Fungal plant pathogens that attempt to penetrate and feed on living cells frequently trigger a localized plant defence response that results in fungal penetration failure. In the current study we demonstrate that breakdown products of the cell wall released by the localized application of hemicellulase elicit localized responses including, sequentially, extracellular H2O2 generation; accumulation of phenolic compounds; and cross-linking of proteins in the cell wall. In a detailed time-course study of three plant–fungus interactions that result in a high frequency of penetration failure, only one plant–fungus combination displayed a similar profile of responses to that induced by localized cell-wall degradation. The additional generation of extracellular O2– in one interaction, and the absence of phenolic compounds in the cell wall in another, demonstrate that plant responses to the penetration process may be influenced by activities of the penetrating fungus. Significantly, H2O2 generation was the only response detected in all three plant–fungal combinations at the correct time and place to account for penetration failure, and in all three combinations the enzymatic removal of H2O2 resulted in increased penetration success. Pharmacological studies suggest that in two of the three interactions, H2O2 generation required cytoskeletal involvement but was independent of transcription or translation, although inhibition of the latter processes increased fungal penetration. In at least one of these two interactions, the data suggest that H2O2 generation and new gene expression act within the same penetration-inhibiting pathway, possibly through the involvement of phenolic materials. However, enzymatic removal of H2O2 from the third interaction almost completely eliminated penetration failure, while interference with cytoplasmic processes had no effect, suggesting that H2O2 generation in this system did not require protoplast involvement and, alone, was necessary and sufficient to account for fungal penetration failure.
Fungal plant pathogens, unlike other pathogenic microbes such as bacteria, are unique in their invasiveness, often directly penetrating the plant cell wall to gain access to cell contents. For those fungi for which penetration of a plant cell wall is an essential part of the initial pathogenesis process, stopping the fungus during penetration can be an effective and energetically frugal means of plant defence. Not surprisingly, therefore, penetration failure is commonly associated with disease resistance, and numerous fungal-induced modifications of the plant cell wall have been reported that could theoretically hinder the pathogen as it attempts to enter the cell. These include cross-linking of pre-existing or induced cell-wall proteins and phenolics (Aist and Bushnell, 1991; Bolwell et al., 2001; Brisson et al., 1994; McLusky et al., 1999; Thordal-Christensen et al., 1997); formation of calcium pectate gels (Kieffer et al., 2000); accumulation of hydroxyproline-rich glycoproteins (HRGPs) (Mazau and Esquerré-Tugayé, 1986), phenolic compounds or silica (Aist and Bushnell, 1991); the deposition of callose-containing papillae (Aist and Bushnell, 1991); and the extracellular generation of reactive oxygen species (ROS) (Thordal-Christensen et al., 1997). Commonly, many of these responses occur in combination (Perumalla and Heath, 1991), and despite the fact that chemical (Sherwood and Vance, 1980) or fungal (Mellersh and Heath, 2001) inhibition of all detectable wall modifications allow successful fungal entry into cells, each individual response may not contribute equally, or at all, to the inhibition of fungal growth (Perumalla and Heath, 1989). Evidence that a particular plant response provides an effective barrier to fungal ingress is scarce, in part because elimination of one effective response can, in some cases, lead to an increase in the efficacy of another (Heath and Stumpf, 1986).
Although localized extracellular generation of ROS has been reported during penetration events in a number of plant–fungal interactions (Doke, 1983; Hückelhoven and Kogel, 1998; Thordal-Christensen et al., 1997), and has been suggested to have an antimicrobial effect in plant–bacterial interactions (Bestwick et al., 1997), the role of ROS in the inhibition of wall penetration by fungi remains enigmatic. There are a number of potential enzymatic mechanisms for the extracellular generation of ROS, and the type and source of ROS has been shown to vary depending on the plant–pathogen combination (Bolwell, 1999). There also are a number of ways in which ROS have been suggested to participate in plant defence, including acting as a signalling agent (Lamb and Dixon, 1997), or causing reinforcement of the cell wall through oxidative cross-linking (Brisson et al., 1994; Brown et al., 1998; McLusky et al., 1999). In addition, ROS have the potential to be directly toxic as exogenous H2O2 can inhibit fungal growth in vitro (Lu and Higgins, 1999) and can prevent infection in leaf discs (Peng and Kuc, 1992). Nevertheless, despite some evidence that transgenic plants expressing H2O2-generating enzymes display increased protection against bacterial and fungal pathogens (Schweizer et al., 1999; Wu et al., 1995), there is little direct evidence that ROS generation in naturally occurring plant–pathogen interactions actually determines resistance. Probably the best evidence is the requirement of some (Santos et al., 2001; Xu and Pan, 2000), but not all (Miguel et al., 2000), plant pathogenic bacteria for ROS scavenging enzymes as virulence factors. However, no similar data are currently available for plant pathogenic fungi.
