semi‐solid minimal medium. The values are averages for three replicates for each strain; bars represent the associated standard error of the mean. OD, optical density. (B) The percentage of disease coverage on wheat leaves infected with the wild‐type M. graminicola strain K4418 (wt) and the two ΔMgKU70 M. graminicola strains K4902 and K4903, 17 days post‐infection. The values are averages for 10 replica pots infected with each strain; bars represent the associated standard errors of the mean.

Table 1.

Analysis of variance performed on the in vitro growth of the wild‐type strain (K4418) and the MgΔKU70 strains (K4902 and K4903) at each time point.

Strain	Day 0	Day 1	Day 2	Day 3	Day 4	Day 7
K4418 (wt)	0.0426	0.0868	0.3261	0.4667	0.4513	0.4156
K4902	0.0615	0.0826	0.2692	0.3535	0.3771	0.3474
K4903	0.0623	0.0943	0.3530	0.4967	0.4930	0.4894
F‐test prob.	30.3%	77.9%	25.1%	16.5%	17.1%	9.6%
Pooled SD	0.0151	0.0198	0.0525	0.0764	0.0603	0.0583

The growth of the wild‐type and ΔMgKU70 strains on wheat was compared in an in planta infection assay. Seventeen days after inoculation, the level of disease caused by the three strains on the wheat leaves was visually assessed. The ΔMgKU70 strains displayed similar symptoms to the wild‐type strain, with all three strains developing traces of pycnidia (data not shown). The average percentage of leaf surface covered with disease was calculated per pot, with the associated standard error of the mean, and these were plotted graphically (Fig. 3B). Analysis of variance was performed on the three strains, and the resultant F‐test probability indicated that the differences observed between the wild‐type and the ΔMgKU70 strains were not statistically significant (Table 2). These results show that the ΔMgKU70 strains are fully pathogenic, behaving like the wild‐type parental isolate.

Table 2.

Analysis of variance performed on the in planta disease assessment of the wild‐type strain (K4418) and the MgΔKU70 strains (K4902 and K4903).

Strain	Mean disease strength
K4418 (wt)	56.9
K4902	52.1
K4903	57.5
F‐test prob.	48.9%
Pooled SD	10.9

Gene targeting in the ΔMgKU70 strain

To test the efficiency of targeted gene replacement in the ΔMgKU70 strain compared with the wild‐type, a construct was designed to replace the native MgTUB1 gene with a mutant allele (pNOV_βtubulinA198_replace) (see Appendix S1 in Supporting Information for construct design details). The construct was transformed into both M. graminicola wild‐type and one of the single‐copy ΔMgKU70 strains as described. Following cocultivation, the transformations were plated onto Aspergillus minimal medium (AMM) containing 100 µg/mL hygromycin to select for transformants. Putative transformants were generated, transferred onto fresh plates twice for two further rounds of hygromycin selection, and verified by PCR for correctly targeted gene replacement (Fig. 4A–D). The results showing the numbers of transformants generated in which targeted gene replacement had occurred in both the wild‐type and ΔMgKU70 strain are given in Table 3. The frequency of homologous recombination resulting in correctly targeted gene replacement was better than 97% in the ΔMgKU70 strain, compared with approximately 70% for the wild‐type strain.

Figure 4.

Targeted gene replacement at the MgTUB1 locus. (A) Diagram showing the construct used to replace the wild‐type MgTUB1 with a resistant allele. (B) Diagram showing the native MgTUB1 locus. (C) Diagram showing the replaced MgTUB1 locus after transformation in the correctly targeted strains. (D) Agarose gel showing the results of the polymerase chain reaction (PCR) verification of the MgTUB1 gene replacement transformants. Primers BtubHRverF2/hygromycin1 produce a 1577‐bp product for all successfully transformed strains. The primers BtubHRverF1/BtubHRverR1 amplify a product of 1925 bp when the construct successfully integrates at the MgTUB1 locus. The same primer pair amplifies a 501‐bp product when the wild‐type locus is retained, indicating ectopic transformants. Samples 1–8 were from wild‐type transformed M. graminicola and the transformants 1, 3 and 7 show correctly targeted gene replacement. Samples 9–16 were transformed into the ΔMgKU70 strain and all show correctly targeted gene replacement.

