Triticale is the intergeneric hybrid between wheat and rye. With the expansion of the triticale growing area, powdery mildew has emerged and become a significant disease on this new host. Recent research demonstrated that this ‘new’ powdery mildew on triticale has emerged through a host range expansion of powdery mildew of wheat. Moreover, isolates sampled from triticale still infect their previous host, wheat, but isolates sampled from wheat hardly infect triticale. Race-specific and adult-plant resistance have been identified in triticale cultivars. The main objective of this study was to characterize the cellular basis of powdery mildew resistance in triticale. Commonalities with resistance responses in other cereals such as wheat, barley and oat are discussed. A detailed comparative histological study of various resistance responses during cross-inoculation of either virulent or avirulent wheat and triticale isolates on both hosts was carried out. The present data provide evidence that for incompatible interactions, the formation of non-penetrated papillae is the predominant resistance response, while the hypersensitive response (HR) acts as a second line of defence, to cut the fungus off from nutrients, if penetration resistance fails. It is not clear yet what causes the slower growth and reduced colony size of triticale isolates when inoculated on wheat. Possibly, post-penetration resistance mechanisms, other than HR, are switched on during these (semi-) compatible interactions. Molecular studies on gene expression and gene function of defence-related genes might reveal further insights into the genetic basis of these resistance responses.
Triticale (× Triticosecale) is the intergeneric hybrid between the female parent wheat (Triticum spp.) and the male parent rye (Secale spp.). This artificial cereal combines the cold and disease tolerance of rye and its adaptation to unfavourable soils and climates with the productivity and nutritional qualities of wheat (Walker et al., 2011). However, with the expansion of the triticale growing area in Europe (FAOSTAT), its disease resistance has changed dramatically (Oettler, 2005). During the last decade powdery mildew has emerged on triticale and become a significant disease on this new host. This was simultaneously observed in several European countries (Walker et al., 2011; Troch et al., 2012).
Powdery mildew, caused by Blumeria graminis, is a major problem in cereal production, as it can reduce quality and yield (Everts et al., 2001; Conner et al., 2003). Blumeria graminis is an obligate biotrophic fungus, which implies that it depends on living plant cells for survival and reproduction. In this way, B. graminis has evolved eight distinct formae speciales (ff. sp.) that are adapted to specific hosts (Marchal, 1902; Oku et al., 1985). Recent combined pathology and genetic research demonstrated that this ‘new’ powdery mildew on triticale has emerged through a host range expansion of the wheat powdery mildew B. graminis f. sp. tritici (Walker et al., 2011; Troch et al., 2012), which has acquired the capacity to colonize a new host species, triticale, in addition to the host of origin, wheat. Multilocus phylogeographical analysis revealed that this expansion occurred recently and multiple times at different locations in Europe (Troch et al., 2012). Race-specific (R-gene mediated) and adult-plant resistance have been identified in triticale cultivars (Troch et al., 2013), but nothing is known about the cellular basis of this resistance.
