• Blast disease (causal agent Magnaporthe oryzae) has presented as a new and serious field disease of wheat in South America. Here, we investigated the responses of wheat to both adapted and nonadapted isolates of the blast fungus Magnaporthe, examining cellular defence and transcriptional changes.
• Resistance towards the nonadapted isolate was associated with the formation of appositions, here termed halos, beneath attempted Magnaporthe grisea penetration sites that wheat-adapted, M. oryzae isolates were able to breach.
• Transcriptome analysis indicated extensive transcriptional reprogramming following inoculation with both wheat-adapted and nonadapted isolates of Magnaporthe. Functional annotation of many of the differentially expressed transcripts classified into the categories: cell rescue and defence, plant metabolism, cellular transport and regulation of transcription (although a significant number of transcripts remain unclassified).
• Defence-related transcripts induced in common by adapted and nonadapted isolates were differentially regulated in response to M. oryzae and M. grisea isolates over time. Differential expression of genes involved in cellular transport indicated the importance of this process in plant defence. Functional characterisation of these transcripts and their role in defence may eventually lead to the identification of broad-spectrum resistance mechanisms in wheat towards Magnaporthe.
Plants are under constant threat from a wide variety of microorganisms and have evolved different strategies to protect themselves against these would-be invaders. Initial lines of defence include physical barriers such as a waxy cuticle that helps prevent pathogen penetration (Carver et al., 1990), or the presence of constitutive phytotoxins such as avenacin in oats (Osbourn, 1996). These preformed defence mechanisms are supported by a range of inducible responses that must be overcome by the pathogen to successfully cause disease (Thordal-Christensen, 2003). Conserved molecules known as microbe- or pathogen-associated molecular patterns (MAMPs/PAMPs) enable the plant to recognise the microorganism as foreign, triggering early, PAMP-triggered immunity (PTI) defence responses (Zipfel, 2008). Microorganisms able to overcome PTI are believed to produce effector molecules that suppress early defence responses, enabling these now adapted pathogens to cause disease. In response plants have evolved specific R-genes that recognise pathogen effectors, either directly or indirectly, resulting in accelerated and stronger defence responses collectively termed effector-triggered immunity (ETI) (Jones & Dangl, 2006). When a pathogen is able to evade both PTI and ETI a compatible interaction is established, resulting in disease.
Most pathogens, however, are only adapted to cause disease on a limited number of plant species, resistance to nonadapted pathogens being referred to as nonhost resistance (Heath, 2000; Mysore & Ryu, 2004; Nürnberger & Lipka, 2005). Nonhost resistance is considered to be both genetically complex and durable, although it has been proposed that similar basal defence mechanisms operate against adapted and nonadapted pathogens (Thordal-Christensen, 2003). Where a close evolutionary relationship exists between the nonhost and the host plant species of a pathogen there is often evidence for a gene-for-gene-like resistance system operating against the nonadapted pathogen (Murakami et al., 2000; Oh et al., 2002; Tosa et al., 2006). Nonadapted pathogens may lack the effectors required to suppress PTI, or possess avirulence effectors that trigger multiple R-gene mediated ETI, either way the plant species expresses broad-spectrum resistance to the nonadapted pathogen (Mysore & Ryu, 2004; Jones & Dangl, 2006).
Blast disease caused by the ascomycete fungi Magnaporthe species can occur on > 50 graminaceous species (Ou, 1985). Phylogenetic analysis of Magnaporthe isolates has led to the classification of two distinct species (Couch & Kohn, 2002). Isolates pathogenic on cultivated cereals belong to the species Magnaporthe oryzae, while isolates pathogenic on wild grasses, including those of the wild crabgrass genus Digitaria, belong to the species Magnaporthe grisea.
A global picture of the reprogramming of the transcriptome that occurs following inoculation with adapted and nonadapted pathogens can now be obtained using array technologies. In cereals this technology has been used to identify changes in the transcriptome caused by a number of pathogens, including yellow and leaf rust in wheat (Hulbert et al., 2007; Coram et al., 2008a,b), powdery mildew (Caldo et al., 2004, 2006) and Polymyxa infection in barley (McGrann et al., 2009) and M. oryzae infection in rice (Vergne et al., 2007). Using the Affymetrix Wheat GeneChip (Santa Clara, CA, USA) differences and commonalities in the reprogramming of the wheat transcriptome have been studied in host compatible and race-specific incompatible interactions (Coram et al., 2008b) and in wheat cultivars expressing durable, adult plant-expressed resistance (Hulbert et al., 2007; Bolton et al., 2008; Coram et al., 2008a).