Table 3.

The frequencies of correctly targeted gene replacement strains at the MgTUB1 and MgERG11 loci, generated in the wild‐type and ΔMgKU70 parental strains.

Gene	Parent strain	Ectopic transformants	Gene replacement transformants
MgTUB1	Wild‐type	13/46	33/46
		28.3%	71.7%
MgTUB1	ΔMgKU70	1/46	45/46
		2.2%	97.8%
MgERG11	Wild‐type	63/129	66/129
		48.8%	51.2%
MgERG11	ΔMgKU70	1/88	87/88
		1.1%	98.9%

To show that this result was not specifically associated with the MgTUB1 locus on chromosome 1, a second construct was designed to replace the native MgERG11 gene present on chromosome 7 with a mutant allele (pNOV_ERG11_replace) (see Appendix S1 in Supporting Information for construct design details). The transformations were performed as before, with the PCR strategy to verify the correctly targeted gene replacement strains shown in Fig. 5A–C. The results showing the frequency of homologous recombination in the wild‐type and ΔMgKU70 strain are given in Table 3. This increased to more than 98% in the ΔMgKU70 strain, compared with approximately 50% in the wild‐type.

Figure 5.

Targeted gene replacement at the MgERG11 locus. (A) Diagram showing the construct used to replace the wild‐type MgERG11 with a resistant allele. (B) Diagram showing the native MgERG11 locus. (C) Diagram showing the replaced MgERG11 locus after transformation in the correctly targeted strains. The primers used to verify the transformants in which the MgERG11 gene has been correctly replaced are shown. Primers erg11 fwd‐int/erg11 rev‐int amplify a product of 492 bp when the wild‐type MgERG11 locus is present and a 1876‐bp product in all transformants. Primers erg11 fwd‐ext/erg11 rev‐ext produce a product of 1750 bp only when correctly targeted gene replacement has occurred, but no product when an ectopic insertion is present.

Insertion of a single point mutation into a native gene by targeted gene replacement

The construct generated to replace the native MgTUB1 gene with a mutant allele included the point mutation E198A, inserted by site‐directed mutagenesis. To investigate whether the MgTUB1 homologous recombinants generated above contained the A198 allele, a fragment of the MgTUB1 gene was amplified from the transformant, and a subsequent diagnostic digest was performed using HgaI to distinguish between E198 and A198 (Fig 6A, B). The results in Table 4 show that not all the homologous recombinants verified by PCR contain the A198 mutation. At this particular locus, 9% (7/78) of homologous recombinants (generated in both the wild‐type and ΔMgKU70 parental strains) lacked the mutated allele. Although the incorporation of the A198 allele occurred at a similar frequency regardless of whether the wild‐type or the ΔMgKU70 strain was used for the transformation, the overall frequency was higher in the ΔMgKU70 strain because the homologous recombinants were generated more efficiently in this strain. The presence of the A198 allele was confirmed by sequencing three M. graminicola strains generated in both the ΔMgKU70 and wild‐type background. The E198 allele was also confirmed by sequencing two M. graminicola strains generated in both the ΔMgKU70 and wild‐type background. Southern blot analysis was then performed on eight strains, two of each allele in the wild‐type and ΔMgKU70 parental background, and in all strains there was only a single copy of the inserted DNA, indicated by a shift in hybridization product from 1.395 to 2.797 kb in the transformed strains compared with the wild‐type, as a result of the presence of the hygromycin resistance cassette (4, 7). In addition to this expected shift, Southern blot analysis also showed the presence of an extra band in both wild‐type and transformed strains. A blast search was performed against the M. graminicola genome sequence, using the MgTUB1 DNA sequence, and this indicated that there was a region on chromosome 19 of approximately 300 bp which showed high identity (>90%) to part of the MgTUB1 gene. This additional band may therefore represent this MgTUB1 pseudogene.