In this study we demonstrate that localized generation of H2O2 is one of the earliest cytologically detectable defence responses to both experimentally induced localized wall degradation and the penetration of plant epidermal cell walls by three taxonomically distinct fungal pathogens. A detailed examination of the timing of H2O2 generation relative to the expression of other wall-associated responses, as well as pharmacological studies, implicate the rapid generation of H2O2 in response to cell-wall penetration as one of the most important determinants of fungal penetration failure in invading epidermal cells. However, our results suggest that although H2O2 generated in the wall may have a direct role in restricting fungal growth in one plant–fungal combination, in at least one of the others it most likely acts within a pathway involving transcription/translation and the expression of wall-associated responses such as the accumulation of phenolic compounds.
Mimicking the localized wall degradation caused by fungal penetration elicits wall-associated responses
Previous studies have implicated cell-wall components released during fungal penetration as potential elicitors of localized cytoplasmic plant responses (Heath et al., 1997). To determine whether localized wall degradation can elicit wall-associated responses, we applied hemicellulase (a mixture of hydrolytic enzymes) to small scratches in the cuticle of cowpea epidermal cells, as described previously (Heath et al., 1997). The hemicellulase solution contained 0.05% aqueous calcofluor white, which binds cell-wall polymers, allowing visualization of the scratches under UV irradiation. Leaves were treated for visualization of extracellular generation of O2– and H2O2, accumulation of phenolic compounds, and protein cross-linking at 6, 24 and 48 h after hemicellulase was applied. Callose was detected as refractive localized deposits (papillae) between the cell wall at the plasma membrane at the scratch site, and confirmed by its fluorescence under UV illumination after aniline blue treatment.
Nitroblue tetrazolium (NBT) staining for O2– was not associated with calcofluor-stained hemicellulase-treated scratches at any of the three time points examined. However, treatment with 3,3′-diaminobenzidine-tetrahydrochloride (DAB) revealed a reddish-brown precipitate indicative of H2O2 production associated with 20–30% of scratches within 6 h of hemicellulase application (Table 1; Figure 1a). The percentage of scratches associated with DAB staining increased to 50–60% by 24 h, and staining was still frequent at 48 h (Table 1), although calcofluor staining had faded by this time, hindering numerical data collection. No DAB staining was seen in response to small scratches in the cuticle in the absence of hemicellulase (not shown).
Table 1. Time-course of responses in cowpea epidermal cell walls at the sites of hemicellulase-treated scratches
Hours after application
Values are given as a percentage of hemicellulase application sites visualized by calcofluor fluorescence. The number of scratches varied from leaf to leaf; in most cases the values presented represent the range of percentages seen over four leaf pieces with at least 25 scratches examined per leaf piece.
Numerical data were difficult to collect at 48 h because of fading of the calcofluor fluorescence.
NBT staining for superoxide
DAB staining for hydrogen peroxide
Toluidine blue staining for phenolics
Coomassie blue staining for protein cross-linking
Accumulation of phenolic compounds, as determined by toluidine blue staining, did not occur as rapidly as H2O2 production, and was associated with less than 10% of scratches 6 h post-treatment (Table 1; Figure 1b). However, phenolics were associated with 50–60% of scratches by 24 h, and were frequently seen in a similar pattern at 48 h (Table 1).
Protein cross-linking, visualized using a modified version of a Coomassie blue staining technique (Thordal-Christensen et al., 1997), was not seen prior to 48 h following hemicellulase application, and even then was only seen infrequently (Table 1; Figure 1c). However, the SDS treatments associated with this technique appeared to interfere with calcofluor staining, making scratch visualization difficult.
Callose deposition was not seen following hemicellulase treatment at any time point unless the scratch was deep enough to have broken through the cell wall, in which case its deposition was seen even in the absence of hemicellulase.