In this study we used the Affymetrix Wheat GeneChip to identify those genes and genetic pathways in wheat in common and specific to the interaction with wheat-adapted M. oryzae and nonadapted M. grisea isolates. The microscopic development of each Magnaporthe isolate and the cellular defence responses observed in wheat are presented in relation to the reprogramming of the wheat transcriptome. This study sets the foundations to identify those defence responses, genes and genetic pathways specific to resistance in wheat to adapted and nonadapted pathogens.
Materials and Methods
Magnaporthe isolates and plant material
Magnaporthe isolates BR32, BR37 and BR29 were obtained from Didier Tharreau (CIRAD, Montpellier, France). Magnaporthe oryzae isolates BR32 and BR37 were isolated from wheat while the M. grisea isolate BR29 was isolated from Digitaria sanguinalis (D. Tharreau, pers. comm.). Previous phylogenetic analyses placed the wheat-adapted and Digitaria-adapted isolates in distinct groups (Zellerhoff et al., 2006; Faivre-Rampant et al., 2008). Seed of the winter wheat cultivars Renan and Thésée (Triticum aestivum L.) were provided by Jean-Benoit Morel (INRA, Montpellier, France). ‘Renan’ was selected for this study because of the genetic resources available for this wheat cultivar. Digitaria sanguinalis seed was provided by Nina Zellerhoff (RWTH, Aachen University, Germany). The Magnaporthe isolate referred to as BR29 in the publication by Tsurushima et al. (2005) is not the same as the BR29 isolate used in this study.
Inoculation of wheat seedlings with Magnaporthe isolates
Magnaporthe isolates were grown for 11 d on Complete Media Agar at 25°C under a 16 h : 8 h light–dark cycle. Conidia were harvested by flooding the plates with 5 ml of sterile inoculation solution (0.25% (w : v) gelatine and 0.01% (v : v) Tween 20) and scraping the conidia from the surface using a sterile glass rod. Conidia were filtered through sterile miracloth and the density adjusted to 1 × 105 conidia ml−1 with inoculation solution.
Plants were grown in a peat and sand (1:1) mix at 23°C in a Sanyo Fitotron growth cabinet (Sanyo Gallenkamp PLC, Loughborough, UK) with a 16 : 8 h light–dark cycle. Fourteen-day-old wheat seedlings and 19-d-old D. sanguinalis seedlings were mist inoculated with 4 ml of conidia suspension. Plants were sealed in plastic propagators to maintain relative humidity at c. 100% and kept at 25°C in the dark for the first 24 h postinoculation (hpi). Subsequently, the lids of the propagators were removed and plants were returned to the growth cabinet. Control plants were misted with inoculation solution, without conidia (mock inoculation). Each Magnaporthe isolate was inoculated onto a pot of D. sanguinalis (10 seedlings per pot) and a pot of the wheat cv. Thésée as controls and six pots of the test cv. Renan (10 seedlings per pot). Leaf samples were collected at 4, 11, 18, 24, 48 and 72 hpi for both histopathology and transcriptomic analysis from three independent biological experiments. Unsampled seedlings were checked for macroscopic symptoms 96 hpi.
Histopathology of wheat–Magnaporthe interactions
Leaf tissue was cleared and fixed by immersing in chloral hydrate solution (300 ml 95% ethanol, 125 ml 90% lactic acid, 800 g chloral hydrate, made up to 1 l with chloroform; Garrood, 2001) for 5 d, changing the solution every 48 h, or until leaf tissue was translucent. Prepared leaf samples were stored in lactoglycerol (1 : 1 : 1, v : v, lactic acid–glycerol–water) at 4°C. Uvitex-2B (Ciba-Geigy, Basel, Switzerland) staining of fungal tissue was carried out as described in Moldenhauer et al. (2006), but with only 15 min preincubation in 0.05 m NaOH. Leaf samples were directly mounted onto glass slides in 40% glycerol and stored in the dark.
Stained fungal and plant autofluorescent cellular structures were observed on a Zeiss LSM 510 META confocal microscope using ×25 LD LCI Plan-Apochromat (numerical aperture 0.8) or ×40 EC Plan-Neofluar (numerical aperture 1.3) oil immersion objective lenses. Spectral data were collected by excitation with 488 nm Argon (30-mW) and 405 nm Diode (30-mW) lasers. Fungal structures stained with Uvitex-2B and plant autofluorescing cellular structures were differentiated using a Uvitex-2B specific filter (BP 420–480 nm) and an autofluorescence-specific filter (LP 530 nm), respectively. Bright-field light microscopy was used to identify epidermal and mesophyll cells. From each of the three independent biological experiments 3–4 cm of leaf tissue, from at least two seedlings, was prepared for analysis. At each time-point a minimum of 100 infection sites per Magnaporthe isolate were scored.