Figure 6.

Verification of the MgTUB1 gene replacement transformants containing the A198 allele. (A) Diagram showing the position of the primers used to amplify a fragment of the MgTUB1 gene, prior to restriction digestion of the polymerase chain reaction (PCR) product with the enzyme HgaI. Primers BtubHRverF4/BtubHRverR3 amplify a 501‐bp product in all transformants. The A198 mutation encodes an HgaI site not present in the wild‐type E198 allele and, when the PCR product is digested with HgaI, results in two diagnostic bands at 340 and 161 bp. (B) Agarose gel showing the HgaI digest of the PCR products for seven of the MgTUB1 gene replacement transformants. Sample 5 contains the E198 allele, as only the undigested 501‐bp product is present, where as samples 1–4, 6, 7 all have the desired A198 allele, the 501‐bp product is absent and the diagnostic 340‐bp and 161‐bp bands are present. As expected, the wild‐type control displays only the 501‐bp band, whereas an ectopic transformant shows all three bands.

Table 4.

The frequencies at which the A198 allele was incorporated into the MgTUB1 gene in the gene replacement strains generated in both the wild‐type and ΔMgKU70 parental strains.

Gene	Parent strain	Digest E198	Digest A198
MgTUB1 gene replacement transformants	Wild‐type	4/33	29/33
		12.1%	87.9%
MgTUB1 gene replacement transformants	ΔMgKU70	3/45	42/45
		6.7%	93.3%

Figure 7.

Southern blot of the MgTUB1 gene replacement transformants transformed into the wild‐type (wt) (K4418) or ΔMgKU70 (K4903) background. Genomic DNA was digested with EcoRI and probed with a fragment of MgTUB1, which spans an internal EcoRI site. K4913 (T1), K4914 (T2) = A198 in ΔMgKU70, K4918 (T3), K4919 (T4) = E198 in ΔMgKU70, K4920 (T5), K4921 (T6) = E198 in wild‐type, K4922 (T7), K4924 (T8) = A198 in wild‐type. The ∼10‐kb band is upstream from the internal EcoRI site in the MgTUB1 probe and is common to transformed and untransformed strains. Untransformed wt and ΔMgKU70 strains display a 1.395‐kb band. T1–T8 all contain a single‐copy insertion of the MgTUB1 gene replacement cassette, indicated by a shift in size from 1.395 kb to 2.797 kb, corresponding to the presence of the hygR cassette. The ∼1.5‐kb band in all strains may result from hybridization to an MgTUB1 pseudogene (see main text for further details).

The presence of the A198 or E198 allele in the gene replacement strains provides an interesting insight into where the crossover occurs between two pieces of DNA in M. graminicola during homologous recombination. All the samples analysed for the presence of the E198A mutation were homologous recombinants confirmed by PCR. For the strain to contain the hygromycin gene, but the wild‐type allele at position 198, the crossover between the two pieces of DNA would have had to occur somewhere within the 910 bp between the hygromycin resistance cassette and the mutation, rather than nearer the 5′ end of the 3566‐bp fragment of DNA (Fig. 8A). In the strains with the A198 mutant allele, the crossover between the two pieces of DNA would have occurred at a distance greater than 913 bp from the hygromycin resistance cassette (Fig. 8B). This is the first indication in M. graminicola of the variability of the position of homologous recombination between two fragments of DNA.

Figure 8.

Homologous recombination at the MgTUB1 locus. (A) Diagram showing homologous recombination occurring in the 0.910 kb between the hygromycin resistance cassette and the A198 mutation, thus resulting in the incorporation of the E198 allele in the gene replacement transformant. (B) Diagram showing homologous recombination occurring in the 2.653 kb between the A198 mutation and the 5′ end of the DNA fragment, thus resulting in the incorporation of the A198 allele in the gene replacement transformant.

The A198 allele of β‐tubulin confers resistance to benomyl in M. graminicola.