The powdery mildew pathogenesis and its cell biological aspects have been reviewed comprehensively (Eichmann & Hückelhoven, 2008; Hückelhoven & Panstruga, 2011). Penetration of the plant epidermal cell walls and subsequent haustorium formation is a critical stage in the infection process. Attacked plants may trigger different host cell defence responses that can act before, during or after cell penetration to arrest fungal development. A molecular framework for the evolution of plant immunity has been proposed by Jones & Dangl (2006) and is now a widely accepted hypothesis. Initially, recognition of conserved pathogen-associated molecular patterns (PAMPs) leads to PAMP-triggered immunity (PTI) (Jones & Dangl, 2006; Dodds & Rathjen, 2010). This early race-nonspecific basal pre-invasion resistance is exhibited in susceptible as well as R-gene resistant cultivars and is associated with cell wall modifications, including the formation of papillae (cell wall appositions) (Zeyen et al., 2002a). Papilla formation involves many cellular and biochemical processes, including the early accumulation of nitric oxide, hydrogen peroxide, phenolics and localized deposition of callose below the attempted site of penetration. As discussed by Zeyen et al. (2002a), papillae may effectively stop pathogen penetration, or alternatively, if penetration succeeds, the papilla becomes a ‘collar’ for the neck region of the haustorium. The second layer of defence is activated when the host recognizes specific pathogen effector proteins, which are produced to suppress PTI and facilitate infection, leading to effector-triggered immunity (ETI) (Jones & Dangl, 2006; Dodds & Rathjen, 2010). This race-specific resistance is based on the interaction between host resistance gene (R) products and pathogen avirulence gene (AVR) products in a classical gene-for-gene relationship (Flor, 1971). In contrast to race-nonspecific resistance, race-specific resistance usually permits fungal penetration into the host cell, but restricts further spread of the fungus by triggering a hypersensitive response (Hückelhoven & Kogel, 2003). On the other hand, if the plant does not recognize this pathogen effector, penetrated host epidermal cells survive and a biotrophic interaction is established. PTI is generally effective against non-adapted pathogens in a phenomenon called non-host resistance, whereas ETI is active against adapted pathogens. However, these relationships are not exclusive and depend on the effector molecules present in each infection (Dodds & Rathjen, 2010).
The main objective of this study was to characterize the cellular basis of powdery mildew resistance in triticale. Previous research showed that isolates sampled from triticale still infect their previous host, wheat, but isolates sampled from wheat hardly infect triticale (Troch et al., 2012, 2013). A first experiment was set up to clarify the cellular basis of these different responses. Two appropriate and highly compatible interactions of a wheat (Bgt) and triticale (BgTR) isolate on their respective hosts were compared with inappropriate inoculations of Bgt on triticale and BgTR on wheat, resulting in an incompatible and semicompatible infection type, respectively. In addition, a second experiment was set up to examine possible isolate-specific effects, by cross-inoculation of a virulent and an avirulent isolate from wheat and triticale on both hosts. In particular, fungal development, the role of papillae, accumulation of phenolic compounds and HR were monitored over a 72 h time course following inoculation. Altogether, the results provide a comprehensive comparative view of cellular defence responses in triticale associated with PTI and ETI.
Materials and methods
Plants, pathogens and inoculation
Two wheat (Triticum aestivum) and two triticale (× Triticosecale) cultivars were used: wheat cultivars Holger and Amigo, carrying resistance genes Pm6 and Pm17 respectively; and triticale cultivars Lamberto (DANKO, Poland) and Cultivo (SW Seed, Sweden), both carrying Pm17 and other unknown powdery mildew resistance genes (Troch et al., 2013). Plants were grown in a growth chamber at 18°C and a photoperiod of 16 h. Two B. graminis isolates sampled from wheat (Bgta and Bgtb) and two isolates sampled from triticale (BgTRa and BgTRb) were used for the inoculations. Isolates were made from single colonies and cultured on leaves detached from 2-week old seedlings on water agar (5 g L−1) amended with benzimidazole (40 mg L−1) in Petri dishes (Troch et al., 2012, 2013). Cultures were maintained at 18°C with a 16 h photoperiod and transferred to new leaves every 14 days. For the maintenance of the isolates collected from wheat and triticale, the susceptible host cultivars Cerco and Lamberto were used, respectively.
Inoculation was performed in a settling tower measuring 300 mm high and 103 mm diameter by uniformly dispersing conidia (Troch et al., 2012, 2013). Inoculation was carried out on 2-week old seedlings of the above-mentioned plants, including a susceptible control. Inoculum density was c. 20 conidia mm−2. Cultures were maintained on the same medium and under the same conditions as mentioned above. Infection types (IT) were scored 12 days post-inoculation (dpi) using a modified 0–4 scale (Wang et al., 2005), where 0 = no colonization; 1 = minute colonies with few conidia produced; 2 = colonies with moderately developed hyphae, but few conidia; 3 = colonies with well-developed hyphae and abundant conidia, but colonies not joined together; and 4 = colonies with well-developed hyphae and abundant conidia, and colonies mostly joined together. The term ‘compatible’ interaction is used for scores 3–4, ‘semicompatible’ for score 2 and ‘incompatible’ for scores 0–1. Additionally the term ‘appropriate interactions’ is used for inoculations of isolates on their respective hosts and ‘inappropriate interactions’ for inoculations of wheat isolates on triticale and vice versa.