Aniline blue staining of leaf tissue cleared with chloral hydrate was used to identify callose deposition associated with Magnaporthe infection sites using confocal microscopy as described earlier, but substituting the aniline blue-specific filter (LP 420 nm) for the Uvitex-2B specific filter (BP 420–480 nm). Leaf tissue was stained in 100 mm KH2PO4 buffer (pH 9.6) with 0.01% aniline blue overnight at room temperature.
The histopathological examination of the Magnaporthe isolates BR29, BR32 and BR37 on the wheat cv. Renan, in three independent biological experiments was analysed using generalised linear mixed modelling (GLMM; Welham, 1993). A binomial distribution, with a logit transformation was used to compare the ratio of the growth stages APP, HALO, HYPFLUO and MULTIFLUO to the total number of infection sites observed. The model fitted compared replicate experiment, isolate and time point effects. Differences between isolates and time points significant at an F-value probability of P < 0.001 were further compared by t-test analysis. Quantitative reverse transcription polymerase chain reaction (qRT-PCR) relative transcript levels obtained for the three biological replicate experiments were compared using general linear regression. The model fitted compared replicate experiments, isolates and time points and isolate × time point interactions. Differences between isolates and time-points significant at an F-value probability P < 0.01 were further compared by t-test analysis. All analyses were carried out using the statistical package genstat for Windows, 9th edition (GenStat Release 9 Committee, 2007).
Leaf tissue was ground under liquid nitrogen and total RNA extracted using a QIAquick RNeasy Plant Extraction Kit (Qiagen), followed by TURBO DNase (Ambion) treatment. RNeasy Mini Spin column purification (Qiagen) was used to further purify RNA samples for array hybridization. RNA quality checks, cRNA conversion and Affymetrix genome array hybridization was carried out by the Nottingham Arabidopsis Stock Centre (NASC) array hybridisation service (http://affymetrix.arabidopsis.info/). The 24 hpi time-point, from three biological replications of the BR29, BR32, BR37 and mock-cv. Renan inoculations were hybridized to the Affymetrix Wheat genome array GeneChip following standard Affymetrix protocols, resulting in a total of 12 hybridizations (http://www.affymetrix.com). Data were normalised using Robust Multiarray Average measure (RMA) (Irizarry et al., 2003) and subsequently analysed with the LIMMA (Smyth, 2004) software package. Probe sets were identified as differentially expressed if they had a fold change greater than two relative to the control and an Empirical Bayes moderated t-statistic of P < 0.001 (Loennstedt & Speed, 2002).
Hierarchical cluster analysis
Transcript levels of probe sets identified as differentially expressed in response to all three Magnaporthe isolates compared with the mock-inoculated control were log2 transformed. The expression of each probe set was assumed to be stable in replicate mock-inoculated samples and therefore set at 1 (log2 = 0). cluster 3.0 (Eisen et al., 1998) was used to cluster the data by both probe sets and by treatments, using a complete-centred similarity matrix, and complete-linkage clustering. treeview, v.1.0.13 was used to visualise the data.
Quantitative RT-PCR time-course analyses of selected transcripts
RNA isolated from the 4, 11, 18, 24, 48 and 72 hpi time-points from each of the three wheat cv. Renan–Magnaporthe isolate inoculations and the mock-inoculated control were used to investigate selected transcript expression profiles over time in each interaction. Primers were designed for the wheat sequences covering the target region from which the probe sets were developed (see Supporting information Table S1). Each primer pair was validated with an efficiency > 80%. Primers for the probe sets TaAffx.92008.1.A1_s_at and Ta.1174.1.S1_x_at were from Coram et al. (2008b). cDNA for qRT-PCR analysis was synthesised from 1 µg of total RNA using the Superscript III reverse transcription kit (Invitrogen) primed with random hexamers. cDNA was amplified using the Sybr Green JumpStart Taq Ready mix (Sigma) and gene-specific primers (Table S1). The qRT-PCR amplification was done using a DNA engine Opticon2 Continuous Fluorescence Detector (MJ Research Inc., Alameda, CA, USA) with an initial activation step at 95°C for 4 min, followed by 40 cycles of 30 s at 95°C, 30 s at 60°C and 30 s at 72°C. Melt-curve analysis was performed at the end of each reaction to monitor primer–dimer formation and the amplification of gene-specific products. The average threshold cycle (CT) value for each gene was calculated from duplicate samples for each experiment. Data were analysed using Opticon Monitor analysis software v2.02 (MJ Research Inc.). Three housekeeping genes – ubiquitin, GAPDH and elongation factor-1α– were selected for normalisation using published primer sequences that had perfect matches to wheat sequences (Van Riet et al., 2006; McGrann et al., 2009). Using genorm v3.5 (http://medgen.ugent.be/~jvdesomp/genorm/; Vandesompele, et al., 2002) the expression of all three normalisation genes was shown to be stable under our experimental conditions. Gene expression levels were adjusted at each time point, in each cv. Renan–Magnaporthe isolate inoculation relative to the mock-inoculated control using the ΔCT method, following application of the normalisation factor calculated by genorm using GAPDH and ubiquitin as reference genes.