In field isolates of M. fijiensis (Canas‐Gutierrez et al., 2006), O. yallundae and O. acumformis (Albertini et al., 1999) and a laboratory‐generated strain of N. crassa (Orbach et al., 1986), the A198 allele has been reported to confer resistance to the fungicide benomyl. The benomyl in vitro sensitivity assay showed that the presence of the A198 allele in β‐tubulin gave very high resistance to the fungicide when the allele was inserted into either the wild‐type or the ΔMgKU70 strain. The IC₅₀ value for the strains with the E198 allele ranged from 0.135 to 0.203 p.p.m. benomyl (average of 0.155 p.p.m.), whereas the strains containing the A198 allele were not inhibited at 1000 p.p.m. The estimated resistance factor is therefore >5000 (Fig. 9). These results show that it is now possible to make very precise manipulations to a specific gene in M. graminicola.

Figure 9.

The percentage growth inhibition to benomyl of MgTUB1 gene replacement strains, generated in both the wild‐type and ΔMgKU70 parental background. K4903 = ΔMgKU70 strain (wt MgTUB1); K4418 (wt) = wild‐type; K4913 and K4914 = A198 in ΔMgKU70; K4918 and K4919 = E198 in ΔMgKU70; K4920 and K4921 = E198 in wild‐type background; K4922 and K4924 = A198 in wild‐type background. The average IC₅₀ value for the strains with the wild‐type E198 allele is 0.155 p.p.m. benomyl. All four strains with the A198 allele are not inhibited by 1000 p.p.m. benomyl, a resistance factor estimated to be >5000.

DISCUSSION

In this study, we have reported the development of new capabilities for the transformation of M. graminicola which, together with the release of the M. graminicola genome sequence and the large‐scale gene expression analysis studies (Kema et al., 2008; 2000, 2005a, 2005b), will aid in the functional analysis of genes that play key roles in the biology, pathogenicity and virulence of this economically important pathogen.

We have shown the successful adoption of two further selectable markers for M. graminicola transformation, which have not been shown previously to work in this pathogen. The successful establishment of G418 and carboxin selectable markers now gives researchers a choice of three selectable markers for their studies. This will allow the selection of multiple gene disruptions or mutations simultaneously in M. graminicola for the first time. This ability will aid in the elucidation of the roles of multiple proteins, or how proteins within a family or within a pathway or network complement one another's functions.

The availability of an additional selectable marker enabled the disruption of the MgKU70 gene, followed by subsequent transformations into this ΔMgKU70 strain using a different resistance gene to select for transformants. The aim of the generation of this strain was to improve the efficiency of homologous recombination. This was successfully achieved, with homologous recombination in more than 95% of transformants for MgTUB1 and MgERG11 in the ΔMgKU70 strain, compared with 71% and 51%, respectively, in the wild‐type strain. As with many other fungi (Ninomiya et al., 2004; Villalba et al., 2008), the lower frequencies of homologous recombination in the wild‐type indicate that the NHEJ pathway is the major mechanism for double‐stranded DNA break repair in M. graminicola. The comparative in vitro growth assay and in planta pathogenicity assays revealed that the ΔMgKU70 strains grow in a similar manner to the wild‐type strain, thus validating their applicability for performing targeted gene disruptions of any hypothesized pathogenicity genes. These results are similar to those reported for Magnaporthe oryzae (Villalba et al., 2008), A. nidulans (Nayak et al., 2006) and A. fumigatus (Krappmann et al., 2006) when the KU70 gene was disrupted. Targeted gene disruptions in M. graminicola performed in this ΔMgKU70 strain will be achieved much more efficiently in the future.

