Identification and functional characterization of proteases and protease inhibitors involved in virulence of fungal tomato pathogens

Pathogens cause disease on both animal and plant hosts. For successful infection and establishment of disease, pathogens need proper weaponry to protect themselves against host defenses and to promote host colonization to facilitate uptake of nutrients for growth and reproduction. Indeed, plant pathogens secrete various types of effector molecules (proteins and secondary metabolites) to manipulate host responses for their own needs. Secreted proteases and protease inhibitors (PIs) are such effector molecules. Proteases can hydrolyze plant defense proteins and PIs can inhibit plant proteases that are part of the host surveillance system. Despite the importance of proteases and PIs secreted by fungal pathogens, little information about their role in virulence is available. The recent advances in genomics, bioinformatics, transcriptomics and proteomics have facilitated identification and functional analysis of proteases and PIs relevant to plant-fungus interactions. <strong>Chapter 1</strong> is an introduction to the thesis outlining the general concept of plant-microbe interactions. It briefly describes the current knowledge of pathogenicity mechanisms employed by fungal plant pathogens and defense mechanisms employed by their host plants. It further introduces proteases and PIs and their potential role in modifying pathogenesis-related (PR) proteins to facilitate fungal virulence. It completes with an outline of the PhD research project. In <strong>chapter 2</strong>, we analyzed and compared the number of putatively secreted proteases present in the genomes of 30 fungi with different lifestyles. The analysis showed that fungi with a saprotrophic and hemibiotrophic lifestyle contain more secreted protease genes than biotrophs. Surprisingly, the number of protease genes present in the genome of <em>Cladosporium fulvum</em>, a biotrophic tomato pathogen, is comparable with that of hemibiotrophs and saprotrophs. We analyzed all <em>C. fulvum</em> protease genes both at the transcriptome and proteome level by means of RNA-Seq/RT-qrtPCR and mass spectrometry analyses, respectively. Results showed that many proteases of <em>C. fulvum</em> are not expressed during growth <em>in planta</em>, likely sustaining the biotrophic growth pattern of this fungus. In <strong>chapter 3, </strong>using an alignment-based gene prediction tool, we identified pseudogenes containing disruptive mutations (DMs) that likely lead to the production of nonfunctional proteins, including a group of putatively secreted proteases from <em>C. fulvum.</em> Fewer DMs were observed in other fungi including <em>Dothistroma septosporum</em>, a hemibiotrophic pine needle pathogen and close relative of <em>C. fulvum,</em> and suggested that the difference in pseudogenization of proteases between these two pathogens might in part explain their different lifestyle. In <strong>chapter 4, </strong>we analyzed the tomato genome and identified 30 candidate chitinases genes, of which six encoded chitin binding domain (CBD)-containing chitinases. Transcriptome and proteome data were collected after inoculation of tomato with several fungal pathogens and allowed the identification of two CBD-chitinases (SlChi2 and SlChi13) with a putative role in protecting tomato against <em>C. fulvum</em> and<em> F. oxysporum</em> f. sp. <em>lycopersici</em> (<em>F. oxysporum</em>), respectively. Purified CBD-chitinases SlChi1, SlChi2, SlChi4 and SlChi13 were incubated with secreted protein extracts (SPEs) from seven fungal tomato pathogens and we could show that SPEs from <em>F. oxysporum</em>, <em>Verticillium dahliae</em>, and <em>Botrytis cinerea</em> modified SlChi1 and SlChi13. LC-MS/MS analysis revealed that incubation with SPE from <em>F. oxysporum </em>removed the N-terminal 37 and 49 amino acids, comprising part and complete CBD domain from SlChi1 and SlChi13, respectively. Removal of the CBD of SlChi1 and SlChi13 by SPE of <em>F. oxysporum</em> reduced the antifungal activity of the two chitinases. We identified a fungal metalloprotease (FoMep1) and a subtilisin serine protease (FoSep1) that synergistically cleaved both SlChi1 and SlChi13. Transgenic <em>F. oxysporum</em> in which the genes encoding these two proteases were knocked out by homologous recombination lost the ability to cleave the two chitinases and were compromised in virulence on tomato compared to the parental wild type. These results suggest an important role of the two chitinases in defense of tomato against this pathogen. In <strong>chapter 5, </strong>we searched for host target(s) of the apoplastic effector Avr9 secreted by <em>C. fulvum</em> during infection of tomato. Based on the structural homology of Avr9 with carboxy peptidase inhibitors, we hypothesized that the host target of Avr9 might be apoplastic proteases. To isolate and identify Avr9 targets in apoplastic fluids, we used synthetic biotinylated Avr9, and performed pull-down and far-western blotting assays with apoplastic fluids from tomato inoculated with a <em>C. fulvum</em> race lacking the <em>Avr9</em> gene. However, we found no specific Avr9-interacting proteins from pull-down complexes analyzed by mass spectrometry or by far-western blotting. Then, we hypothesized that glycosylation of Avr9 might be required for its biological function. The results of mass spectrometry analysis revealed that Avr9 is N-glycosylated when secreted by <em>C. fulvum</em>, containing at least two GlcNac and six mannose residues. The necrosis-inducing activity of glycosylated and non-glycosylated Avr9 was assayed but appeared not significantly different; however, we could not produce sufficient amounts of (biotinylated)-glycosylated Avr9 to perform pull-down assays for identification of potential glycosylated Arv9-interacting proteins by mass spectrometry. Previous studies as well as the results present in this PhD thesis showed that fungal pathogens secrete a plethora of effectors including proteases and PIs. Many of identified proteases and PIs mediate effector-triggered immunity in host plants. In <strong>chapter 6, </strong>we reviewed the recent advances on the various roles of proteases and PIs in compromising basal defense responses induced by microbe-associated molecular patterns. <strong>Chapter 7</strong> is a summarizing discussion of the PhD thesis. We showed determinative roles of proteases and PIs in shaping plant-pathogen interactions. The expression and pseudogenization studies on proteases of <em>C. fulvum</em> showed that the genome content does not necessarily reflect the lifestyle of this fungus. This is true for many classes of fungal genes, including proteases. Fungi contain many different types of proteases whose functions may partly overlap. This hampers the discovery of their biological functions. We could demonstrate that two different types of proteases (metalloprotease (FoMep1) and subtilisin serine protease (FoSep1)) of <em>F. oxysporum</em> act synergistically to modify and reduce antifungal activity of two plant CBD-chitinases. Identifying additional proteases is achievable by a targeted proteomics approach using known targets as we did in chapter 4. However, identification of biological functions of proteases is a technical challenge when targets are not known. Multi-gene targeting of protease and PI genes is required to reveal their function in plant-pathogen interactions, which can only be addressed by using advanced genetic tools in future research.