Infection phenotypes of adapted and nonadapted Magnaporthe isolates on wheat
Magnaporthe infection phenotypes were observed on wheat cvs Renan and Thésée 96 hpi (Fig. 1). On cv. Renan both wheat-adapted isolates produced lesions ringed by brown, necrotic tissue typical of the partial resistance phenotype seen in rice to M. oryzae (Ribot et al., 2008). The lesions were generally less than 3 mm in diameter and in addition to the ring of necrotic tissue the lesion sites sat within chlorotic leaf tissue, demonstrating that cv. Renan has partial resistance to isolates BR32 and BR37. On cv. Thésée lesions were not associated with necrotic rings of tissue. The grey lesions eventually coalesced, resulting in the complete blighting of the leaf blade typical of a susceptible interaction.
Neither wheat cultivar exhibited any macroscopic phenotype when inoculated with the nonadapted D. sanguinalis isolate BR29 (Fig. 1). An extensive screen of 36 wheat cultivars with BR29 failed to identify a wheat genotype on which BR29 produced infection symptoms (Table S2) supporting the selection of BR29 as a nonadapted Magnaporthe isolate for wheat. Inoculation of BR29 onto D. sanguinalis seedlings resulted in the development of large grey, water-soaked lesions confirming that BR29 had retained its pathogenicity on its host species (Fig. 1). The cv. Renan was used in all further experiments.
Magnaporthe development and corresponding cellular responses in the wheat cv. Renan
To determine the extent of fungal invasion and the plant responses to the wheat-adapted and nonadapted Magnaporthe isolates the infection process was microscopically evaluated on the wheat cv. Renan over 72 h following inoculation. Time-points were selected for examination based on the infection of Magnaporthe on rice, with 4, 11 and 18 hpi and 24, 48 and 72 hpi predicted to cover preinvasion and postinvasion stages of pathogen development, respectively (Kankanala et al., 2007). To facilitate quantification, infection sites defined as the formation of an appressorium were classified into four interaction stages: APP, appressorium formation with no associated plant response (Fig. 2a); HALO, autofluorescing halo formed beneath an appressorium (Fig. 2b); HYPFLUO, hyphae in first invaded epidermal cell associated with epidermal and mesophyll autofluorescence (Fig. 2c); and MULTIFLUO, hyphal growth spanning multiple cells, associated with epidermal and mesophyll cell autofluorescence in colonised as well as adjacent, non-colonised cells (Fig. 2d). Where hyphal growth was observed in an epidermal cell the mesophyll cells lying directly below it exhibited autofluorescence. Cytosolic granulation of the first invaded epidermal cell and collapse of the underlying mesophyll cells was also observed (Fig. 2c,d).
All three Magnaporthe isolates produced infection sites on cv. Renan. No appressoria were observed at 4 hpi, while by 11 hpi almost all germinated conidia had reached the APP stage (Fig. 3). The transition from APP to the HALO stage occurred more rapidly with isolate BR32, significant differences being seen at 18 hpi (t-test probability < 0.001). The number of infection sites at the APP stage then decreased over time in all wheat–isolate interactions, with only 4–5% of infection sites remaining at the APP stage at 72 hpi.
The nonadapted isolate BR29 was able to produce appressoria and attempt to infect the wheat cv. Renan as effectively as the wheat-adapted isolates (Fig. 3a). However, from 24 hpi it became clear that the majority of the BR29 infection sites were arrested in their development at the HALO stage (compared with BR32 and BR37 t-test probability < 0.001). The BR29 infection sites were seen to form hyphae within epidermal cells (HYPFLUO) by 72 hpi, but only rarely was a BR29 infection site seen to develop hyphae in additional plant cells (MULTIFLUO).
