To feed the increasing world population, agricultural production needs continuous improvement. Especially protection of crops from disastrous diseases is crucial. The interaction between the pathogenic fungus Cladosporium fulvum and its host, tomato, serves as a model system for plant-pathogen interactions. Some tomato plants carry resistance ( R ) genes that confer recognition of fungal strains carrying complementary avirulence ( Avr ) genes. A number of these R genes have been cloned, as well as their complementary Avr genes. The aim of the research described in this thesis was to examine how R gene products confer recognition of fungal strains carrying the matching Avr genes. Profound understanding of the molecular basis of this interaction might help us to improve the protection of other crop plants against economically important diseases.
Chapter 1 introduces the state of the art on interaction between Cladosporium fulvum and tomato at the time the research described in this thesis was initiated. C. fulvum is a specialised, biotrophic pathogen, causing tomato leaf mold. The fungus infects tomato leaves by entering stomata at the lower side of the leaf. The infection will proceed if no resistance R genes of the plant match any of the Avr genes of the fungus. However, the plant recognises the fungus when it carries an R gene that matches an Avr gene present in the fungus. This recognition results in the induction of plant defence responses, including a rapid death of cells surrounding the infection site, called the hypersensitive response (HR). Further fungal growth is prohibited by these defence responses. During its lifecycle on susceptible plants, C. fulvum is restricted to the extracellular space of the tomato leaves and secretes many proteins that potentially play a role in virulence. Also the elicitor proteins encoded by the Avr9 and Avr4 are secreted. Injection of these proteins is sufficient to trigger HR in tomato plants carrying Cf-9 and Cf-4 resistance genes, respectively. Both AVR9 and AVR4 are small, stable, cysteine-rich proteins. The complementary Cf-9 and Cf-4 genes encode highly similar, membrane-anchored, receptor-like proteins with extracytoplasmic leucine-rich repeats (LRRs) and a short cytoplasmic tail. Differences between Cf-9 and Cf-4 proteins are located in the N-terminal half, predominantly in amino acid residues at putative solvent-exposed positions of the LRRs, which is thought to form the 'recognition surface' of these proteins.
To examine the role of the various domains of Cf proteins in perception of AVR proteins of C. fulvum in more detail, a functional, transient expression system was developed for the Cf-4 and Cf-9 resistance genes ( chapter 2 ). This expression system is based on infiltration of tobacco leaves with Agrobacterium strains that carry Cf genes on the T-DNA of binary plasmids (agroinfiltration). The AVR proteins are delivered either by injection, agroinfiltration, Potato Virus X-mediated expression or by using Avr -transgenic tobacco plants. This chapter also describes differences between Avr9/Cf-9 - and Avr4/Cf-4 -induced necrosis, which are mainly due to a difference in Avr gene activity upon expression in the plant. Finally, it is shown that the signal transduction pathway leading to HR is conserved in solanaceous plants, but likely not in non-solanaceous plant species. An exception is the non-solanaceous plant lettuce, in which the Avr4/Cf-4 gene pair is functional.
The agroinfiltration assay is an excellent expression system to study the effect of mutations in Cf genes. In chapter 3 , agroinfiltration was used to determine specificity determinants in Cf proteins by exchanging domains between Cf-4 and Cf-9 and subsequently examining the effect of these mutations on specificity of perception of AVR proteins. Cf-4 differs from Cf-9 at 67 amino acid positions and also contains three deletions. Significantly, Cf-4 lacks two LRRs compared to Cf-9, which appears essential for Cf-4 function. The two additional LRRs in Cf-9 are required for Cf-9 function. Specificity determinants in Cf-4 reside not only in the LRR domain but also in the B-domain. In contrast, specificity determinants in Cf-9 reside entirely in the LRR domain and are likely scattered throughout this domain. The specificity determinants in the LRRs of Cf-4 cluster in a few adjacent LRRs and reside in only three amino acid residues at putative solvent-exposed positions. Thus, most of the 67 amino acids that vary between Cf-4 and Cf-9 appear not to be required for specificity, but probably serve as a source to generate new specificities.
To learn more about specificity determinants of Cf-9 proteins occurring in natural populations, we examined the molecular variation of Cf-9 in Lycopersicon pimpinellifolium (Lp) , from which the Cf-9 locus has been introgressed into cultivated tomato ( chapter 4 ). It appears that AVR9 recognition occurs frequently throughout the Lp population. In addition to Cf-9 , a second gene, designated 9DC , confers AVR9 recognition in Lp . Compared to Cf-9 , 9DC is more polymorphic, occurs more frequently and is more widely spread throughout the Lp population, suggesting that 9DC is older than Cf-9 . The second half of the 9DC gene is nearly identical to the second half of Cf-9 , whereas the first half is nearly identical to Hcr9-9D , a Cf homolog adjacent to Cf-9 at the Cf-9 locus. This suggests that Cf-9 has evolved by intragenic recombination between 9DC and another Cf homolog. The fact that 9DC and Cf-9 proteins both confer recognition of AVR9 but differ in 61 amino acid residues shows that Hcr9 proteins can be highly variable, without affecting their recognitional specificity.
After having examined their specificity determinants, we subsequently focused on the cellular location of Cf proteins. The presence of a dilysine motif in the G-domain of Cf-9 ( KK RY) suggests that the protein resides in the endoplasmic reticulum (ER) instead of the plasma membrane (PM). Previously, two conflicting reports on the subcellular location of Cf-9 were published. One report showed that Cf-9 accumulates in the ER and is absent in the plasma membrane, whereas the other showed that Cf-9 resides in the plasma membrane. In chapter 5 we have mutated the dilysine motif and show that the mutant Cf-9 protein remains functional in AVR9 recognition and mediation of HR. The data presented in this chapter, in combination with the two previous reports on Cf-9 localisation, can be explained by assuming that proteins that interact with Cf-9 mask the dilysine motif. This theory suggests that functional Cf-9 protein resides in small quantities in the plasma membrane, where it mediates recognition of the extracellular AVR9 protein as a component of a receptor complex.
AVR9 recognition in tomato plants carrying Cf-9 most likely involves the high-affinity binding site (HABS) for AVR9 that was identified in plasma membranes. However, the HABS is not encoded by Cf-9 because it is also present in tomato plants that lack Cf-9 and in many other plant species. As it is likely that both the HABS and the Cf-9 protein reside in the plasma membrane and may be present in the same receptor complex, it is essential to isolate the HABS in order to get more insight in the molecular mechanism of AVR9 perception. In chapter 6 , a procedure is described that allows solubilisation of the HABS without affecting its AVR9-binding activity. Of the 19 detergents that were tested, only octyl glucoside appeared to be suitable for solubilisation of the HABS. Removal of the detergent is crucial in this procedure, as it interferes with AVR9 binding. The described procedure may become an essential tool to study the AVR9 receptor complex at the biochemical level.
In the final chapter ( chapter 7 ), the experimental data presented in the previous chapters are discussed. In addition to AVR9/Cf-9 there are many other examples of gene-for-gene interactions where no direct interaction was found between R and Avr gene products. In many cases, there are indications for the involvement of an additional host protein, which may represent the virulence target of the Avr protein. The prevalence of R proteins that 'guard' virulence targets can be explained by natural selection for R genes that are maintained in the plant population through 'trench-warfare', resulting in recognition events that cannot be circumvented by the pathogen without taking a virulence penalty. The 'guard' hypothesis significantly changes the focus of current research to the role of virulence targets of Avr proteins, and might explain absence of functionality of R genes in heterologous plant species, despite the fact that they belong to conserved gene families.