Visualization of defence responses
For histological studies, cultures were maintained in the growth chamber until fixation at different time points. Leaf segments of 3 cm length were collected from the middle part of the inoculated leaves and the mid-vein was removed. Leaf segments were transferred to a solution of alcoholic lactophenol until they were completely cleared of chlorophyll, according to Adam & Somerville (1996).
Intracellular hyphae were visualized using a modified KOH–aniline blue technique (Hood & Shew, 1996). Cleared leaf segments were stained in a solution of 0·05% aniline blue dye in 0·067 m K2HPO4 at pH 9·0 for 5 min, followed by three rinses in demineralized water and examined under bright-field microscopy and UV light excitation (Olympus U_MWU2 filter set; excitation 330–385 nm; DM 400 dichroic beam splitter and BA420 long-pass filter).
For analysis of callose deposition, cleared leaves were stained for 5 min in a solution containing 0·01% (w/v) aniline blue and 0·15 m K2HPO4, followed by three rinses in demineralized water (De Vleesschauwer et al., 2010). Callose-stained segments were visualized using epifluorescence microscopy with UV filter (Olympus U-MWU2 filter set; excitation, 330–385 nm; DM 400 dichroic beam splitter and BA420 long-pass filter).
Phenolic compounds were analysed by staining cleared leaf segments in a solution of 0·05% toluidine blue in 50 mm citrate buffer pH 3·5 (Mellersh et al., 2002), followed by three rinses in demineralized water and visualized as autofluorescence under blue light epifluorescence (Olympus U-MWB2 GPF filter set; excitation, 450–480 nm; dichroic beam splitter, 500 nm; barrier filter BA515).
Detection of H2O2 by 3,3-diaminobenzidine (DAB) staining was performed as described previously (Thordal-Christensen et al., 1997) with minor modifications (De Vleesschauwer et al., 2010). Six hours before each time point, leaf segments were vacuum infiltrated with an aqueous solution of DAB-HCl (1 mg mL−1, pH 3·8) for 30 min. Infiltrated leaf segments were then further incubated at room temperature in the above-mentioned solution until sampling. DAB polymerizes in the presence of H2O2 and endogenous peroxidase to form a brownish-red precipitate that can be visualized using bright-field microscopy.
After staining, leaf segments were mounted on glass slides in 70% glycerol. Images were acquired digitally (Olympus Color View II camera) and further processed with the Olympus analySIS cell^F software.
Experimental design and data analysis
Two separate experiments were carried out. All interactions of one experiment were inoculated at the same time and harvested at the times indicated. Three leaf segments originating from different Petri dishes were used for each sampling point. The plant cell responses were assessed for 50 germinated conidia evenly distributed on each of the three fixed leaves. Epidermal cells challenged by more than one conidium were not considered. In experiment 1, samples were taken at 24, 48 and 72 h post-inoculation (hpi), whereas in experiment 2, samples were taken at the same time points and an additional time point of 8 days post-inoculation (dpi) to observe possible differences in pathogen growth between interactions later in the infection process. Interaction sites were scored for fungal developmental stages, papilla formation, accumulation of phenolic compounds, H2O2 accumulation in response to primary and appressorial germ tubes (PGT and AGT) and hypersensitive response (HR). Note was taken of whether papillae had effectively stopped fungal development or if papillae became haustorial neck collars. Percentage data were calculated for each leaf replicate. For statistical analysis, percentages were transformed to arcsine square roots (transformed value = 180/π × arcsin [√(%/100)]) to normalize data and stabilize the variance throughout the data range. sas software (SAS Inc.) was used to compute analyses of variances (anova) followed by Tukey's multiple range comparison tests among interactions; differences at P <0·05 were considered statistically significant.
