Evolutionary divergence of the insect disease-encoding Serratia plasmid pADAP : A thesis submitted in partial fulfilment of the requirements for the degree of Doctor of Philosophy at Lincoln University

The larval stage of the endemic New Zealand grass grub (Costelytra giveni) (Coleoptera: Scarabaeidae), causes estimated damage worth between 215 – 585M NZD annually to dairy and meat farms throughout New Zealand. To date, two diseases of grass-grub larvae, instigated by strains of Serratia entomophila and S. proteamaculans, have been identified and are being used as commercial biocontrol agents. As chemical pesticides are rapidly being banned due to hazardous side effect or soil contamination, together with documented increases of resistance to these pesticides, biological control approaches such as the Serratia based products are slowly taking their place. The two main virulence determinants in these Serratia strains, an Anti-feeding prophage (Afp) and an ABC-toxin-complex (Serratia entomophila pathogenicity toxin complex), are encoded on the previously described 153-kb conjugative megaplasmid pADAP (Amber Disease Associated Plasmid). Variants of the pADAP have been identified containing only one of the two virulence determinants and homologs have been found in bacteria from other genera. With the goal of defining evolutionary points of divergence between plasmid variants, and to examine potential co-evolution between plasmid and host, plasmids from 76 Serratia isolates and three Yersinia isolates with varying disease phenologies were sequenced. Phylogenetic analysis of the conserved pADAP plasmid “backbone”, residing between a conserved point of demarcation found in this study and the end of a sex pili cluster (spa1), revealed clustering of all the S. entomophila plasmids. Within the predicted backbone region, several intergenic regions in the areas of replication and conjugation contained DNA inserts, one of which, positioned between TraG and TraC, demarcates chronic disease related plasmids from the hypervirulent and non-pathogenic plasmids. These inserted regions, together with other insertions, deletions and mutations identified in this project, are clear markers for evolutionary divergence between the plasmid variants. In addition to virulence determinant variants, several novel gene clusters have been identified. Some of these include antimicrobials, putative accessory virulence determinants, toxin-antitoxin clusters and secretion systems, and two novel regions of unknown function. In this study, several of the novel regions were investigated to determine correlation between disease states in C. giveni and the occurrence of these genes, but none of the regions were found to influence amber disease in C. giveni, suggesting these strains might be associated with other hosts or provide the bacteria with other competitive advantages. Of note is a bacteriophage present on a pADAP-type plasmid, that is also present in a non-pADAP type plasmid, both obtained from S. proteamaculans strains sequenced in this project. The non-pADAP phage carrying strain also has an orthologous Sep region. This phage and Sep carrying S. proteamaculans isolate was ineffective towards C. giveni larvae but had significant mortality response in Pyronata festiva, a beetle species morphologically similar to grass grub, but genetically very distinct. This is an interesting finding, as most Sep carrying strains without Afp have the opposite phenotype in that they are only bioactive in C. giveni and not P. festiva. Numerous Sep variants were observed, including several variants residing on non-pADAP plasmids (i.e. without the conserved backbone). All but one of the Sep variants were co-located with the Sef fimbrial cluster, which was highly conserved compared to the Sep cluster. The one non-Sef encoding plasmid was associated with a Yersinia frederiksenii isolate and appeared to have no bioactivity in C. giveni. The co-location of the Sef cluster, the high conservation of the sef genes and the absence of plasmids lacking the Sef region while still being bioactive, leads to the assumption that fimbriae are an integral part to the functionality of Sep, through a yet to be determined mechanism. The hypothesis of the study was that there would be a tightly regulated network of HGT mechanisms that made pADAP a clonal plasmid capturing device with actively interchangeable pathogenicity associated clusters. It was assumed virulence determinants can be lost in situations where the bacterium and its plasmid were not in contact with the insect host, or that the plasmid could be discarded in periods of low nutrients, such as the summer period in New Zealand where grass grub larvael populations are at their lowest and nutritional stress and heat stress are at their highest. However, the exact opposite of this hypothesis was observed. The plasmids analyzed appear to be highly stable and loss of plasmid happens rarely even under the hashest conditions (extreme heat, extreme stress, nutritional depletion). Numerous analysed isolates with peculiar phenotypes thought to have a pADAP harboured non-pADAP megaplasmids of similar size. This study also showed that pADAP plasmids do not seem to be a burden to their host cell, but instead actually convey some fitness benefits. Instead of observing active horizontal gene transfer (HGT) of pathogenicity clusters, it was observed that the genetic diversity is much more correlated with speciation. The overall conclusion of this study is that perturbation of the bacterium system is more likely caused by genetic diversity than an tightly regulated active HGT mechanisms.