We have shown that it is possible to perform precise manipulations to specific genes in M. graminicola, and believe that this is the first report of this approach in this pathogen. Here, a specific mutation E198A, conferring resistance to the fungicide benomyl in M. graminicola, as mentioned by Fraaije et al. (2007), was successfully inserted into MgTUB1. The resistance to benomyl was confirmed in an in vitro sensitivity assay, where strains with the A198 allele had a resistance factor of more than 5000 compared with those with the E198 allele. These data indicate that the MgTUB1(E198A) benomyl resistant gene could also be used as a selectable marker for Agrobacterium‐mediated M. graminicola transformation. This potential has been reported previously by Payne et al. (1998). We also observed that 9% of the MgTUB1 gene replacement strains from both the wild‐type and ΔMgKU70 parental strains did not contain the A198 allele, despite the allele being 2.6 kb from the end of the fragment. This scenario can only have resulted from homologous recombination occurring downstream of the single nucleotide polymorphism, in this case at a distance of less than 0.910 kb from the hygromycin cassette (Fig 8A, B). This variability is particularly important to consider when designing constructs to engineer a precise change to the sequence of a gene of interest, so as to maximize the chances of the transformants generated containing the desired sequence.

The new tools for M. graminicola research reported here, used in combination with the M. graminicola genome sequence and associated genomics technologies, open up the way for many exciting functional genomics studies in the future, with the results arising from such studies helping to guide future research in this field.

EXPERIMENTAL PROCEDURES

Fungal strains and growth conditions

Mycosphaerella graminicola isolate IPO323 was used for all experiments. The isolate was inoculated from stocks stored in liquid nitrogen onto solid V8 agar (200 mL V8 juice, 3 g CaCO₃, 800 mL H₂O, 20 g agar oxoid number 3). Yeast‐like cells were harvested from these plates and used immediately as inoculum for all experiments. The following media were used throughout: AMM (6 g NaNO₃, 0.52 g KCl, 1.52 g KH₂PO₄, 0.52 g MgSO₄.7H₂O, 10 g glucose, 1 mL trace element solution, 1 L H₂O, pH 6.5 with KOH, 15 g bacto agar); glycerol yeast extract (GlYE) medium (10 g yeast extract, 6 g NaNO₃, 1.5 g KH₂PO₄, 0.5 g KCl, 0.5 g MgSO₄.7H₂O, 16 mL 100% glycerol, 984 mL H₂O, 20 g bacto agar); induction medium (IM) (10 mm KH₂PO₄, 10 mm K₂HPO₄, 2.5 mm NaCl, 2 mm MgSO₄.7H₂O, 0.7 mm CaCl₂, 9 µm FeSO₄, 4 mm (NH₄)₂SO₄, 10 mm glucose, 0.5% glycerol, 40 mm MES buffer, 1 L H₂O, pH 5.6–5.8, with 20 g bacto agar added for preparing the plates); yeast sucrose medium (YSM) (10 g yeast extract, 10 g sucrose, 1 L H₂O); potato dextrose agar (PDA) (24 g potato dextrose broth, 1 L H₂O, 20 g bacto agar); Vogels minimal medium [15 g sucrose, 20 mL Vogels concentrate (125 g C₆H₅Na₃O₇.2H₂O, 250 g KH₂PO₄, 100 g (NH₄)₂SO₄, 10 g Mg(NO₃)₂.6H₂O, 5 g CaCl₂ dissolved in 20 mL H₂O, 2.5 mL biotin solution, 5 mL trace element solution, 750 mL H₂O, 3 mL chloroform added as a preservative and stored at 4 °C), 1500 mL H₂O and 4.5 g bacto agar added for 0.3% semi‐solid agar].

Bacterial strains

DH5α was used for the maintenance of plasmids in E. coli, unless otherwise stated. Agrobacterium tumefaciens strain EHA105 (Hellens et al., 2000; Hood et al., 1993) was used for the maintenance of constructs and for A. tumefaciens‐mediated transformation. Both were grown in Luria–Bertani (LB) broth, unless otherwise stated.

Identification of MgKU70 homologue, MgTUB1 gene and MgERG11 gene

The M. graminicola homologue of the N. crassa MUS‐51 (KU70) gene was identified prior to M. graminicola genome release by performing a blastp search with the N. crassa sequence (GenBank accession no. Q7SA95) against the M. graminicola single reads, when they were first released by JGI. This identified a single hit. On release of the complete M. graminicola genome sequenced by the JGI (http://genome.jgi‐psf.org/Mycgr3/Mycgr3.home.html), the MgKU70 gene was identified as fgenesh1_pm.C_chr_3000110, protein id 85040, which resides on chromosome 3. The sequence accession number for the TUB1 gene, encoding β‐tubulin, from GenBank is Q6QDC9, and from the JGI complete genome is geneID estExt_fgenesh1_kg.C_chr_11047, protein ID 102950, and it is present on chromosome 1. The identifiers for the ERG11 gene, encoding lanosterol 14‐α‐demethylase, from the JGI complete genome are estExt_fgenesh1_pg.C_chr_70359, protein ID 110231, and it is present on chromosome 7.

