The accumulation of H2O2 (oxidative burst) and the progress of pathogen development were studied in compatible and incompatible wheat-brown rust interactions. The accumulation of H2O2 was detected in 98·7% of guard cells with appressoria 8 h post inoculation (hpi). The reaction in both susceptible and resistant plants declined 2–3 days post inoculation (dpi). The second phase of the oxidative burst was observed in the mesophyll and/or epidermis. In susceptible plants it began 4–5 dpi and was detected only in the epidermis. In resistant plants the response was observed in the mesophyll. In moderately resistant plants it was induced 1–3 dpi, and the percentage of infection units reached 80–90% 8 dpi. This corresponded with severe necrotic symptoms. In highly resistant plants, the oxidative burst was short and transient. The percentage of infection units with H2O2 accumulation reached its highest level (60–70%) 2 dpi, and decreased thereafter. Four days later, the low percentage and weak DAB staining indicated very low H2O2 accumulation. The localization and the time-course changes of the oxidative burst correlated with the profiles of the micronecrotic response, haustorium mother cell formation and pathogen development termination. An early and localized induction of oxidative burst followed by its rapid quenching correlated with high resistance and almost no disease symptoms. The possible correlation of the oxidative burst and pathogen development patterns with the level and durability of resistance conferred by Lr genes are discussed.
Brown or leaf rust caused by Puccinia triticina is one of the most devastating foliar diseases of wheat (Triticum aestivum), and can be responsible for large yield losses every year (Marasas et al., 2003; Park, 2008). Close to 60 leaf rust resistance genes (Lr) conferring various levels of resistance against Puccinia triticina in seedlings and/or adult plants have been identified. A significant number have been bred into cultivars released since the 1960s. Despite the high number of identified Lr genes and their economic importance, understanding of the mechanisms and processes involved in leaf rust resistance is still very limited. This knowledge would be useful for the selection of a particular resistance gene for breeding and for resistance gene pyramiding.
Based on histopathological analyses, two types of resistance have been identified, termed pre- and post-haustorial. The former is characterized by a high percentage of early aborted infection units and is often observed in non-host interactions (Anker & Niks, 2001). Post-haustorial resistance is typical for race-specific plant-pathogen interactions. It is characterized by a hypersensitive response induced in the host cells shortly after the formation of haustoria (Southerton & Deverall, 1989). This type of resistance is thought to be less durable than pre-haustorial resistance (Niks & Rubiales, 2002). Although the role of the hypersensitive response in plant resistance against fungal pathogens is well established, this single response is not necessarily decisive in inhibiting pathogen development. The hypersensitive response can be activated in resistant, partially resistant and in susceptible plants (Niks & Dekens, 1991). Also, it does not simply correlate with resistance durability (Bender et al., 2000; Niks & Rubiales, 2002).
The purpose of this study was to analyse the early stages of host-pathogen interaction in wheat cultivars expressing race-specific resistance conferred by selected Lr genes. The objective was to check whether certain types of race-specific resistance could activate the host response to affect pathogen growth at very early stages in a manner similar to pre-haustorial resistance. The possible correlation of the oxidative burst and pathogen development patterns with the level and durability of resistance conferred by Lr genes are discussed.
Materials and methods
Plant and pathogen material
The wheat cultivar Thatcher, which is susceptible to brown rust (leaf rust), and lines isogenic to Thatcher carrying single leaf rust resistance genes (Lr): TcLr9, TcLr19, TcLr24, TcLr25, TcLr26, TcLr29 and TcLr34 (McIntosh et al., 1995) were used. Seeds were kindly supplied by Dr M. Csosz, Cereal Research Institute, Szeged, Hungary. The wheat seedlings were grown in a growth chamber at 22°C with a 16-h photoperiod and an illumination intensity of 60 μE m−2 s−1. A single spore isolate of P. triticina with an established avirulence/virulence formula was used. The primary leaves of 7-day-old seedlings were inoculated with urediniospores suspended in water with Tween 20 at a density of 1 mg mL−1. Inoculation with a soft brush gave a final spore density on the leaf surface of 40–150 spores cm−2.
Plants were incubated for 24 h in a dark growth chamber at 18°C and 100% humidity, and then cultivated in the same conditions as seedlings prior to inoculation. The pathogen development and disease symptoms were observed and scored 10 days post inoculation (dpi). The infection type was determined according to the scale for seedlings established by Roelfs & Martens (1988).
