The cell wall of the fungus and the interaction apparatus The photomicrographs presented in this paper document the existence of β-1,3-glucans and chitin in the cell wall of C. trifolii. Both polymers are known to be the most important structural components of ascomycetous cell walls (Kuhn et al., 1990; Ruiz-Herrera, 1992). The fact that they are also present in the IA, as shown by dense labelling in either case, provides evidence for the wall-like nature of that structure (Simon et al., 2004). It is worth noting the different labelling patterns of the various methods employed to detect chitin. While WGA in combination with an ovomucoid–gold complex reliably labelled the wall and the IA, it was almost impossible to avoid gold clusters. Furthermore, large areas of the hyphal walls remained unlabelled, although it is unlikely that chitin should not be present at these places.
This method only worked with Epon-embedded samples. We therefore tested biotinylated WGA, various secondary or even tertiary antibodies. Not all of the WGAs and antibodies worked well (not shown), but the ones referred to in here resulted in very neat labelling. Although WGA from Sigma-Aldrich consistently labelled the fungal wall, only Vector WGA regularly labelled the IA. Furthermore, the latter showed no preference for any of the tested embedding resins, while the former resulted in only weak labelling of Epon-embedded samples. Therefore we recommend to try more than one procedure when using WGA to locate chitin.
The plant cell wall and its alterations within the tube Most antibodies against plant cell wall constituents used in the present work gave reliable and reproducible results. There were, however, some exceptions: Three polyclonal antibodies, directed against xyloglucans (XGs), rhamnogalacturonan-I (RG-I)/polygalacturonic acids (PGAs) and extensins (Moore et al., 1986; Moore et al., 1991; Lynch & Staehelin, 1992), heavily labelled the clover cell wall. Unfortunately, they also had a very high affinity for unknown epitopes in the wall of the fungus and in the IA, and were therefore useless for the present study. Such cross-labelling has been reported for Trichoderma cellobiohydrolase-I recognizing β-1,4-glucans in the cell walls of Calluna vulgaris and Hymenoscyphus ericae (Bonfante, 1994). Very little to no labelling was obtained with the monoclonal antibody JIM 5, which recognizes pectin epitopes with a low degree of esterification (Knox et al., 1990), and CCRC-M2 with a high affinity for an unknown epitope in RG-I (Puhlmann et al., 1994). The respective epitopes might have been masked since, especially in the former case, it is unlikely that the middle lamella and cell corners, which in many plants and plant organs have been shown to be reactive with JIM 5 (VandenBosch et al., 1989; Knox et al., 1990; Rioux et al., 1995; Wisnieswski & Davies, 1995; Majewska-Sawka & Münster, 2003), were almost completely devoid of such homogalacturonans.
The labelling pattern of those antibodies that worked gave a sufficiently detailed overview of the changes occurring in the tube linking IA and the host bubble. Unlike intracellular fungi, which must break through the host wall and therefore need to dissolve its cellulose microfibrils and xyloglucan network, those polymers are apparently not attacked by C. trifolii. The consistent labelling obtained for cellulose and xyloglucan at the interaction zone suggests that these main structural elements of dicotyledonous plant cell walls (Carpita & Gibeaut, 1993; Schindler, 1993) are left unaltered by the fungus.
The antibody used against xyloglucan, CCRC-M1, was primarily generated against RG-I, but exhibited a much higher affinity for several XGs (Puhlmann et al., 1994). It cannot be ruled out that CCRC-M1 and CCRC-M7, the latter binding to epitopes in RG-I present in the inner half of the plant cell wall and, to a smaller degree, plant membrane proteins (Puhlmann et al., 1994), may have recognized similar antigens. Yet, the slightly different labelling patterns of CCRC-M1 and CCRC-M7 in our study, and the absence of CCRC-M7 antigens within the tube indicate that CCRC-M1 has not recognized RG-I but xyloglucan. Other authors have also proven the predominant xyloglucan affinity of CCRC-M1 (Sherrier & VandenBosch, 1994; Rodriguez-Galvez & Mendgen, 1995).
Furthermore, two pectins were successfully analysed: RG-I and homogalacturonans with a medium to high degree of esterification. These polysaccharides represent the bulk part of the matrix in most dicotyledonous plant cell walls (Carpita & Gibeaut, 1993; Schindler, 1993). Yet, even though the unaltered cell wall was consistently labelled by the respective antibodies, CCRC-M7 and JIM 7, we never found any labelling in the tube. Consequently, it has to be assumed that the pectin matrix is degraded there by the fungus. As one piece of evidence to support this point we could detect the presence of a polygalacturonase (PGase) in functioning but not in degenerated IAs using an antibody against a PGase isolated from Fusarium oxysporum f. sp. radicis-lycopersici (Blais, 1991; Charest et al., 2004). Apparently, this enzyme is not needed at this site once the interaction has come to an end. The slight anti-PGase labelling along the fungal wall may be explained by the intercellular spread of the pathogen. On its way through the plant tissue, the pectinaceous middle lamella has to be dissolved, otherwise C. trifolii would be restricted to naturally occurring air chambers.