Blumeria graminis infection phenotypes
Blumeria graminis infection phenotypes 12 dpi of highly compatible, semicompatible and incompatible interactions are presented in Figure 1b and d for experiments 1 and 2, respectively. Powdery mildew colonies with well-developed hyphae and abundant conidia were observed on compatible interactions (IT 3–4), although colonies were not always joined together (IT 3). Colonies with moderately developed hyphae and only few conidia were observed on the semicompatible interaction (IT 2). No visible symptoms were observed on the incompatible interactions (IT 0).
Blumeria graminis development was monitored over a 72-h time course in both experiments 1 and 2, taking samples at 12, 24, 48 and 72 hpi. The most significant differences between interactions were found at 72 hpi and therefore only data from this time point are presented (Fig. 1a,c). Data at other time points can be found in Figures S1 and S2 for experiment 1 and 2, respectively.
Conidial germination with a short primary germ tube (PGT) and an appressorial germ tube (AGT) was observed at 12 hpi for all interactions. For some interaction sites (between 16 and 44%) the AGT already formed a swollen, hooked apical appressorium. Small but significant differences were found between interactions at this time point (only for experiment 2), indicating that both wheat and triticale isolates germinated at similar rates on any wheat or triticale cultivar, irrespective of whether the interaction was compatible or incompatible. Subsequently at 24 hpi, appressorial lobes were predominantly observed in all interactions of both experiments, and again no significant differences were found between compatible, semicompatible or incompatible interactions. Moreover, for some interactions, the development of elongating secondary hyphae (ESH) was initiated (<4% of interaction sites). In experiment 1 at 48 hpi, significantly higher percentages of interaction sites forming ESH were observed for the compatible and semicompatible interactions compared to the incompatible interaction. Generally in experiment 2, a higher percentage of interaction sites succeeded in the formation of ESH in compatible interactions than in incompatible interactions. However, the incompatible and inappropriate interaction of BgTRa on wheat cv. Amigo did result in a percentage of interaction sites forming ESH similar to the percentage in compatible interactions.
Highly significant differences were found for fungal development stages at 72 hpi in experiment 1 (Fig. 1a). In particular, the percentage of interaction sites showing ESH was significantly higher on both the compatible and semicompatible interactions, compared to the incompatible interaction. Interestingly, no significant difference in fungal development was found between the semicompatible and compatible interactions. Also, in experiment 2 at 72 hpi, the highest percentage of interaction sites forming elongating secondary hyphae was observed for the compatible interactions, although this was not significantly higher than all incompatible interactions (Fig. 1c). In particular, the compatible interaction of the triticale isolate BgTRb on wheat cv. Amigo (IT 3) resulted in a similar percentage of interaction sites forming elongating secondary hyphae as all incompatible interactions. Surprisingly, at 8 dpi (Fig. S2) only the compatible interactions on wheat cv. Amigo (Bgtb–Amigo and BgTRb–Amigo) had proceeded to the formation of conidiophores (51 and 32% of interaction sites, respectively), while this was not (yet) observed for the compatible interaction on triticale cv. Cultivo (BgTRb–Cultivo).
Callose deposition in papillae
In both experiments, callose deposition was frequently observed in papillae after pathogen attack from 24 hpi (Tables 1 & 2). Papillae either prevented penetration (designated as non-penetrated papilla) or became a haustorial neck collar after successful penetration (Fig. 2).
Table 1. Papilla formation during different Blumeria graminis interactions in experiment 1
Interaction sites (%)
Bgta Holger IT 4
BgTRa Holger IT 2
Bgta Lamberto IT 0
BgTRa Lamberto IT 4
Data represent the mean of 50 interaction sites on each of three replicate leaves. For each time point, different letters indicate statistically significant differences according to Tukey (P <0·05). hpi, hours post-inoculation; IT, infection type; AGT, appressorial germ tube.