Calcofluor white staining
Leaf samples were collected on each of the 8 days after inoculation, and were cleared and fixed in ethanol/dichloromethane (3:1 v/v) with 0·15% trichloroacetic acid for 24 h. The specimens were washed twice in 50% ethanol for 15 min, twice in 0·05 m sodium hydroxide for 15 min, three times in water, and once in 0·1 m Tris–HCl, pH 8·5, and then stained in 3·5 mg mL−1 calcofluor white dissolved in 0·1 m Tris–HCl, pH 9, for 5 min. They were then washed once in water, and stored in 25% glycerol with 0·1% lactophenol as per a modified version of the procedure in Bender et al. (2000). The stained samples were examined under a fluorescence microscope (Nikon Diaphot, epifluorescence optics with excitation 340–380 nm, barrier filter 420 nm and dichroic mirror 400 nm). These conditions allowed the visualization of germinating spores, appressoria, penetration pegs and haustorium mother cells (HMC) of the pathogen, and the autofluorescence of the host cell necrosis. Observations were made for at least 30 infection sites per leaf sample with three replications (plants).
Evans blue staining
Samples of inoculated leaves were infiltrated for 15 min under a vacuum with a 0·1% aqueous solution of Evans blue. Directly after infiltration, the leaf samples were examined under a light microscope (Wang et al., 2007). The dead cells accumulated the stain and were visible as dark blue spots/areas. The same leaf samples were observed under a fluorescent microscope to check whether the cells displaying yellow-red autofluorescence were also stained with Evans blue. The observations of calcofluor white autofluorescence and Evans blue staining allowed the co-localization of the plant’s necrotic reactions with the pathogen structures.
Hydrogen peroxide detection (DAB staining)
Leaf samples were collected 8, 13 and 18 hpi and on each of the 8 days after inoculation, and were placed in 1 mg mL−1 DAB aqueous solution, pH 3·8, to be incubated in the dark for 4 h, according to a modified version of the procedure in Tada et al. (2004) and Wang et al. (2007). The leaves were de-stained overnight in ethanol/chloroform (4:1 v/v) containing 0·15% trichloroacetic acid as per a modified version of the procedure in Schweizer (2008). The hydrogen peroxide accumulated in the leaf tissue converts DAB into a brown insoluble precipitate which is visible under a light microscope, and its presence allows the detection and relative quantification of hydrogen peroxide. Observations were made for at least 30 infection sites per leaf sample with three replications (plants).
The reference for the quantitative evaluation of the plant-pathogen interactions was the number of infection units (iu). For early interactions (8, 13, 18 hpi), iu was defined as the mean number of appressoria formed. For all other interactions (1–8 dpi), it was the mean number of appressoria with HMC and/or necrosis.
Characterization of the tested lines
Disease symptoms were scored 10 dpi (Table 1, Fig. 1) allowing the separation of the tested lines into three distinct groups. Group I contained the susceptible plants (infection type 4 with large uredinia without chlorosis or necrosis). Group II contained the intermediate resistant plants (infection type 1 or 2 with few small to medium-sized uredinia surrounded by excessive chlorosis and necrosis, and showing restricted pathogen development). Group III contained resistant lines (infection type 0 with few and very weakly visible disease symptoms).
Table 1. Infection types of selected isogenic wheat lines inoculated with urediniospores of Puccinia triticina (brown rust) isolate. 0 – no uredinia or other macroscopic signs of infection, 0; – no uredinia, but hypersensitive necrotic or chlorotic flecks of varying size present, 1 – small uredinia often surrounded by necrosis, 2 – small to medium-sized uredinia often surrounded by chlorosis or necrosis, 3 – medium-sized uredinia that may be associated with chlorosis or rarely necrosis, 4 – large uredinia without chlorosis or necrosis (Roelfs & Martens, 1988)
Profiles of compatible and incompatible wheat-rust interactions
Calcofluor white staining allowed visualization of the pathogen structures. Spore germination started 4 hpi. After 8 h, 70% had formed appressoria, and over 80% had formed appressoria at 1 dpi, with similar results in all the tested lines. The final density of infection units observed in the stained leaf segments varied between 20–100 iu cm−2. This indicated that the first steps of pathogen growth (spore germination and appressoria formation on the stomata) were not affected by the presence or type of Lr gene (data not included). After the formation of appressoria, all the subsequent steps of P. triticina development and the corresponding wheat response were described by five distinct profiles of plant-pathogen interactions: (i) the presence of HMC without micronecrotic reactions; (ii) the simultaneous presence of HMC and micronecrosis; (iii) the presence of micronecrotic reactions without HMC; (iv) the formation of uredinia with an associated necrotic response; and (v) the formation of uredinia without any necrotic reaction.