Both BR32 and BR37 developed past the HALO stage (Fig. 3b,c) forming hyphae within epidermal (HYPFLUO) and mesophyll cells (MULTIFLUO), with similar numbers of multicellular infection sites being observed at 72 hpi. No significant differences were seen between the three repeat experiments, supporting the biological reproducibility of the interactions between cv. Renan and the three Magnaporthe isolates. The halo response was occasionally accompanied by the formation of a denser, papillae-like structure. These papillae-like structures were observed in cv. Renan in interactions with both the adapted and nonadapted Magnaporthe isolates (data not shown). The halo structures formed beneath infection sites of both the adapted and nonadapted Magnaporthe isolates all contained callose, as indicated by aniline blue staining (Fig. S1).
Transcriptome profiling in wheat cv. Renan–Magnaporthe interactions
Transcriptional changes were monitored in the wheat cv. Renan 24 hpi with the adapted and nonadapted Magnaporthe isolates using the Affymetrix wheat genome array GeneChip. The 24 h time-point was selected to focus the study around the earliest point showing a clear difference in the histopathology of the adapted and nonadapted isolates. Comparable transcriptome studies in rice inoculated with adapted and nonadapted isolates of Magnaporthe also indicated that the 24 h time-point after inoculation was the most informative (collaborative programme with J. B. Morel, unpublished; Jantasuriyarat et al., 2005). Three separate comparisons (Magnaporthe-inoculated vs mock-inoculated) were performed on the transcriptome array data to identify differentially expressed transcripts responding to each of the three Magnaporthe isolates. Each probe set was assumed to represent a single transcript (Coram et al., 2008b) and functional annotation was assigned using those provided by Affymetrix (http://www.affymetrix.com/products/arrays/specific/wheat.affx) and from Plexdb (http://www.plexdb.org/plex.php?database=Wheat). Plexdb regularly updates probe set annotations through similarity searches to multiple publicly available databases (Wise et al., 2007). Transcripts were grouped into functional categories based on those outlined in the MIPS (Munich Information Centre for Protein Sequences) classification system (http://mips.gsf.de/).
A total of 961 probe sets were identified as differentially transcribed in response to the nonadapted isolate BR29 compared with the mock-inoculation, of which 424 probe sets were upregulated and 537 were downregulated. The wheat-adapted isolates BR32 and BR37 caused changes to the transcriptional expression of 564 and 414 probe sets, respectively. BR32 inoculation induced 322 transcripts and repressed 242, while BR37 inoculation resulted in 281 probe sets being upregulated and 133 downregulated (Tables S3, S4). Functional classification of the differentially expressed transcripts indicated that functionally similar transcripts were being regulated in each wheat–Magnaporthe isolate interaction (Fig. 4). The categories cell rescue and defence, metabolism, cellular transport and functions unclassified were most highly represented in the upregulated probe sets, while transcripts involved in transcription or with unclassified functions were most prevalent within the downregulated sequences.
A large proportion of the transcriptionally regulated probe sets were common to all three wheat–isolate interactions tested. There were 169 upregulated and 104 downregulated transcripts that followed the same pattern of expression irrespective of the inoculated Magnaporthe isolate. Inspection of the annotations of these probe sets highlighted the upregulation of general defence-related genes, including β-1,3-glucanases (PR2) and chitinases and the cell wall defence-related WIR proteins (Table S3). Probe sets predicted to code for enzymes relating to the production of phenolic-based secondary metabolites were also upregulated and included shikimate kinase and phospho-2-dehydro-3-deoxyheptonate aldolase of the shikimate pathway, and phenylalanine ammonia-lyase (PAL), caffeoyl-CoA O-methyltransferase, cinnamyl alcohol dehydrogenase and 4-coumarate-CoA ligase of the phenylpropanoid pathway. Furthermore, probe sets predicted to code for key enzymes in the biosynthesis of compounds involved in defence signalling pathways were upregulated in all three interactions, including ethylene biosynthesis (1-aminocyclopropane-1-carboxylic acid (ACC) synthase) and the biosynthesis of oxylipins such as jasmonic acid (12-oxophytodienoic acid reductase, allene oxide synthase). Transcripts downregulated in response to inoculation by BR29, BR32 and BR37 included expansin-like genes, putative cold-induced proteins and EF-hand calcium-binding proteins (Table S4). Hierarchical cluster analysis of the 273 probe sets transcriptionally regulated by all three of the Magnaporthe isolates showed that although these genes were responding in a nonspecific manner to the Magnaporthe isolates, within this cluster the responses to the wheat-adapted Magnaporthe isolates BR32 and BR37 grouped together, while the transcription profile responding to the nonadapted, BR29 isolate formed an outlying group (Fig. S2).