AGT + papilla
Table 2. Papilla formation during different Blumeria graminis interactions in experiment 2
Interaction sites (%)
Bgta Amigo IT 0
Bgtb Amigo IT 4
BgTRa Amigo IT 0
BgTRb Amigo IT 3
Bgta Cultivo IT 0
Bgtb Cultivo IT 0
BgTRa Cultivo IT 0
BgTRb Cultivo IT 3
Data represent the mean of 50 interaction sites on each of three replicate leaves. For each time point, different letters indicate statistically significant differences according to Tukey (P <0·05). hpi, hours post-inoculation; IT, infection type; AGT, appressorial germ tube.
AGT + papilla
In experiment 1, the majority of interaction sites were associated with papilla formation at 72 hpi (81–92%) and no significant differences were found between compatible and incompatible interactions (Table 1). Nevertheless, the fungal penetration peg overcame a significantly higher percentage of papillae in the compatible and semicompatible interactions than in the incompatible interaction, as indicated by the percentage of non-penetrated papillae.
In experiment 2 at 72 hpi, a significantly higher percentage of interaction sites associated with papilla formation was found on wheat cv. Amigo after infection with wheat isolates (Bgt) compared to infection with triticale isolates (BgTR), irrespective of the compatibility of the interaction (Table 2). On triticale cv. Cultivo, the compatible interaction (BgTRb–Cultivo) resulted in a lower percentage of interaction sites associated with papilla formation compared to the incompatible interactions. Similarly, as in experiment 1, the number of non-penetrated papillae was significantly lower in the compatible interactions than in the incompatible interactions. However, even for the compatible interactions more than 50% of papillae were not penetrated.
Accumulation of phenolic compounds
The accumulation of phenolic compounds, as an important plant defence mechanism to prevent fungal penetration, was demonstrated by the autofluorescence of attacked cells after toluidine blue staining at different time points (24, 48 and 72 hpi). Phenolic compounds accumulated in epidermal cells in direct contact with infection structures and occasionally also in epidermal cells adjacent to the attacked epidermal cells (Fig. 3a). Statistical analysis of the accumulation of phenolic compounds can be found in Tables S1 and S2 for experiment 1 and 2, respectively.
In experiment 1, autofluorescence was detectable as early as 24 hpi, irrespective of the compatibility of the interaction (Fig. 3b). However, by 48 and 72 hpi, significantly more phenolic compounds accumulated in the incompatible (Bgta–Lamberto) and semicompatible (BgTRa–Holger) interactions (reaching a level of autofluorescence at 81 and 74% of interaction sites at 72 hpi, respectively) compared to the incompatible interactions Bgta–Holger and BgTRa–Lamberto (reaching a maximum of autofluorescence at 49 and 40% of interaction sites at 72 hpi, respectively).
In experiment 2, autofluorescence was also detectable at 24 hpi for all interactions, with as many as 37% of interaction sites accumulating phenolic compounds for the compatible interaction Bgtb–Amigo at this time point (Fig. 3c). By 48 hpi, the accumulation of phenolic compounds increased for all interactions, although generally the percentage of interaction sites showing autofluorescence was lower for the interactions on triticale cv. Cultivo compared to those on wheat cv. Amigo. At 72 hpi, only two compatible interactions (BgTRb–Cultivo and BgTRb–Amigo) resulted in a significantly lower percentage of interaction sites that autofluoresced (43 and 38%, respectively) than all other interactions, for which the percentage of interaction sites that autofluoresced reached between 71 and 93%. Strikingly, the highly compatible interaction Bgtb–Amigo also showed this high percentage of interaction sites accumulating phenolic compounds.
H2O2 localization at interaction sites
Histochemical localization of H2O2 using DAB in epidermal cells attacked by B. graminis showed an accumulation of this active oxygen species in cell wall appositions as well as in cells undergoing HR (Fig. 4). Typically, many DAB-stained particles were observed near the site of attack of the epidermal cell, with slight development of a haustorium inside the cell, which resembles a slow HR (Fig. 4c).