In cv. Thatcher and TcLr34, the two lines representing group I, the interactions were of the profile type i (HMC without micronecrosis), and reached 100% very rapidly. This high level was maintained until 6 dpi, when the onset of uredinia formation indicated the completion of the pathogen’s life cycle. This last step of rust development also did not induce micronecrotic reactions (Figs 2d & 3a,b). This pattern of host-pathogen interaction indicated unrestricted pathogen growth, and occurred only in the case of the compatible interaction of P. triticina with the plants from group I. The incompatible interactions observed in the plants from groups II and III were always associated with the micronecrotic response (group II, Fig. 2f–h, and group III, Fig. 2j–l), with the intensity and time of initiation differing between the two.
In group II, the highest percentage of infection units with both HMC and necrosis (84·8–92·8%) was found 3–5 dpi (profile ii, Fig. 3c–e). The percentage of iu with necrosis but without HMC (profile iii) was very low. The pattern of profile ii (Fig. 3c–e) indicated that the plants’ micronecrotic response was not sufficient to arrest the formation of HMC and stop the pathogen’s growth. In group III, almost all the appressoria with HMC (95·5–99·6%) were associated with necrotic reactions 2 dpi (profile ii, Fig. 3f–h). Over the next 3 days, this profile steadily declined with a simultaneous increase of profile iii (the presence of micronecrosis but a lack of HMC; Fig. 3f–h). After 5 dpi, the majority of appressoria formed shortly after inoculation were not associated with HMC. There was a close association between the growing number of necrotic reactions and the decreasing number of HMC. This indicated a strong suppression of the pathogen’s growth. Macroscopic observations confirmed this as no uredinia were produced in these plants (Fig. 1). Concurrent changes in these profiles fell into three distinct patterns (Fig. 3a–h) that corresponded to the three groups established earlier based on disease symptoms (Fig. 1).
Accumulation of hydrogen peroxide during compatible and incompatible interactions
The accumulation of hydrogen peroxide, seen as a brown precipitate in samples stained with DAB, occurred in all samples, and was localized close to the infection site (Fig. 4). The response was induced shortly after inoculation and was initially detected in guard cells with appressoria (Fig. 4a,e,i) and 8 hpi, guard cell H2O2 accumulation was observed in 72·6–98·1% of infection units (i.e. stomata with appressoria; Fig. 5). The percentage of infection units responding with an oxidative burst reached a high level 8 hpi, then sharply declined 2 dpi and between 2–3 dpi became undetectable. The temporal pattern of this very rapid and transient response was similar for both susceptible and resistant plants (Fig. 5). The accumulation of H2O2 in the guard cells was followed by an oxidative burst in the mesophyll and/or epidermis (Figs 4e–h & 5c–e) or only in the mesophyll cells (Figs 4i–k & 5f–h). This response was always localized in close proximity to developing rust structures. The initiation time, localization and time-course changes of the oxidative burst constituted a pattern that was distinct for the three groups of lines (Figs 4 & 5). In susceptible plants (group I), the oxidative burst was not activated (Figs 4a–d & 5a,b) and the accumulation of H2O2 in the stomata was the only reaction. Accumulation of H2O2 in the epidermis was only detected by the end of the pathogen cycle (4–8 dpi), when it was observed around newly formed uredinia (Figs 4c,d & 5a,b). Incompatible wheat-rust interactions (group II and III) were always associated with the induction of an oxidative burst in the mesophyll and epidermis (Figs 4e–h & 5c–e) or in the mesophyll alone (Figs 4i–l & 5f–h). Although the response was induced shortly (1–2 dpi) after the infection in all the resistant plants, its pattern varied between groups II and III.
In group II (Fig. 5c–e), the accumulation of H2O2 in the mesophyll and epidermis was induced 8–13 hpi and started to increase 1 dpi (TcLr24, TcLr29) and 3 dpi (TcLr25). The percentage of infection units showing this response steadily increased to reach a level of 81·8–91·5% by 8 dpi. Once induced in these lines, the accumulation of hydrogen peroxide was observed until the end of the pathogen’s life cycle (Figs 4e–h & 5c–e). Although the steady increase of H2O2 accumulation in group II coincided with the limited pathogen development (Fig. 2h), the rate of pathogen suppression was different between the lines. The percentage of infection units with uredinia and the necrotic reaction (profile iv) was low (34·7%) in TcLr24, intermediate (38%) in TcLr29 and high (79·9%) in TcLr25 8 dpi (Fig. 3c–e). This corresponded to the disease symptoms scored for these lines: necrosis and chlorosis with small uredinia in TcLr24 and TcLr29, compared with all the above symptoms and small to medium sized uredinia on TcLr25.