Expression of wall-associated responses is influenced by the penetrating fungus
To compare wall-associated plant responses to experimentally induced localized wall degradation with those accompanying attempted fungal penetration, we studied three pathosystems. Two were non-host interactions involving the plantain powdery mildew fungus (Erysiphe cichoracearum) and cowpea plants, and the cowpea rust fungus (Uromyces vignae) and pea plants. The third was a host interaction involving the tomato anthracnose fungus (Colletotrichum coccodes) and cotyledons of young tomato plants. In the non-host interactions, successful penetration into leaf epidermal cells resulted in fungal growth being curtailed by fungal encasement in callose and/or hypersensitive death of the invaded cell. In tomato, successful epidermal penetration eventually resulted in continued fungal growth and disease symptoms. Nevertheless, in all three interactions the majority (>75%) of penetration attempts failed, and the fungus became trapped either within the epidermal cell wall or, often in the case of the powdery mildew fungus, within a callose papilla between the cell wall and the plasma membrane.
In a detailed time-course study, only the responses of pea to penetration by the cowpea rust fungus resembled those detected in response to hemicellulase-treated scratches in cowpea leaves (Figure 1d–f). Despite the absence of detectable O2– following hemicellulase treatment of cowpea leaves, a diffuse, transient NBT staining was seen prior to attempted penetration under newly formed appressoria of the powdery mildew fungus on this plant (Figure 1g) as early as 6 h post-inoculation. Such staining had not been detected in a previous, less detailed investigation of this pathosystem in which NBT staining for O2– had transiently been seen only in anticlinal and periclinal cell walls near (but not at) the site of fungal penetration (Mellersh and Heath, 2001). The improved digital camera optics used in the current study also allowed us to determine that NBT staining was localized between the plant plasma membrane and cell wall, rather than in the wall itself. O2– generation was not seen in either of the other two plant–fungal interactions.
DAB staining revealed that localized, catalase-sensitive H2O2 generation was the earliest cytologically detectable wall-associated response common to all three plant–fungal interactions. In the interaction between the powdery mildew fungus and cowpea, DAB staining was seen in both periclinal (Figure 1h) and anticlinal cell walls, as well as in callose papillae, underlying appressoria as early as 12 h post-inoculation, and was common by 15 h. This fungus is unusual in that it may produce up to three penetration attempts from the single appressorium, and DAB staining was seen at all three even though papillae generally were observed only at sites of the second and third attempts. DAB staining was often less intense at the second penetration attempt, which was also most frequently associated with successful cell penetration. A correlation between the absence of H2O2 in papillae and penetration success has also been seen for the barley powdery mildew fungus (Hückelhoven et al., 1999; Hückelhoven et al., 2000), and in our system this absence could reflect a period of desensitization after the first induction of H2O2, as demonstrated in cell suspension culture (Chandra et al., 2000). No callose papillae were seen in the other two plant–fungus combinations although intense DAB staining was also seen in periclinal and anticlinal cell walls underlying appressoria of the cowpea rust fungus on pea plants (Figure 1d) and the tomato anthracnose fungus on tomato plants (Figure 1j). Staining was detected as early as 12 h post-inoculation in the latter interaction and as early as 6 h post-inoculation in the former. In all three plant–fungal combinations, the first DAB staining coincided with the time when penetration pegs first become visible in the cell wall.
Although the accumulation of phenolic compounds followed H2O2 generation in scratched walls treated with hemicellulase, there were almost no (<1%) cytologically detectable signs of phenolic accumulation in the tomato anthracnose fungus/tomato interaction. However, accumulation of autofluorescent and/or toluidine blue-stained phenolics were seen in both the rust fungus/pea interaction (Figure 1e) and the interactions between the powdery mildew fungus and cowpea plants (Figure 1i). Detectable accumulation of phenolic compounds in the latter interaction was correlated with the first appearance of H2O2 at around 12 h post-inoculation, but was first detectable in the cowpea rust fungus/pea interaction at about 15 h post-inoculation, which was considerably later than the appearance of H2O2.
Protein cross-linking is a potential side effect of H2O2 generation that could provide cell-wall reinforcement against fungal penetration. We therefore examined protein cross-linking using a modified version of the Coomassie blue staining technique (Thordal-Christensen et al., 1997). The appearance of protein cross-linking was seldom detected prior to 48 h post-inoculation in either the tomato anthracnose fungus/tomato interaction or the cowpea rust fungus/pea interaction, and it was normally found in anticlinal epidermal walls (Figure 1f,k), and to a lesser extent in periclinal walls, under the appressorium. Protein cross-linking was not detected in the powdery mildew fungus/cowpea interaction using this technique, perhaps because of the eventual impregnation of the plant cell wall and papillae with impermeable compounds such as silica that are known to accumulate in response to powdery mildew penetration (Aist and Bushnell, 1991).