To determine the temporal profile of H2O2 accumulation in defence reactions, the frequency of interaction sites showing accumulation of H2O2 at different time points was analysed (Fig. 5). In experiment 1, no significant differences between interactions were found at 24 and 48 hpi for the accumulation of H2O2 (Fig. 5a). Moreover, only a few interaction sites showed H2O2 accumulation both in cell wall appositions and in cells undergoing HR at these time points (<35% of all interaction sites). By 72 hpi, the number of interaction sites showing no accumulation of H2O2 at all was still prevalent for all interactions. Moreover, compatible interactions (Bgta–Holger and BgTRa–Lamberto) showed a significantly higher percentage of interaction sites without H2O2 accumulation and consequently a significantly lower percentage of interaction sites with H2O2 accumulation, both in cell wall appositions and in cells undergoing HR, compared to the incompatible (Bgta–Lamberto) and semicompatible (BgTRa–Holger) interactions.
In experiment 2 at 24 hpi, the compatible interaction Bgtb–Amigo showed a significantly lower percentage of interaction sites without H2O2 accumulation and consequently a significantly higher percentage of interaction sites showing H2O2 accumulation in cell wall appositions than all other interactions (Fig. 5b). However, only at 48 hpi were considerable differences found for the accumulation of H2O2 between different interactions. In particular, differences were found between compatible and incompatible interactions on wheat cv. Amigo at this time point. Compatible interactions (Bgtb–Amigo and BgTRb–Amigo) showed a higher percentage of interaction sites without any accumulation of H2O2 and a lower percentage of interaction sites with H2O2 accumulation both in cell wall appositions as well as in cells undergoing HR, compared to incompatible interactions (Bgta–Amigo and BgTRa–Amigo). On the other hand, no significant differences were found between the incompatible and compatible interactions on triticale cv. Cultivo at 48 hpi. By 72 hpi, the percentage of interaction sites undergoing HR increased for the incompatible interactions on triticale cv. Cultivo, while it decreased on wheat cv. Amigo. In particular, the incompatible interactions of wheat isolates (Bgta and Bgtb) on triticale cv. Cultivo showed a significantly higher percentage of cells undergoing HR after attack than all other interactions.
The data presented provide evidence of multiple defence pathways involved in triticale resistance to the recently emerged powdery mildew. By determining the timing of various resistance responses, some clues are provided as to which of these responses may have effective defence functions.
In a first experiment, two appropriate and highly compatible interactions of a wheat and triticale isolate on their respective hosts were compared with inappropriate inoculations of Bgt on triticale and BgTR on wheat, resulting in an incompatible and semicompatible infection type, respectively. The main resistance responses induced in these interactions are summarized in Table S3. As expected, papilla formation was frequently observed in both compatible and incompatible interactions. This early race-nonspecific basal pre-invasion resistance is exhibited in susceptible as well as R-gene resistant cultivars (Zeyen et al., 2002a). Notwithstanding, significantly more non-penetrated papillae were observed in the incompatible interaction compared to the (semi-) compatible interactions. Li et al. (2005) also found that resistant wheat lines had higher frequencies of non-penetrated papillae than susceptible lines after inoculation with B. graminis f. sp. tritici. Additionally, the presence of H2O2 in papillae is a biochemical marker that distinguishes non-penetrated from penetrated cells in the interaction of barley and wheat with powdery mildew (Hückelhoven et al., 1999; Li et al., 2005; Trujillo et al., 2004), and enzymatic removal of H2O2 supports fungal penetration (Mellersh et al., 2002). However, in this study, local H2O2 accumulation in papillae was observed both in incompatible and compatible interactions at different time points, and no unambiguous link could be found with the percentage of non-penetrated papillae. As it appears that speed of deposition and composition of papillae is pivotal for successful defence (Assaad et al., 2004), it may be that the crucial local H2O2 accumulation, which distinguishes non-penetrated from penetrated papillae, takes place before the earliest time point of 24 hpi measured in the present study. Accumulation of phenolic compounds is also associated with penetration resistance (Zeyen et al., 2002a). Significantly more accumulation was observed from 48 hpi in both the incompatible and semicompatible interaction compared to the compatible interactions. When penetration resistance fails and haustoria develop