In group III, the oxidative burst was transient. The percentage of infection units surrounded with DAB-stained cells reached its maximum level (54·7–81·5%) 2 dpi and thereafter, it decreased (Fig. 5f–h). Four days later (6 dpi), the value was low (19·8%) in TcLr19 and even lower (14·0 and 9·1%, respectively) in TcLr9 and TcLr26 (Fig. 5f–h). Weak DAB staining (Fig. 4l) showed that the remaining level of H2O2 accumulation in all three lines was very low. All the infection units activated the oxidative response, but the percentage of infection units with oxidative burst never reached 100% (Fig. 5f–h). The difference (18·5–45·3%) was the result of the asynchronous induction of the reaction.
The time-course changes in the oxidative burst in group III coincided with an increasing rate of two processes: the formation of HMC and the development of micronecrosis (profile ii, Fig. 3f–h). The maximum rate (77·3–91·5% profile ii) was reached 2 dpi, and was similar for all lines. The later decrease in the presence of this profile was gradual in TcLr19 and TcLr9 and sharp in TcLr26. Simultaneous with the decrease in the occurrence of profile ii, there was an increase in that of profile iii, i.e. an increase in the percentage of infection units that no longer had any HMC but were still associated with micronecrosis. This profile indicated the complete arrest of pathogen growth (no HMC) with the remains of a micronecrotic response. This value was high (54·5%) for TcLr19 and very high (81·6 and 86·7%, respectively) for TcLr9 and TcLr26. The time-course changes in the oxidative burst and the profiles ii and iii (Figs 3f-h & 5f-h), indicated a tight correlation of the oxidative burst with the termination of the pathogen’s growth. The early and localized induction of the oxidative burst combined with its rapid quenching correlated with almost no disease symptoms developed on the leaves from group III. Here again, almost no disease symptoms were observed in TcLr9 and TcLr26 (Fig. 1).
One of the first cytologically detectable plant reactions to pathogen infection is the accumulation of H2O2 (Mellersh et al., 2002). In the wheat-P. triticina interaction, this response is first activated in the stomatal guard cells and later in the mesophyll and/or epidermis. The accumulation of H2O2 in the stomata was detected in both susceptible and resistant plants. Wang et al. (2007) described a similar reaction of wheat to infection with stripe rust (P. striiformis f.sp. tritici). It is conceivable that this response was induced by pressure generated by the appressoria, as the mechanical stimulus can induce defence responses including an oxidative burst (Gus-Mayer et al., 1998).
The second phase of H2O2 accumulation was detected in the mesophyll and/or epidermal cells adjacent to the pathogen structures. The pattern of this response, the induction time and its dynamics was genotype dependent. The initiation was different in susceptible, moderately resistant and highly resistant plants. In susceptible plants (Thatcher and TcLr34), H2O2 did not accumulate in the mesophyll cells, and was detected only in the epidermal cells neighbouring the area of developing uredinia 4–8 dpi.
Effective plant resistance against brown rust (incompatible interaction) was associated with an oxidative burst in the mesophyll cells. This concurs with the results of a study on H2O2 accumulation in wheat inoculated with avirulent and virulent strains of P. striiformis f.sp. tritici (Wang et al., 2007). The authors found that a higher rate of mesophyll cells responding with an oxidative burst was strongly correlated with more efficient resistance. In the present experiments, the accumulation of H2O2 was always restricted to the site of fungal infection. Constitutive accumulation of H2O2 that was not related to the infection site was never detected, neither in the susceptible nor in the resistant lines. Although the oxidative response was strong and H2O2 accumulation was easily observable by DAB staining, the amount of accumulated H2O2 in the leaf tissue measured using the FOX method (results not included) was barely detectable. This was probably due to the relatively small number of responding cells in the whole leaf sample taken for the quantitative H2O2 assay.
The pattern of H2O2 accumulation presented in this study was related to the pattern of nitric oxide (NO) accumulation in wheat infected with stripe rust (Guo et al., 2004). Although the interrelation of NO and H2O2 in the wheat response to rust infection still requires analysis, there is data indicating that NO is involved in defence signalling (Floryszak-Wieczorek et al., 2007) and that it interacts with other ROS molecules in plant resistance (Romero-Puertas et al., 2004).