H2O2 generation plays a role in restricting fungal penetration in each of the three plant–fungal interactions
The ubiquitous presence of localized H2O2 generation at the initial stages of fungal penetration in all three plant/fungal interactions suggests that it might be an important component of resistance to fungal penetration. To explore this hypothesis, we treated plants with the H2O2 scavenger catalase prior to inoculation. Removal of H2O2 in this manner is unlikely to have unwanted pleiotropic effects on plant cells as the enzyme has a specific, well characterized reaction mechanism (which results solely in the production of molecular oxygen and water), and is unlikely to cross the plant plasma membrane. In all three interactions, prior treatment of plants with catalase caused a significant reduction in the number of sites of attempted penetration with DAB staining (Table 2), and this corresponded with a significant increase in fungal penetration efficiency (Table 2). Strikingly, while removal of H2O2 was not 100% effective in any of the three plant fungal interactions, and only partially increased fungal penetration efficiency in both the powdery mildew fungus/cowpea and rust fungus/pea interactions, catalase treatment almost completely eliminated penetration failure in the tomato anthracnose fungus/tomato interaction (Table 2).
Table 2. Effect of reactive oxygen scavengers on the presence of localized DAB staining, NBT staining and fungal penetration efficiency in three plant–fungal interactions
Asterisk indicates treatment is significantly different from control at P
Percentage of appressoria associated with reddish-brown precipitates in the plant cell wall.
Percentage of appressoria associated with diffuse blue staining between the plant cell wall and plasma membrane.
Percentage of appressoria that subsequently successfully penetrated the plant cell wall to form intracellular structures or that penetrated successfully at the second penetration attempt for powdery mildew fungus/cowpea interaction.
Percentage of sites with DAB response at 24 h post-inoculation; penetration efficiency at 48 h post-inoculation.
N/A, not applicable.
Cowpea rust fungus on pea (18 h)
DAB staining (%)c
75.0 ± 9.5
47.0 ± 5.7*
75.0 ± 6.6
NBT staining (%)d
Penetration efficiency (%)e
22.5 ± 7.5
43.0 ± 7.7*
23.0 ± 4.7
Plantain powdery mildew fungus on cowpea (22 h)
DAB staining (%)
86.5 ± 7.5
44.7 ± 3.0*
89.0 ± 3.5
NBT staining (%)
12.0 ± 8.6
2.0 ± 4.0*
Penetration efficiency (%)
17.0 ± 4.7
38.5 ± 5.9*
13.0 ± 3.8
Tomato anthracnose fungus on tomato (24/48 h)f
DAB staining (%)
77.0 ± 5.1
26.0 ± 2.8*
73.5 ± 3.7
NBT staining (%)
Penetration efficiency (%)
2.5 ± 2.9
96.0 ± 7.8*
3.0 ± 2.6
The relationship of H2O2 generation to subsequent protein cross-linking within the wall was illustrated by the effect of catalase on this response. This effect was most readily examined in the tomato system because the melanized appressoria of the anthracnose fungus best survived the SDS treatments needed to visualize cross-linked proteins. In this system, catalase treatment reduced the percentage of sites displaying protein cross-linking from 75.5 ± 5.7 to 31.0 ± 6.6 (significant at P ≤ 0.01).
As would be expected, the O2– scavenger superoxide dismutase (SOD) had no effect either on DAB staining or on fungal penetration efficiency in either the rust fungus/pea or anthracnose fungus/tomato interactions (Table 2), both of which lacked any NBT-detectable O2–. However, although SOD caused a significant reduction in NBT staining in the powdery mildew fungus/cowpea interaction, there was no effect on fungal penetration efficiency (Table 2). Together these data suggest that H2O2 plays a more important role than O2– in restricting fungal growth during wall penetration in this system.
Transcription/translation-dependent and actin cytoskeleton-dependent responses also play a role in preventing fungal penetration in two of the three plant–fungal interactions
To examine whether activities of the protoplast are involved either in the generation of H2O2 and/or in restricting fungal penetration of the plant cell wall, we employed the transcription inhibitor actinomycin D, the translation inhibitors blasticidin S or cycloheximide (the latter was used in the case of the powdery mildew fungus due to adverse effects of blasticidin S on fungal growth), and the actin microfilament-disrupting agent cytochalasin E. H2O2 generation (as indicated by DAB staining) was not dependent on either transcription or translation in any of the three plant–fungal systems, but the transcription inhibitor increased penetration efficiency in both the rust fungus/pea and the powdery mildew fungus/cowpea combinations (Table 3). In these same combinations, H2O2 generation was reduced, and penetration efficiency increased, by interference with the actin cytoskeleton (Table 3). However, none of the treatments reduced H2O2 generation or increased penetration efficiency in the anthracnose fungus/tomato interaction (Table 3), even though the concentrations of transcription and translation inhibitors were sufficient to delay a bacterially induced hypersensitive response in the same plants (not shown).