within host cells, another defence mechanism, i.e. the hypersensitive response (HR), can be triggered (Hückelhoven & Kogel, 2003). However, irrespective of the compatibility of the interaction, almost no HR was observed in experiment 1, probably because no appropriate and incompatible interaction was studied in this experiment. Altogether, these findings might indicate that resistance of triticale to inappropriate wheat powdery mildew is mainly associated with penetration failure due to papilla deposition. This corresponds with previous studies of resistance to inappropriate B. graminis ff. sp. (Hückelhoven et al., 2001; Zeyen et al., 2002b; Olesen et al., 2003; Trujillo et al., 2004). On the other hand, for inappropriate inoculations of triticale isolates on wheat some post-penetration resistance mechanisms cause colonies to be reduced in area or slower in growth, but are not strong enough to abort the colony. Slow growth and development rate of fungal colonies in quantitatively resistant host plants has been associated with reduced quality, size or efficiency of the haustorium (Sillero et al., 2001). However, based on observations in the present study, the possibility cannot be ruled out that the reduced area of colonies observed in the semicompatible interaction is caused by the accumulation of phenolics, although this is known to act as a pre-penetration resistance mechanism (Zeyen et al., 2002a). Recently, a sow thistle powdery mildew isolate (Golovinomyces cichoracearum) that can overcome penetration resistance but is invariably stopped by post-invasion non-host resistance of Arabidopsis thaliana was identified (Wen et al., 2011). This post-invasion non-host resistance was manifested as the formation of a callosic encasement of the haustorial complex. Previously, Meyer et al. (2009) found that essentially all B. graminis f. sp. hordei haustoria on the non-host Arabidopsis were partially encased at 24 hpi, while this was never observed on barley leaves, even at 72 hpi. It is likely that this post-penetration resistance mechanism is switched on in wheat attacked by triticale powdery mildew.
In a second experiment a virulent and avirulent isolate of wheat and triticale were cross inoculated on both hosts. The main resistance responses induced in these interactions are summarized in Table S4. Even in incompatible interactions, the pathogen occasionally developed ESH (which is an indicator of successful host cell penetration), although no macroscopically visible symptoms were observed. This was also found by Trujillo et al. (2004) after inoculation of barley genotypes with non-adapted wheat powdery mildew. Although the compatible interaction of the triticale isolate on wheat (BgTRb–Amigo) resulted in a similar level of ESH as all incompatible interactions, this interaction proceeded to the formation of conidiophores at 8 dpi, whereas this was not the case for the incompatible interactions. Therefore, it seems that the slower growth observed following the inappropriate inoculation of triticale isolates on wheat, may be a result of post-penetration resistance mechanisms, possibly callosic encasement of the haustorial complex, as suggested for experiment 1. The development of conidiophores in the compatible interaction on triticale cv. Cultivo may occur later than on wheat, as no conidiophores were formed at 8 dpi, but at 12 dpi abundant conidia were observed macroscopically. More non-penetrated papillae were associated with incompatible interactions, as found in experiment 1, but the compatible interactions in experiment 2 resulted in 50–70% non-penetrated papillae while this was <50% in experiment 1. Additionally, in general more local H2O2 accumulation was observed in papillae for the interactions studied in experiment 2 compared to experiment 1, which is a characteristic of non-penetrated papillae. Sánchez-Martín et al. (2011) found that in adult plant resistance in oat, the main resistance component to powdery mildew was penetration resistance mediated by papillae. This is in agreement with the findings of this study for wheat cv. Amigo and triticale cv. Cultivo, which are both known to be highly resistant at the adult plant stage (Hsam & Zeller, 2002; Troch et al., 2013). In experiment 2, only the compatible interaction BgTRb–Cultivo resulted in a significantly lower percentage of interaction sites accumulating phenolics than all incompatible interactions, which is in contrast to the findings in experiment 1. This seemingly conflicting observation of high phenolic accumulation on the highly compatible Bgtb–Amigo interaction might indicate that phenolics do not provide effective pre-invasion defence against powdery mildew, probably because their accumulation is too late (starting after 24 hpi). Moreover, whether phenolics function by toxicity or by cell wall strengthening is poorly understood (Hückelhoven, 