Plasmid construction

The binary vector pNOV2114 was used for all A. tumefaciens‐mediated transformation experiments. This is based on the pPZP family of binary vectors (Hajdukiewicz et al., 1994), but has been modified to contain a mutated copy of the virG (N54D) gene on the plasmid backbone (Pazour et al., 1992; Scheeren‐Groot et al., 1994). The following plasmids were generated: (i) pNOV2114gateway; (ii) pNOV_SDHS_H267Y; (iii) pNOVΔMgKU70; (iv) pNOV_βtubulin_replace; (v) pNOV_βtubulinA198_replace; (vi) pNOV_ERG11_replace. A detailed description of their construction is given in Appendix S1 in Supporting Information.

Mycosphaerella graminicola transformation

Transformation was performed as described by Zwiers and De Waard (2001) with the following modifications. Agrobacterium tumefaciens strain EHA105 was transformed with the binary vectors for M. graminicola transformation by electroporation following standard procedures, and subsequent strains were maintained in LB medium supplemented with 100 µg/mL spectinomycin (Sigma, Gillingham, Dorset, UK) and 100 µg/mL rifampicin (Sigma). Transformed A. tumefaciens strains were grown overnight in LB medium amended with 100 µg/mL spectinomycin and 100 µg/mL rifampicin at 28 °C, 250 r.p.m. The overnight cultures were diluted to an optical density (OD) of 0.15 at 660 nm in IM, amended with 100 µg/mL spectinomycin, but no acetosyringone, and cultured at 28 °C, 250 r.p.m. until an OD of 0.3–0.35 at 660 nm was reached. The A. tumefaciens cultures were then mixed in equivolume with M. graminicola yeast‐like cells which had been harvested from 4–5‐day‐old V8 plates, and diluted to a concentration of 5 × 10⁷ or 1 × 10⁷ cells/mL in H₂O. One hundred microlitres of the A. tumefaciens–M. graminicola mixtures were plated onto nitrocellulose filters placed on IM agar plates containing 100 µg/mL spectinomycin, but no acetosyringone, and incubated at 19 °C for 2 days. No acetosyringone was required as the pNOV2114 plasmid contains a mutated virG gene (N54D), allowing the constitutive expression of virG even in the absence of acetosyringone (Pazour et al., 1992; Scheeren‐Groot et al., 1994). Filters were then transferred onto AMM plates containing 250 µm cefotaxime (Sigma), 10 µg/mL ampicillin (Sigma), 100 µg/mL streptomycin (Sigma) and either 100 µg/mL hygromycin (Sigma) or 250 µg/mL G418 (Sigma) for selection with the hygromycin resistance cassette or nptII resistance gene, respectively. For selection with carboxin, filters were transferred onto GlYE plates containing 250 µm cefotaxime, 100 µg/mL ampicillin, 100 µg/mL streptomycin and 20 or 40 µg/mL carboxin. The plates were incubated at 19 °C for 10–14 days, and colonies which appeared after this time were transferred to PDA amended with hygromycin at 100 µg/mL or G418 at 250 µg/mL and the antibiotics mentioned above when hygromycin or G418 was used as the selectable marker, or to GlYE medium amended with 40 µg/mL carboxin and the antibiotics mentioned above when carboxin was used as the selectable marker. These plates were incubated at 19 °C for about 7–10 days until mycelial colonies were present. The transformants were then subjected to a further round of selection as described above, prior to verification by PCR to confirm that they were true transformants.