A significant number of the differentially expressed transcripts responded specifically to inoculation with the nonadapted Magnaporthe isolate BR29, 156 probe sets being upregulated and 300 transcripts being downregulated (Fig. 5; Tables S3, S4). Many of these transcripts represented genes involved in the regulation of transcription. Transcription factors of the WRKY, BHLH and NAC domain-containing families were generally found to be upregulated in all three isolate interactions, while transcription elongation factors appeared to be repressed. Transcripts annotated as different MYB or bZIP transcription factors appeared to show both induced and repressed expression patterns (Tables S3, S4). BR29 also specifically induced 24 transcripts annotated to be involved in cellular transport (37% of the transcripts in this group). Of the 24 transcripts many were putative sugar transporters, a classification not induced in response to inoculation with the wheat-adapted isolates. However, most of the transcripts differentially expressed in response to BR29 had no annotation, 31% (48/156) of the upregulated probe sets and 47% (141/300) of the downregulated transcripts having no predicted function. Furthermore, isolate-specific transcriptional changes were also evident between the two wheat-adapted isolates (Fig. 5; Tables S3, S4). Although no specific pathways could be clearly identified, these isolate-specific transcripts may have important roles in defining the specific responses to adapted and nonadapted Magnaporthe isolates in wheat.
Quantitative RT-PCR time-course analyses of selected transcripts
A qRT-PCR analysis was used to confirm the validity of the data obtained from the transcriptome arrays. Probe sets were selected that represented transcripts involved in the common response towards all three Magnaporthe isolates, as well as those identified as responding specifically to the wheat-adapted isolates, BR32 and BR37, or the nonadapted BR29 isolate. Statistical analysis confirmed that all transcripts tested showed significant differences in transcript levels over time and that no significant differences were apparent between the biological replications.
Three defence-related transcripts induced by all three isolates (PAL, TaAffx.92008.1.A1_s_at; ß-1,3-glucanase, Ta.1174.1.S1_x_at; and WIR1B, Ta.97.1.S1_at) typically showed higher transcript levels at the earlier time-points in cv. Renan inoculated with BR29. However, from 18 hpi onwards transcript levels were higher when inoculated with the two wheat-adapted isolates, with a stronger response being induced by BR32. An isolate × time-point interaction was detected for isolate BR32 which induced higher transcript levels of the three defence-related transcripts at specific time-points (Fig. 6a–c; t-test probabilities between 0.001 and 0.050).
12-Oxophytodienoic acid reductase (TaAffx.128684.1.S1_at), a key enzyme in oxylipin biosynthesis was upregulated at 24 hpi in the array analysis following inoculation by all three isolates. In the time-course analysis amounts of 12-oxophytodienoic acid reductase were not substantially different over the first 48 h following inoculation, but increased significantly at 72 hpi in the two wheat-adapted isolate inoculations compared with BR29 (Fig. 6d; t-test probabilities BR32 = 0.1 and BR37 = 0.15).
The putative peptide transporter Ta.1057.1.A1_at and the cellulose synthase-like transcript Ta.4084.1.S1_at were induced specifically by BR29 in the array analysis. The qRT-PCR showed Ta.1057.1.A1_at to be strongly induced at 11 hpi and 24 hpi by BR29 (t-test probabilities between 0.05 and 0.09) while transcript levels varied little over the time-course in cv. Renan inoculated with the two wheat-adapted isolates (Fig. 6e). The cellulose synthase-like transcript showed slightly higher amounts in cv. Renan inoculated with BR29 at 11 hpi and 24 hpi; however, at 18 hpi the amounts were higher in cv. Renan inoculated with the wheat-adapted isolates (Fig. 6f; t-test probability BR29 vs BR32 = 0.014).
The expression profiles of two wheat-adapted specific transcripts TaMPK3, a wheat mitogen-activated protein kinase (Ta.236.1.S1_at; Rudd et al., 2008) and Blufensin1, a proposed negative regulator of penetration resistance towards Blumeria graminis f. sp. hordei in barley (TaAffx.26815.1.S1_at; Meng et al., 2009) were examined over time. TaMPK3 showed similar levels of induction by BR29 at all time-points, while increasing in cv. Renan in response to inoculation with both wheat-adapted isolates between 48–72 hpi (Fig. 6g; t-test probabilities between 0.150 and 0.200 compared with BR29). Blufensin1, showed similar transcription patterns to the defence-related transcripts, with the level of transcription being higher in cv. Renan inoculated with BR29 at 4 hpi and 11 hpi, with subsequent higher levels being induced by BR32 (t-test probability 0.033) and BR37 (t-test probability 0.01) at later time-points.
