A GENOME-WIDE APPROACH TO PLANT-HOST PATHOGENICITY IN XYLELLA FASTIDIOSA: MULTIGENIC METHODS FOR IDENTIFYING STRAINS, FOR STUDYING THE ROLE OF INTER-STRAIN RECOMBINATION, AND FOR IDENTIFYING PATHOGENICITY CANDIDATE GENES Project Leader:

We have developed a multilocus sequence typing (MLST) system for identifying the known pathovars of Xylella fastidiosa (including subsp. fastidiosa that causes Pierce’s disease). This identification system is based on allelic variation at seven housekeeping genes (holC, nuoL, leuA, gltT, cysG, petC and lacF) and a public MLST database has been established at www.mlst.net (see Scally et al. 2005). This easily accessible system will allow for the rapid recognition of novel variants as they arise, since it can be continuously updated by researchers who have sequenced new isolates for the seven genes. We tested the effectiveness of the MLST system using 25 isolates of Xylella fastidiosa (Xf) from five different host plants: grapevine (Pierce’s disease, PD), oleander (oleander leaf scorch, OLS), oak (oak leaf scorch, OAK), almond (almond leaf scorch, ALS) and peach (phony peach scorch, PP). An eBURST analysis identified six clonal complexes (CCs), using the grouping criterion that each member of a CC must be identical to at least one other member at five or more of the seven loci. These clonal complexes corresponded to phylogenetic clades that we had previously identified (Schuenzel et al. 2005), including subspecific clades of fastidiosa and sandyi (CC1 & 2), while CC3-5 defined host-specific sub-clades of the subsp. multiplex. CC6 identified a multiplex-like group characterized by a high frequency of inter-subspecific recombination. To begin to understand the potential role of recombination in the creation of new pathovars, we also used the MLST data (plus three additional loci) to estimate the relative contribution of recombination and mutation to the observed variability. Recombination between different alleles was estimated to give rise to 76% of the nucleotide changes and 31% of the allelic changes observed. However, sequence data also suggests that inter-subspecific recombination has started relatively recently. This new phenomenon may lead to an increased rate of pathovar formation. INTRODUCTION The availability of four genome sequences of Xf (two completely annotated and two non-annotated), allows us to exploit the extraordinary power of genomic research to investigate Xf’s genetic diversity. This diversity can provide information essential for understanding the plant-host specificity of the Xf subspecies. So far, the only form implicated in causing PD is subspecies fastidiosa, however a very real possibility exists that new pathovars may arise by recombination among the three North American subspecies. Sequence data from the Xf genomes suggests that recombination was historically rare; however this has never been quantified. We have shown that the PD pathovar (subsp. fastidiosa) has very low sequence variability (Schuenzel et al. 2005). This suggests that the PD pathovar has been subject to intense selectiona result probably reflecting significant genetic constraint imposed by the grapevine host on the bacterium. Identifying this constraint is likely to lead to a mechanism for pathogen control. Our first priority was to place the PD strain within a statistically robust phylogeny, extending earlier work defining the interrelationships of the plant-host strains of Xf (e.g. Hendson et al. 2001; Lin et al. 2005). Schaad et al. (2004) identified two North American subspecies based on DNA hybridization: subsp. multiplex, found on a range of hosts including almond, peach and plum, plus the PD pathovar, subspecies fastidiosa (initially named piercei). Using DNA sequence data, we added a third North American subsp. (sandyi) isolated originally from oleander and we estimated that these three subspecies have been separated for more than 15,000 years (Schuenzel et al. 2005). Given a robust phylogeny, our challenge was to develop an effective method for identifying the known host pathovars. The “state-of-the-art” approach is to use MLST (multiple locus sequence typing) (Maiden et al. 1998). This technique has been applied primarily in the identification of pathovars of human pathogens, and a public database has been established at www.mlst.net, which is located at Imperial College, London and is funded by the Wellcome Trust. Unambiguous identification of strains is of considerable importance for understanding the epidemiology of PD and the other plant diseases caused by this bacterium. Previously, this has been approached using a variety of DNA based methods (Banks et al. 1999; Hendson et at. 2001; Rodrigues et at. 2003; Meinhardt et al. 2003; Lin et al. 2005). As yet, it has not been established if results from these methods have a clear relationship to the true underlying phylogenetic relationships. The simple sequence repeat (SSR) approach of Lin et al. (2005) shows up high variability, useful for uniquely distinguishing isolates, however it relies on many (34) loci. In contrast, MLST methods rely on the allelic variability of just seven