Analysis of transformants

For PCR verification of transformants, genomic DNA was extracted using the Wizard® magnetic 96 DNA plant system (Promega, Southampton, Hampshire, UK) which had been adapted for use on the Biomek FX robot (Beckman Coulter, High Wycombe, Buckinghamshire, UK). In all cases, PCRs were set up using 2 × taq PCR master mix, 2 µL of genomic DNA as template and oligo concentrations of 1 mm.

For transformations using G418 as the selectable marker, positive transformants were confirmed using the primers kanRverF and kanRverR (Table S1, see Supporting Information). For transformations using carboxin as the selectable marker, positive colonies were confirmed using the primers SDHBverF and SDHBverR (Table S1). To verify that the MgKU70 gene had been correctly deleted by homologous recombination, the following primers were used as described in Fig. 2A–C: Ku70verF1/Ku70verR1 and Ku70verF2/Ku70verR2 (Table S1). To verify that the wild‐type MgTUB1 gene had been correctly targeted by homologous recombination, and replaced with the mutant allele, the following PCR primers were used as described in Fig. 4A–C: βtubHRverF1/βtubHRverR1 and βtubHRverF2/hygromycin1 (Table S1). To verify that the wild‐type MgERG11 gene had been correctly targeted by homologous recombination, and replaced with the mutant allele, the following PCR primers were used as described in Fig. 5A–C: erg11 fwd‐int/erg11 rev‐int and erg11 fwd‐ext/erg11 rev‐ext (Table S1).

To confirm the presence of the A198 allele in the MgTUB1 homologous recombinants, a fragment of the MgTUB1 gene covering position 198 was amplified using primers βtubHRverF4/βtubHRverR3 (Table S1), and subsequently digested using HgaI (New England Biolabs, Hitchin, Herts, UK) (Fig. 6A). The PCR product was also cloned into TOPO TA vector pCR2.1 (Invitrogen, Paisley, Renfrewshire, UK), and sequenced.

Southern blot analysis was performed by standard procedures (Sambrook et al., 1989). One millilitre of M. graminicola yeast‐like cells at a concentration of 1 × 10⁷ cells/ mL was used to inoculate 150 mL YSM and the cultures were grown for 3 days at 19 °C, 140 r.p.m. The mycelia were harvested by vacuum filtration through miracloth onto filter paper and washed in 0.5 m ethylenediaminetetraacetic acid (EDTA). Genomic DNA was extracted following the method of Lee et al. (1988), with the following modifications. On addition of the lysis buffer, the samples were mixed by inverting the tube until the mixture was homogeneous. After incubation at 65 °C, an equivolume of phenol–chloroform (1 : 1) was added, the samples were mixed by inverting the tubes, and centrifuged as described. The aqueous phase was transferred to a fresh tube and the phenol–chloroform extraction was repeated. The aqueous phase was transferred to a fresh tube and an equivolume of chloroform–indoleacetic acid (IAA) (1 : 1) was added, mixed by inverting the tube and centrifuged as described. The aqueous phase was transferred to a fresh tube, the DNA was precipitated with 3 m NaOAc and isopropanol then spooled into 70% ethanol. The pellet was air‐dried and resuspended in 300 µL TE (10 mm Tris, 5 mm EDTA) overnight at 4 °C, with RNase. Following RNase treatment, the DNA was cleaned by performing a phenol–chloroform, followed by chloroform–IAA, extraction, isopropanol precipitation of the DNA, and resuspension in 100–150 µL of TE buffer at 4 °C overnight. The DNA was digested with HindIII (MgKU70 gene disruptants) or EcoRI (MgTUB1 gene replacements) (New England Biolabs).

An 800‐bp fragment of the nptII gene was amplified by PCR with the primers 6R/5Fb (Table S1) as the probe for the MgKU70 gene disruptants. A 1‐kb fragment of the MgTUB1 gene was amplified by PCR using primers Tubulin‐For/Tubulin‐Rev2 (Table S1) as the probe for the MgTUB1 gene replacements. Both probes were labelled with dCTPα32P (Perkin‐Elmer, Cambridge, Cambridgeshire, UK). The Southern transfer and hybridizations were performed as described by Sambrook et al. (1989).