The Digitaria-adapted Magnaporthe isolate BR29 was unable to infect any of the wheat genotypes tested. Development of BR29 on cv. Renan was effectively halted at the HALO stage, with only a few infection sites producing hyphae. Similarly BR29 was unable to infect barley genotypes, as were Magnaporthe isolates collected from Pennisetum (Zellerhoff et al., 2006). On barley a Pennisetum-adapted M. grisea isolate and a barley-adapted M. oryzae isolate both induced a strong papillae response, however by 96 hpi a number of infection sites formed by the barley-adapted isolate had grown away from the papillae, producing hyphae (Zellerhoff et al., 2006). The authors proposed that the papillae produced beneath the appressoria of the barley-adapted isolate were in some way less effective, enabling the adapted isolate to bypass this defence response. A difference in the chemical and/or structural composition of the papillae produced in barley and the halos produced in wheat, in response to attempted penetration by nonadapted isolates of Magnaporthe may account for the greater proportion of infection sites halted in their development at this interaction stage. In goosegrass failed penetration of a nonadapted M. oryzae isolate was associated with the deposition of a silica-based granular deposition beneath appressoria (Heath et al., 1992), while constitutive expression of the barley gene HvRAC1, a member of the Ras superfamily of small G-proteins, prompted callose deposition at sites of attempted Magnaporthe penetration in barley, resulting in a subsequent increase in resistance (Pathuri et al., 2008).
On rice, arrest of the nonadapted Digitaria isolate BR29 was not associated with any visible appositions beneath the appressoria, as could be discerned by epifluorescence microscopy or 3,3′-diaminobenzidine (DAB) staining (Faivre-Rampant et al., 2008). However, BR29 was not seen to produce hyphae and only a small number of infection sites were seen associated with epidermal cell autofluorescence. In rice, therefore, the formation of visible appositions beneath appressoria does not appear to be required to prevent penetration of the first epidermal cell by BR29.
Hyphal growth was almost invariably coupled with autofluorescence and cytoplasmic granulation of the first invaded epidermal cell (HYPFLUO). Cytoplasmic granulation of the first invaded epidermal cell has been associated with race-specific resistance to adapted Magnaporthe isolates in rice (Koga & Kobayashi, 1982; Tomita & Yamanaka, 1983). Escape from the first invaded cell was rarely observed with the nonadapted isolate BR29 whereas hyphae of the adapted isolates went on to infect multiple cells, resulting in collapse and autofluorescence of the adjacent mesophyll cells (MULTIFLUO).
Considerable reprogramming of the wheat transcriptome was observed in cv. Renan after inoculation with both wheat-adapted and nonadapted Magnaporthe isolates. A significant number of transcripts associated with cell rescue and defence were upregulated at 24 hpi, of which 41% were common to all three isolate interactions, the remaining 59% being either specific to one isolate or in common between only two isolate inoculations, with 18% being specific to the BR29 inoculation. This may indicate that specific defence-related transcripts may be involved in resistance towards the nonadapted Magnaporthe isolate. However, subsequent qRT-PCR analysis of three independent time-course experiments indicated that caution must be applied when interpreting the results of a single time-point array analysis.
The early induction of the three defence-related transcripts and the negative-regulator of penetration resistance Blufensin1 in response to the nonadapted isolate BR29 may form part of a PTI response that is suppressed by the wheat-adapted isolates. The high transcript levels present between 4 and 11 hpi may contribute to the significant number of BR29 infection sites halted in their development at the HALO stage, as first observed 18 hpi, while the stronger transcript induction observed at the later time points towards BR32 and BR37 being a part of the ETI response of the partial resistance expressed in cv. Renan (Jones & Dangl, 2006). BR32 was also more aggressive than BR37, producing significantly more HYPFLUO by 18 hpi and this may account for the higher defence-related transcript levels induced by BR32.
