UC Berkeley

Fungal plant pathogens have major impacts on agriculture and global food security and are likely to have an even greater impact in the future. The current tools that we have available to combat them are insufficient, in part because these fungi can quickly adapt to these tools. Understanding fungal plant pathogen evolution is therefore essential to curbing the threat these pathogens pose. In Chapter 1, I describe the motivations for my dissertation work and the state of the field of fungal plant pathogen evolution. I also introduce the model organism I used in my research, Magnaporthe oryzae, which causes the blast disease. Chapter 2 describes my characterization of the extrachromosomal circular DNAs (eccDNAs) of M. oryzae. EccDNAs are a diverse class of molecules that can contribute to phenotypic and genotypic plasticity in eukaryotes, and I hypothesized that these may be involved in fungal plant pathogen evolution. I show that M. oryzae has a more diverse set of eccDNAs than other organisms and that these are enriched in LTR retrotransposons. I also show that many genes are found on eccDNAs in M. oryzae, and that effectors are enriched on eccDNAs. Finally, I show that eccDNAs are associated with gene presence-absence variation (PAV). Next, in Chapter 3, I discuss in greater detail the results presented in Chapter 2, as well as their implications and potential future directions. I also further discuss evidence in Chapter 2 that led me to believe that eccDNAs do not play a major role in fungal plant pathogen evolution and led me to focus directly on gene PAV in M. oryzae in the remainder of my dissertation. Subsequently, in Chapter 4, I describe my characterization of these events in M. oryzae. I find that genes experiencing PAV between lineages of M. oryzae are enriched in disease-causing and non-self-recognition genes. I describe how gene PAV events in the rice and wheat pathotypes show clear differences in their count and genomic location. Through comparing PAV genes to conserved genes, I show that these had distinct distances to TEs, distances to other genes, lengths, GC content, expression, and epigenetic marks. I also describe how a machine learning model can be trained to take advantage of these features to predict genes prone to PAV in the M. oryzae genome. Finally, in Chapter 5, I further discuss the implications of the results I describe in this dissertation, as well as future directions for implementing machine learning models and general knowledge of fungal plant pathogen evolution to help guide rational disease resistance engineering in crops.