The importance of pectinolytic enzymes in plant–pathogen interactions is rather controversial (Mann, 1962; Puhalla & Howell, 1975; Collmer & Keen, 1986; Durrands & Cooper, 1988; Scott-Craig et al., 1990; Annis & Goodwin, 1997; Baayen et al., 1997; Murdoch et al., 1999). For Claviceps purpurea, Tenberge et al. (1996) showed that PGase, probably in synergy with other enzymes such as pectin methylesterase, was responsible for complete breakdown of galacturonans in rye ovary cell walls. Applying a gene-replacement approach, Oeser et al. (2002) could demonstrate that the product of two endopolygalacturonase genes, which are expressed in all stages of infection, are crucial pathogenicity factors for the development of this fungus on rye ovaries. Claviceps purpurea is a biotrophic pathogen like C. trifolii. It might well be that in such fungi PGases are much more important than in necrotrophic pathogens. Consequently, the suggestion that ‘pectinase is not very important in fungal pathogenesis’ (Walton, 1994) cannot be sustained. The high homology of fungal PGases (Cervone et al., 1990; Bussink et al., 1992; Tenberge et al., 1996) might explain why an antibody generated against a Fusarium PGase worked so well in our system. Although we have examined only polygalacturonase, other enzymes such as pectin methylesterase may be significant as well. Further work is needed to clarify whether such enzymes are produced by C. trifolii.
Taken together, these data indicate that C. trifolii has developed a cunning strategy to gain nutrients from its host. Cellulose and xyloglucan, the main skeletal elements of the plant cell wall, are left intact, while the pectin matrix is broken down (Table 2) by a polygalacturonase and possibly other enzymes. This process is restricted to a wall penetrating tube which links the IA and the host bubble and probably functions as a channel for nutrient transfer from the host bubble into the IA. The result of such a differential degradation of the host cell wall would be a significant increase of wall porosity within the tube. Similar effects of pectinolytic enzymes have been demonstrated by Baron-Epel et al. (1988), Fujino & Itoh (1998) and McCann et al. (1990). Since nutrient molecules like oligopeptides or oligosaccharides are small enough to easily pass through the plant cell wall, it is unlikely that the assumed nutrient flow is much increased by larger pores. Conversely, fungal molecules triggering the formation of the host bubble or facilitating nutrient transfer across the bubble membrane may be required to travel through the host wall and into the bubble. In this respect, we have preliminary data indicating that a protein of C. trifolii becomes integrated into the bubble membrane.
Table 2. Presence or absence of the respective plant cell wall polymers within the tube
|Rhamnogalacturonan I/arabinogalactan proteins||CCRC-M7||–|
|Medium to highly esterified homogalacturonans||JIM 7||–|
To obtain nutrients from a host cell without disrupting its wall may decrease the efficacy of possible defence mechanisms such as a hypersensitive reaction or phytoalexin production. Our studies show that one plant cell can be attacked by more than one IA from the same or different hyphae (unpublished). There are no signs of degradation in the hyphal walls as, for example, in F. oxysporum f. sp. radicis-lycopersici in tomato roots (Benhamou et al., 1990). This indicates that no enzymatic or toxin-producing machinery is turned on after aggression towards the host cell has begun. Yet, although there is no apparent enzymatic fight-back on the host side, the plant cell does react. Wall appositional material is transported to the site of the interaction until the bubble is completely encased and finally collapses. Similar defence mechanisms have been reported in many plant–pathogen interactions, for example against Pseudomonas in bean (Brown & Mansfield, 1991), Xanthomonas in cassava (Boher et al., 1996), F. oxysporum in tomato (Charest et al., 1984) or cotton (Rodriguez-Galvez & Mendgen, 1995), Ophiostoma ulmi in elm trees (Rioux & Biggs, 1994), Glomus mosseae in a nonmycorrhizal pea mutant (Golotte et al., 1993) and Uromyces vigniae on broad bean (Xu & Mendgen, 1997). Many authors have shown that these deposits contain glycoproteins and/or PR-1 proteins (Brown & Mansfield, 1991; Golotte et al., 1995; Rodriguez-Galvez & Mendgen, 1995; Cordier et al., 1998) and callose (Golotte et al., 1993; Rodriguez-Galvez & Mendgen, 1995; Boher et al., 1996; Xu & Mendgen, 1997; Cordier et al., 1998). Rodriguez-Galvez & Mendgen (1995) have furthermore proven the occurrence of fucosylated xyloglucans along the ridges of wall deposits in cotton cells to ward off Fusarium, while their centre remained unlabelled. They also noted that labelling for callose seemed to be denser in more electron-translucent areas. In addition, they observed that the appositions contained pectic epitopes. These results are mostly mirrored by our observations on clover cells infected with C. trifolii. Contrary to Rodriguez-Galvez & Mendgen (1995), we could not see any labelling of wall appositions after incubation with JIM 5 or 7. Instead, CCRC-M7 consistently labelled such deposits. Therefore, the wall appositions in the system studied here contain at least callose, RG-I and XG, but no epitopes reactive with JIM 5 or JIM 7. This also means that the Rodriguez-Galvez & Mendgen's hypothesis (1995) stating that ‘the absence of … callose seems to characterize compatible biotrophic interactions’ cannot be corroborated.
In conclusion, we were able to demonstrate that in the interaction between C. trifolii and Trifolium repens the wall of the plant cell is partly but not completely degraded at a very localized area. This alteration very likely leads to an increased porosity. Although the reasoning behind this remains unclear at present, a possible explanation, the transfer of fungal proteins into the host bubble and its membrane, is under investigation. The interaction apparatus produced by the pathogen for nutrient uptake contains components characteristic of ascomycete walls and is therefore considered apoplastic, confirming ultrastructural observations made earlier (Simon et al., 2004). The wall appositions loaded onto the host bubble by the plant cell are made of at least callose, rhamnogalacturonan-I and xyloglucan and appear to end the interaction.