Gene silencing of transcripts involved in monolignol biosynthesis, including PAL has demonstrated a critical role for these genes in penetration resistance in wheat against both adapted and nonadapted isolates of the powdery mildew fungus B. graminis (Bhuiyan et al., 2009). Early induction of PAL may be indicative of the deposition of lignin-like compounds within the halo structures beneath BR29 that are not deposited within halos beneath the two wheat-adapted isolates. WIR1-like transcripts have been shown to be induced in wheat in response to both biotrophic and necrotrophic pathogens (Bull et al., 1992; Hulbert et al., 2007; Coram et al., 2008a,b; Desmond et al., 2009). WIR1 may have a potential role in maintaining attachment of the plasma membrane to the cell wall, possibly preventing excessive cell damage in regions undergoing a hypersensitive cell death reaction in response to pathogen attack (Bull et al., 1992). The higher levels of TaMPK3 induced by BR32 and BR37 may relate to the greater cellular autofluorescence and granulation seen in cv. Renan towards these wheat-adapted isolates at 48 hpi and 72 hpi during the presumed necrotrophic phase of Magnaporthe infection, TaMPK3 having been implicated in the programmed cell death response in wheat inoculated with compatible isolates of the necrotroph Mycosphaerella graminicola (Rudd et al., 2008).
All three Magnaporthe isolates had a significant effect on plant metabolism, with approximately one quarter of the upregulated transcripts being involved in metabolism and 34% of these transcripts being induced in common following inoculation with all three isolates. Enzymes key to the formation of ethylene and oxylipins were upregulated in all three interactions. Ethylene biosynthesis has been shown to contribute to M. oryzae host resistance in rice (Iwai et al., 2006) and oxylipins, including jasmonic acid have been shown to play a pivotal role in disease resistance (Farmer et al., 2003). Kinetic expression analysis of 12-oxophytodienoic acid reductase 1, an enzyme involved in the biosynthesis of oxylipins, showed a significant increase in transcription by 72 hpi in the wheat-adapted isolate interactions.
The fourth largest group of upregulated transcripts was that predicted to have a role in plant cellular transport. Inoculation with the nonadapted isolate BR29 specifically induced 24 transcripts (37% of transcripts in this group) annotated to be involved in cellular transport. A BR29-specific putative peptide transporter was found to be strongly induced at 11 hpi and 24 hpi in response to the nonadapted isolate. Transport of antimicrobial toxins to sites of attempted pathogen penetration has been shown to contribute towards penetration resistance against nonadapted pathogens in Arabidopsis (Stein et al., 2006). More recently a putative ABC transporter was identified that confers broad-spectrum resistance to a number of pathogens in wheat (Krattinger et al., 2009). The differential transport of distinct compounds within the plant cell may therefore play a key role in resistance to nonadapted pathogens.
Twenty-one per cent of the transcripts upregulated in common between all three Magnaporthe isolates were also upregulated in wheat in response to inoculation with the fungal pathogen Puccinia striiformis f. sp. tritici, the causal agent of yellow rust (Coram et al., 2008a,b). However, only 7% of the transcripts were specifically upregulated following inoculation with the nonadapted isolate BR29 and 4% were upregulated in response to the adapted isolates, were induced in wheat by P. s. f. sp. tritici. The transcripts induced in common by Magnaporthe and P. s. f. sp. tritici encoded for a number of PR proteins, WIR genes, PAL and related enzymes, as well as stress-related alternative oxidases and calmodulin-binding proteins. These transcripts appear to represent general defence responses in wheat against pathogens. The small number of BR29 and BR32/37-specifc transcripts also induced in wheat in response to P. s. f. sp. tritici inoculation included anthocyanin UDP-glycosyltransferases, a MAP kinase homologue and a putative WRKY5 transcription factor. However, the overall low numbers of shared transcripts would suggest that at the level of gene transcription the wheat–Magnaporthe interactions are distinct from the wheat–P. f. sp. tritici interactions and that different defence pathways operate against wheat pathogens with differing modes of infection (Glazebrook, 2005).
By far the majority of transcripts differentially expressed in response to inoculation with both the wheat-adapted and nonadapted Magnaporthe isolates were annotated as function unknown. The genome of rice has been fully sequenced, yet over 60% of the predicted genes are still of unknown function (Ouyang et al., 2007; http://rice.plantbiology.msu.edu/). Many of these unknown transcripts may prove to have important roles in determining the response of wheat to adapted and nonadapted pathogens. The goal now is to determine which of the many transcripts differentially expressed in this study play a significant role in resistance towards Magnaporthe in wheat and what role suppression of these genes plays in breaking otherwise durable, broad-spectrum resistance.
We thank Didier Tharreau for providing the Magnaporthe isolates, Nina Zellerhoff for the Digitaria sanguinalis seed and Ane Sesma for advice relating to Magnaporthe. We also thank Ruth MacCormack, Sarah Tucker and Mélodie Bousquet for technical assistance and Chris Ridout for critical review of the manuscript. This work was funded by a CGIAR, Generation Challenge Project – Cereal Immunity. MIAME/Plant-compliant microarray data from this study have been deposited in PLEXdb (http://www.plexdb.org) with the accession number TA24.