Table 3. Effect of actinomycin D, blasticidin S or cycloheximide and cytochalasin E on the presence of localized DAB staining at fungal penetration sites and on fungal penetration efficiency in three plant–fungal interactions a
Activity of the plant protoplast and H2O2 generation act in concert for the success of defence responses in the powdery mildew fungus/cowpea interaction
The fact that scavenging H2O2 and interference with plant cell transcription both led to increases in fungal penetration efficiency in two of the three pathosystems studied could have at least two explanations. One possibility is that H2O2 primarily acts either as a signalling agent to trigger responses in the protoplast that lead to the secretion of fungus-inhibiting substances, or as a defensive agent that has to act in concert with some other plant activity that requires gene expression. Alternatively, H2O2 may have a direct defensive effect on the fungus or on the structure of the plant wall, and the effect of transcription and translation inhibitors may reflect protoplast involvement in other additional defences. To distinguish between these two hypotheses, we performed a double-inhibitor experiment in which we compared penetration efficiency in plants treated with either the H2O2 scavenger catalase or the transcription inhibitor actinomycin D alone, or in combination. We focused on the powdery mildew fungus/cowpea interaction because this system has less natural variability in penetration efficiency than the rust fungus/pea combination, and unpublished data indicate that pea cell walls contain constitutive features that contribute to fungal penetration failure. In contrast, constitutive defensive features do not appear to be involved in the powdery mildew/cowpea interaction, as treatment of this system with cytochalasin E can almost eliminate penetration failure (Table 3). Nevertheless, in this pathosystem, individual catalase or actinomycin D treatments did not increase penetration efficiency beyond about 50%, and treatment with both pharmacological agents simultaneously did not lead to further increases even though the addition of cytochalasin E to these combined agents increased values by another 30% (Table 4). These results indicate that H2O2 generation and transcription-dependent defences do not act independently as defensive features but, instead, act within a common pathway.
Table 4. Effect of catalase, actinomycin D, and cytochalasin E on penetration efficiency, callose deposition and accumulation of toluidine blue-stained phenolics in the powdery mildew fungus/cowpea interaction
Represents the number of appressoria that successfully formed haustoria in epidermal cells on their second penetration attempt.
Values in the same column followed by an asterisk are significantly different from the MES control at P
≤0.05. Values for treatments followed by
showed a significant difference (P
0.05) when compared with amalgamated data sets from CAT, ACT D and CAT + ACT D treatments within the same column (the latter treatments were never significantly different from one another for any of the traits observed).
Collars and encasements form around haustoria and represent the continued addition of callose to papillae that develop around the fungal penetration peg as it enters the cell lumen.
The penetration-increasing effects of the pharmacological agents, singly or in combination, were directly correlated with the decrease in the incidence of papillae that stained for phenolics with toluidine blue (Table 4). In contrast, only cytochalasin E, alone or with catalase and actinomycin D, had an effect on callose deposition. Papillae in cytochalasin E-treated tissue were often large and diffuse, and the size or presence of callose collars or encasements around haustoria formed at sites of successful penetration was significantly reduced (Table 4). Presumably, this lack of any effect of actinomycin D on callose deposition (Table 4), even though it has been shown to decrease this process in cowpea in an earlier investigation (Škalamera et al., 1997), is due to the much lower concentration used in the present study.