The results indicate that the spatiotemporal pattern of H2O2 accumulation correlates with the similar pattern of the hypersensitive response. These observations are consistent considering that H2O2 can act as a trigger of resistance reactions, including the hypersensitive response (Yoda et al., 2006; Shetty et al., 2008). In the resistant cultivars TcLr9, TcLr19 and TcLr26 (group III), the infection units were associated with few HMC. The micronecrotic reactions were not detectable until 1 dpi, when there was a sharp increase in the oxidative response, which was tightly correlated with an increase of coexisting HMC with necrosis (profile ii). After 2 dpi, the oxidative burst decreased, and at the same time, profile ii (HMC + necrosis) was gradually replaced by profile iii (infection units with micronecrosis but no HMC). In the tested pathosystems, micronecrosis always originated in close association with pathogen structures i.e. HMC, but later it was visible in infection sites without HMC. The relative ratio of both (necrosis with HMC and necrosis without HMC) changed during the infection and was different in groups II and III. Effective resistance (in the lines of group III) was always associated with a high ratio of infection units without HMC, which indicated the rapid elimination of HMC and the termination of pathogen development.
According to Ferreira et al. (2006), the current knowledge of plant-pathogen interactions should alter thinking on the differences between non-host resistance, host-specific resistance and host pathogenesis. Instead of this clear-cut distinction, there is rather a continuum of plant reactions from non-host resistance through host resistance to host pathogenesis. The patterns of oxidative responses (Fig. 5) and wheat-rust interactions (Fig. 3) correspond to this notion in relation to host resistance and host pathogenesis. H2O2 accumulation was detected in all the lines, but the oxidative burst in the mesophyll was only found in incompatible interactions. Although the former reaction was always observed in resistant plants, it varied between the lines: in group II, TcLr24 and TcLr29 responded 1 dpi, while TcLr25 responded 3 dpi. Once initiated, the oxidative response in these lines grew and remained high. In the resistant lines (group III), this response was rapid and transient. It was induced 8 hpi (TcLr9, TcLr19) and 13 hpi (TcLr26). Two dpi, it reached its maximum. The decrease in the oxidative burst was faster in TcLr9 and TcLr26 then in TcLr19. These changes were tightly correlated with the shift from profile ii to iii, indicating rapid pathogen elimination, so one could conclude that the resistance of TcLr9 and TcLr26 was more effective than that of TcLr19. It is worth noting that Lr26 and Lr25 originate from rye (a species that is a non-host for P. triticina), so they could be a part of rye non-host resistance. It would be interesting to learn to what extent the reaction of TcLr26 and TcLr25 to P. triticina infection employs the components of non-host rye resistance.
Histological investigations of host-pathogen interactions revealed that the resistance could be pre- or post-haustorial. Arresting pathogen growth before haustoria formation is common for non-host resistance. Both types of resistance were found to terminate infection early, but only pre-haustorial resistance was not associated with the hypersensitive response. Post-haustorial resistance – typical for race-specific interaction – always involved necrotic reactions (Niks & Dekens, 1991). It was suggested that pre-haustorial resistance was potentially more durable than post-haustorial (Niks & Dekens, 1991). As presented here, the race-specific resistance conferred by the Lr genes could have different characteristics depending on the particular resistance gene. A time-course study of selected processes that are activated during compatible and incompatible wheat-rust interactions might aid the understanding of resistance against P. triticina. The most effective resistance genes (Lr9, Lr19, Lr26) were characterized by a very rapid oxidative and micronecrotic response. Moreover, these reactions were detected in a very restricted area, and they correlated with efficient pathogen arrest in the very early stages of infection. All three genes have a non-wheat origin: Lr26 was derived from rye (Secale cereale), Lr9 from Aegilops umbellulata and Lr19 from Thinopyrum ponticum. Infection of TcLr26 plants led to the development of few HMC per infection unit, associated with reduced host cell necrosis. According to Pathan & Park (2006), the resistance conferred by Lr26 was found in 19 European wheat cultivars. It was the second most common resistance gene (after Lr13), found in 105 tested cultivars. According to data from http://genbank.vurv.cz/wheat/pedigree/Lr9 was inbred into 40 cultivars with its first release in 1965. Lr19 was inbred into 27 cultivars with its first release in 1960. Since then, these three genes have given relatively durable resistance against P. triticina.
Considering the characteristics of rust-wheat interactions and the relatively durable resistance conferred by the three genes, it is hypothesized that certain types of post-haustorial resistance might be durable if they are derived from a non-host species, and if they lead to pre-haustorial type of resistance, i.e. rapid termination of the pathogen with few HMC and strongly reduced host cell necrosis.
This study was financed by grant PBZ-MNiSW-2/3/2006 and the Scientific Network ‘Genomics and transgenesis of agricultural crops’.