2007). They can be polymerized and cross-linked to form lignin or lignin-like polymers that strengthen cell walls and impede fungal penetration. Low molecular weight phenolics are known for their antimicrobial activity. The histological approach of this study did not allow identification of the individual phenolic components in the different interactions. Therefore, it is possible that phenolics accumulate differentially in the different interactions, as was found for broad-spectrum powdery mildew resistance controlled by mlo alleles in barley (von Röpenack et al., 1998). Altogether, these results suggest that non-penetrated papilla may play a substantial role in defence against powdery mildew in the incompatible interactions. However, even in incompatible interactions, a proportion of attacks succeed in successful penetration, as indicated by the percentage forming ESH, but these do not result in further growth and proliferation of the fungus. This points to additional layers of protection switched on during the establishment of an incompatible interaction. HR is typical for race-specific, R gene-mediated resistance triggered by the recognition of specific, pathogen-derived effector molecules, termed avirulence factors (Jones & Dangl, 2006), as exemplified by the Mla gene in barley (Koga et al., 1990). In this way, HR might be expected to be the main line of defence in the incompatible but appropriate interactions of Bgta–Amigo and BgTRa–Cultivo studied in experiment 2. However, in this study, irrespective of the compatibility of an interaction, some epidermal cells showed degrees of degeneration, suggesting cell death at different time points. In experiment 2, an earlier HR was observed in wheat cv. Amigo (mainly at 48 hpi) than in triticale cv. Cultivo (mainly at 72 hpi). This might be related to the faster fungal development in Amigo than in Cultivo (Figs S2 & 1b). Moreover, only slow HR was triggered, at late time points, visible as epidermal cells with many DAB-stained particles near the site of attack and a slight development of the haustorium inside the cell. This post-penetration resistance seems to cause the arrest of fungal development in incompatible interactions, where elongated secondary hyphae were observed in a proportion of interaction sites while no macroscopic symptoms were observed. In incompatible wheat (Li et al., 2005) and oat (Sánchez-Martín et al., 2011) powdery mildew, HR was also found to act as a second line of defence to arrest further fungal development when the papilla defence fails. Moreover, successful host cell penetration by non-adapted powdery mildew fungi is also usually followed by an HR (Eichmann & Hückelhoven, 2008). Additionally, in barley the resistance response of the Mlg locus, which governs powdery mildew resistance in a typical race-specific and semi-dominant manner, was studied (Görg et al., 1993). It was found that, unlike other race-specific barley mildew resistance genes (Mla), Mlg elicits papilla-based defence and that HR is a secondary consequence rather than causally necessary to arrest fungal growth.
In summary, the data in this investigation demonstrate that both for incompatible and appropriate as for incompatible and inappropriate interactions, the formation of non-penetrated papillae is the predominant resistance response, while HR acts as a second line of defence if penetration resistance fails. As papillae formation is typically observed in race-nonspecific resistance such as mlo resistance in barley or quantitative, ‘background’ resistance (Zeyen et al., 1993; Carver et al., 1994), this trait could offer opportunities for durable powdery mildew resistance breeding in triticale. It is not clear yet what causes the slower growth or reduced colony size of triticale isolates inoculated on wheat. Possibly a post-penetration resistance mechanism other than HR, such as callosic encasement of the haustorial complex, is switched on during these inappropriate and (semi)-compatible interactions. Moreover, even in compatible and appropriate interactions, a proportion of epidermal cells do not support fungal growth, because of non-penetrated papillae formation or HR. However, to determine possible genotypic effects of the host on the resistance responses, other triticale cultivars should be tested. Additionally, molecular studies on gene expression and gene function of defence-related genes might reveal further insights into the genetic basis of these resistance responses.
This project was funded by a grant from the University College Ghent (Ghent University Association). The authors gratefully acknowledge James Brown from the John Innes Centre (Norwich, UK) for his expertise and training in powdery mildew manipulations and would also like to thank David De Vleesschauwer from the Laboratory of Phytopathology (UGent, Belgium) for providing help with experimental procedures.