A large number of fungal plant pathogens obtain entry into their host plants by directly penetrating the epidermal wall. It is not surprising, therefore, that plants have evolved to recognize and respond to events associated with this localized damage. For example, penetration of a cell with a needle will result in the rapid encasement of the needle tip in callose (Russo and Bushnell, 1989), and localized pressure applied to a cell surface can result in elevated intracellular ROS levels and upregulation of defence-associated genes (Gus-Mayer et al., 1998). Defence responses can also be elicited by the general application of degradation products of the plant wall (Esquerré-Tugayéet al., 2000), and the localized breakdown of the cell wall leads to translocation of the nucleus to the degradation site (Heath et al., 1997). In the current study we have shown that this localized degradation of cowpea epidermal walls can also elicit extracellular generation of H2O2, accumulation of phenolic compounds in the cell wall, and cell-wall protein cross-linking. Not unexpectedly, the same responses were observed around the penetration point of two of the three different, direct-penetrating fungi that we investigated in this study. However, although the localized hemicellulase treatment did not elicit O2– generation in cowpea epidermal walls, the plantain powdery mildew fungus did, perhaps in response to the activity of cutinases known to be released by some powdery mildew fungi prior to cell penetration (Pascholati et al., 1992). Moreover, although localized hemicellulase treatment led to phenolic accumulation in cowpea epidermal walls, no phenolic accumulation was seen in cell walls of tomato during failed penetration by the tomato anthracnose fungus. The fact that this fungus also did not elicit this response in pea (D.G.M. and M.C.H., unpublished results), despite the fact that phenolic accumulation was triggered in this non-host plant by the cowpea rust fungus, suggests that the tomato anthracnose fungus either inhibits the response or has a different mode of wall penetration from the other two pathogens, perhaps involving mechanical pressure as is the case for other fungi that produce appressoria with rigid melanized walls (Bechinger et al., 1999). Overall, these results suggest that the number and variety of wall-associated responses expressed during penetration by any given fungus are not only a result of recognition and response to wall degradation, but also may be influenced by other fungal activities.
Although O2– generation was observed in the powdery mildew fungus/cowpea interaction, the O2– scavenger SOD had no effect on fungal penetration in any of the three pathosystems studied, and the temporal and spatial separation of this ROS from H2O2 generation in the powdery mildew fungus/cowpea interaction suggests that the two ROS species are produced by different enzymatic systems. The lack of detectable O2– at sites of H2O2 detection is interesting as potential sources of H2O2 theoretically should generate an O2– intermediate. However, in legumes extracellular peroxidases, rather than O2–-releasing NADPH oxidase, are commonly are involved in the generation of H2O2 (Bolwell, 1999), and these enzymes are believed to produce H2O2 via a bound O2– intermediate which, because it is in association with compound III, may not necessarily be detectable.
The timing of the various wall-associated responses differed between the three interactions, providing some clues as to which of these responses may have effective defensive functions. In the plantain powdery mildew/cowpea interaction, H2O2 generation was accompanied by the accumulation of phenolic compounds, as well as the formation of callose papilla around penetration pegs where they protruded into the cell lumen, so all three responses could theoretically have contributed to penetration failure. However, in the cowpea rust fungus/pea interaction, only H2O2 generation was seen at the time that penetration failed. Callose papillae were infrequent and phenolic compounds were not seen until 15 h post-inoculation, much too late to account for penetration attempts that failed between 6 and 9 h post-inoculation. Similarly, in the tomato anthracnose fungus/tomato interaction, H2O2 generation only was seen at the time of penetration failure; callose papillae were not seen; and phenolic compounds were detected infrequently. Cytologically detectable cross-linking of cell-wall proteins was not usually evident in any interaction until long after fungal growth had ceased. However, sensitivity of detection could be an issue as H2O2 can contribute to peroxidative cross-linking within minutes of its generation (Bradley et al., 1992), and undetectable cross-linking may have begun earlier when H2O2 generation was initiated. Nevertheless, it is debatable whether this defence response is likely to be effective when it is present in so low a quantity as to be cytologically undetectable. These results, in total, show that of all the wall-associated responses examined in this study, only H2O2 generation occurred at the time of penetration failure in all three pathosystems.