Phenotypic growth tests

To test the sensitivity to G418, AMM plates were prepared containing different concentrations of G418 (0, 25, 50, 100, 250, 500 and 1000 µg/mL) onto which nitrocellulose filters were placed. Mycosphaerella graminicola was diluted to a concentration of 5 × 10⁷ or 1 × 10⁷ yeast‐like cells/mL, and mixed with an equivolume of IM. One hundred microlitres of this were spread onto each plate, and the plates were incubated at 19 °C for 10–14 days until a lawn had developed on the plates with no selection. To test the sensitivity to the fungicide carboxin, the experiment was repeated with the following alterations. Mycosphaerella graminicola was grown on GlYE medium, containing the following concentrations of carboxin: 0, 5, 10, 20, 40, 80 and 160 µg/mL.

To test the sensitivity of the MgTUB1 A198 gene replacement strains to the fungicide benomyl, a 96‐well plate in vitro assay was performed. Benomyl was added to the plates at the following rates: the highest concentration of 1000 µg/mL was diluted fourfold through 11 dilutions in dimethylsulphoxide (DMSO), giving a lowest concentration of 0.00024 µg/mL with a 0 µg/mL control included. Transformed strains containing either the wild‐type E198 allele, or resistant A198 allele, generated in the wild‐type IPO323 parental strain or ΔMgKU70 strain, were harvested from 4‐day‐old V8 agar plates by adding 10 mL of Vogels minimal medium, agitating the plates to release the cells, and filtering the cell suspension through miracloth. The yeast‐like cells were diluted to 1.6 × 10⁶ cells/mL in Vogels minimal medium. Each cell suspension was mixed with an equivolume of Vogels semi‐solid 0.3% agar to give a final concentration of 8 × 10⁵ cells/mL, and 100 µL were added to the appropriate wells. The plates were incubated at 25 °C, 140 r.p.m. for 7 days, with OD readings at 595 nm being taken on days 0, 1, 2, 4 and 7. From these data, IC₅₀ values were calculated for each strain.

To compare the growth of the wild‐type IPO323 parental strain with the ΔMgKU70 strains generated (K4902 and K4903), yeast‐like cells were harvested from 4‐day‐old V8 agar plates, as described above for the benomyl sensitivity assay. Three replicate samples harvested from three separate plates were compared for each isolate. The inoculum was prepared and added to the appropriate wells on a 96‐well plate as described above. There were eight replicate wells for each sample. The plates were incubated at 25 °C, 140 r.p.m. for 7 days, with OD readings at 600 nm being taken on days 0, 1, 2, 3, 4 and 7.

The growth of the wild‐type IPO323 parental strain and the ΔMgKU70 strains generated (K4902 and K4903) on wheat was compared in an in planta pathogenicity assay, which was set up as follows. Frozen conidial cells were used to inoculate YSM cultures in Erlenmeyer flasks and adjusted to an OD of 0.05 at 600 nm. These were grown for 4 days on a reciprocal shaker at 18 °C. On day 3, the cultures were all adjusted to an OD of 0.5 at 600 nm. After day 4, the conidial cells were allowed to sediment overnight and the liquid medium was removed by aspiration. The resulting pellets were resuspended to 1 × 10⁷ cells/mL with distilled water, supplemented with 0.05% Tween‐20. Fourteen‐day‐old hexaploid wheat plantlets of the susceptible cultivar Riband, four per pot and 10 replicate pots per M. graminicola strain, were sprayed with conidial suspensions and left capped by PVC boxes for 48 h at 20 °C in the dark. Following this incubation period, the PVC boxes were removed and the plants were subjected to a day/night regime of 20 °C 16 h/18 °C 8 h in a controlled environment growth room. At 17 days post‐infection, disease severity was evaluated visually as the percentage of the leaf surface of the second leaf that was covered with symptoms.

Supporting information

Table S1 The primers used in this study.

Appendix S1 Details of plasmid construction.

Please note: Wiley‐Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

Supporting info item