To investigate the function of H2O2 generation in preventing fungal penetration in the various plant–fungal interactions, we chose to employ enzymatic ROS scavengers because not only do they act extracellularly as they are unable to cross the plasma membrane, but they also have specific, well documented reaction mechanisms, making them unlikely to have pleiotropic effects on the plant cells being studied. The results of catalase treatment to specifically scavenge H2O2 demonstrated that H2O2 plays a role in determining penetration failure in all three of the plant/fungal interactions. In both the rust fungus/pea and powdery mildew fungus/cowpea interactions, catalase, which reduced H2O2 generation by about 50%, partly eliminated penetration failure. A similar effect was found for transcription/translation inhibitors even though they did not affect H2O2 generation. These results raise the possibility that H2O2 exerts its effects on fungal penetration indirectly by acting (a) as a signalling molecule that triggers gene activation, or (b) as a cofactor in a process that requires new gene expression. Support for this hypothesis was found in the powdery mildew/cowpea combination in which, unlike the rust fungus/pea combination, constitutive inhibitory features of the plant wall do not significantly contribute to penetration failure. Simultaneous scavenging of H2O2 and inhibiting transcription did not improve fungal penetration beyond that seen following either treatment alone, indicating that H2O2 generation and gene activation work within the same pathway to bring about fungal penetration failure. The single and combined pharmacological agents had no effect on callose deposition, but their effects on fungal penetration were directly the inverse of their effects on the accumulation of toluidine blue-stained phenolic compounds in the callose papillae that form around penetration pegs of the powdery mildew fungus as they begin to enter the plant cell lumen. A correlation between fungal penetration failure and the presence of (autofluorescent) phenolic compounds in papillae has been shown previously (Aist and Bushnell, 1991), and phenolic compounds secreted into the apoplast, together with peroxidative cross-linking of these compounds into the cell wall, have been suggested to produce a physical barrier in onion to fungal penetration (McLusky et al., 1999). In demonstrating the involvement of both H2O2 and transcription in the same fungus-inhibitory pathway, our data support this latter model. However, the temporal difference shown in the present study between H2O2 generation and phenolic deposition within the pea wall after inoculation with the cowpea rust fungus suggests that this plant–fungal combination may differ from the powdery mildew/cowpea combination in the role of phenolic compounds in penetration failure.
In the rust fungus/pea and powdery mildew/cowpea interactions, the induction of H2O2 generation appeared to require the presence of an intact actin cytoskeleton. Indeed, in the powdery mildew/cowpea interaction the actin microfilament-disrupting agent, cytochalasin E, was more efficient at reducing H2O2 generation and phenolic accumulation, and causing an increase in penetration efficiency, than in any other treatment. In stark contrast, transcription/translation inhibitors had no effect on fungal entry into the cell in the tomato anthracnose fungus/tomato combination, and no wall-associated response other than H2O2 was detected at the time of penetration failure. Moreover, no additional cytosolic participation was required for stopping fungal growth as neither treatment with cytochalasin E in this study, nor RGDS peptides (Foulds, 2000) that interfere with communication between the cell wall and the cell interior (Mellersh and Heath, 2001), increased penetration success. These results, together with the fact that H2O2 removal by catalase treatment almost completely eliminated penetration failure for this fungus, strongly suggest that H2O2 generation alone is responsible for arresting fungal growth in this plant–fungal combination, and that this H2O2 can be elicited and generated without any involvement of the plant protoplast.
The overall results of our study therefore reveal that there are at least two models by which H2O2 may act to prevent fungal penetration into plant epidermal cells (Figure 2). In one model, involvement of the plant protoplast via the actin cytoskeleton is required to generate extracellular H2O2, and this ROS either acts as a signal for gene activation resulting in fungus-inhibitory materials such as phenolics being secreted into the plant wall, or has to interact with such materials in the wall to cause penetration-inhibiting changes. In the other model, H2O2 is generated without any involvement of the plant protoplast and alone is necessary and sufficient to account for fungal penetration failure.
Plants and fungal inoculation
Conidia of the plantain powdery mildew fungus (Erysiphe cichoracearum) were produced on susceptible plantain plants and brushed on the upper surface of the primary leaves of 8-day-old cowpea (Vigna unguiculata cv. Dixie Cream) plants as described previously (Mellersh and Heath, 2001).
A conidial suspension (5 × 105 conidia ml−1 containing 0.05% v/v Tween 20) of the tomato anthracnose fungus (Colletotrichum coccodes) was sprayed on the upper surface of cotyledons of 20-day-old tomato (Lycopersicon esculentum cv. Moneymaker) seedlings grown in flats and covered with clear plastic lids to maintain high humidity for the duration of each experiment. The growing tips of the plants were removed after first leaf emergence to ensure that the cotyledons would expand and be retained on the plant.
Cowpea rust fungus (Uromyces vignae) basidiospores were used to inoculate the second set of true leaves of 12-day-old pea (Pisum sativum cv. Alaska) plants as described previously (Mellersh and Heath, 2001).
All plants were grown in a lighted growth chamber at 22°C with 16 h photoperiod at 150 µE m−2 sec−1 (200 µE m−2 sec−1 for tomato).
For all pharmacological treatments, the intercellular spaces of leaves were injected with the appropriate chemical or control solution using a 30-gauge needle and a 1 cc syringe. The water-soaked appearance of leaves was allowed to disappear prior to inoculation with any of the fungi (usually about 1 h).
The following chemicals/concentrations were used for injections: catalase, 1100 U ml−1 in 10 mm MES buffer pH 6.5; superoxide dismutase (SOD), 900 U ml−1 in 10 mm MES buffer pH 6.5; actinomycin D, 1.0 µg ml−1 (10 µg ml−1 for the C. coccodes/tomato interaction) in water; blasticidin S, 1 µg ml−1 in water; cytochalasin E, 1 µg ml−1 in 0.01% DMSO. All chemicals except actinomycin D (Calbiochem, San Diego, CA, USA) and cytochalasin E (Sigma-Aldrich, Ontario, Canada) were from ICN Pharmaceuticals (Costa Mesa, CA, USA).
Visualization of defence responses
For time-course studies of defence responses, leaf pieces were generally harvested at 3, 6, 9, 12, 15, 18, 21, 24, 36 and 48 h post-inoculation. H2O2 was visualized by injecting the intercellular spaces of leaves with an aqueous solution of 2 mg ml−1 3′3-diaminobenzidine-tetrahydrochloride (DAB) (Thordal-Christensen et al., 1997) 2 h before tissue harvest. Leaves were injected with an aqueous solution of 0.05% nitroblue tetrazolium (NBT) 30 min before harvest to visualize O2– (Doke, 1983; Heath, 1998). In either case, leaves were fixed and decolorized in boiling 95% (v/v) ethanol before being cleared in saturated chloral hydrate and mounted in modified Hoyer's medium (Stumpf and Heath, 1985) prior to being examined with a Reichert–Jung Polyvar microscope (Reichert AG, Vienna, Austria) equipped with differential interference contrast optics. Phenolic compounds were visualized as either autofluorescence under blue light epifluorescence (filter cube B1, excitation filter BP 330–380 nm, barrier filter LP 450–495 nm, and dichroic mirror DS 510 nm) in unstained, decolorized, cleared tissue, or by their blue/turquoise appearance in leaves that had been decolorized in ethanol and then stained briefly in a solution of 0.05% toluidine blue in 50 mm citrate buffer pH 3.5 (O'Brien et al., 1964).
In order to visualize protein cross-linking, we used a modified version of the protocol described by Thordal-Christensen et al. (1997). Leaves were fixed and decolorized in 95% (v/v) ethanol, then submerged in 1% SDS for 24 h at 80°C. Leaves were then stained for 10–30 min in 0.1% Coomassie blue in 40% ethanol/10% acetic acid and rinsed in a solution of 40% ethanol/10% acetic acid prior to being mounted in distilled water. Areas of cell wall with protein cross-linking were visualized by their deep purple/blue colour.
Callose was generally visualized as refractive papillae at the penetration site, but in some leaf pieces its presence was confirmed by its fluorescence under UV epifluorescence (filter cube U1, excitation filter BP 330–380, barrier filter LP 418 and dichroic mirror DS 420) as described by Škalamera et al., 1997.
Leaf abrasion experiments and localized degradation of epidermal walls
Small scratches in the cuticle of the upper surface of primary leaves of cowpea plants were created by gently stroking the leaves with a medium-soft artist's paintbrush that had been dipped in 0.05% (aqueous) calcofluor white (Fluorescent Brightener; Sigma-Aldrich Canada, Oakville, ON, Canada) containing 1 mg ml−1 hemicellulase (InterSpex Products Inc., Foster City, CA, USA). At 6, 24 and 48 h after creating these scratches, leaves were treated for visualization of various defensive responses in the same manner as fungus-inoculated tissue. Under the microscope, scratches were detected by their bright blue-white fluorescence under UV epifluorescence, as described previously (Heath et al., 1997).
Images were acquired using a Nikon (Nikon Instruments Inc., Melville, NY, USA) colour digital still camera, model DXM1200 and the act-1 image acquisition software.
Data collection and analysis
The data presented represent means of percentages calculated from at least 50 fungal sites with appressoria on each of four leaf pieces per treatment. For E. cichoracearum, penetration data represent the percentage of appressoria that successfully formed haustoria on their second penetration attempt, as the first attempt almost always failed regardless of treatment (Mellersh and Heath, 2001). Where necessary, data were normalized using the arcsin square-root transformation before application of a t-test. All experiments were performed twice with similar results; typical data from one experiment are shown.
We thank Rosemarie Christopher-Kozjan, Alice Cheung and Pui Tam for excellent technical assistance. This research was supported in part by funds from the Natural Sciences and Engineering Research Council of Canada and by an Ontario Graduate Scholarship to D.G.M.