Study of genetic determinants involved in Salmonella Derby host adaptation

Salmonella is the second most commonly cause of food-borne disease in Europe. The majority of Salmonellae causing disease in human and domestic animals belong to S. enterica subspecies enterica, which is divided in approximately 1,500 serovars, based on their unique somatic and flagellar antigenic formulas. Serovars can exhibit different host-tropism: generalist serovars are able to infect a variety of unrelated hosts, host adapted serovars are mainly isolated from a specific host and specialists serovars are restricted to a single host. Epidemiological evidence supports various levels of host adaptation even among isolates belonging to the same serovar. A number of genetic features have been consistently associated to serovars with a specific host range, but the exact mechanism of host adaptation remains still elusive. Discovering the genetic determinants involved in the ability to infect specific hosts is critically important to assess the risk of pathogen transmission along the food-chain. Data from the Emilia-Romagna Reference Centre for Enteropathogens Surveillance show that Salmonella serovar Derby is mainly prevalent in swine (28.2%) while it is only rarely found in human (2.6%). S. Derby isolates were genotyped by Pulsed-Field Gel Electrophoresis (PFGE). Different proportion of human isolates were observed for the two most prevalent PFGE profiles in swine: SXB_BS.0204 is significantly less isolated in human (0.9%) than in swine (9.4%) while SXB_BS.0056 is proportionally isolated in human (13%) as well as in swine (9.7%). Starting from these data, the aim of this thesis was to identify the genetic determinants responsible for different distribution in human of the two PFGE profiles. It was evaluated if the different prevalences in human of SXB_BS.0204 and SXB_BS.0056 isolates were due to disparity in virulence phenotype. C. elegans infection assays did not detect any difference between isolates belonging to distinct PFGE profiles in the ability to kill the nematode. Invasion and replication assays in human INT-407 and swine IPEC-J2 epithelial cell lines were thus performed. The invasion and replication efficiencies in human cells of SXB_BS.0056 isolates are up to 4 logs higher than those of SXB_BS.0204 isolates. In swine cells SXB_BS.0056 isolates are more virulent than SXB_BS.0204 isolates, but differences are significantly reduced to just one log. Isolates were thus defined virulent (V) and non-virulent (NV) based on their infection efficiency. Whole genomes of 45 isolates were sequenced and core single nucleotide polymorphisms (SNPs) were used as input for generating a phylogenetic tree. V and NV isolates cluster separately, suggesting different evolutionary paths. It was assessed if exclusive genes of V isolates were responsible for their virulence phenotype. Mutants deleted for plasmids and genomic regions found only in V isolates showed no changes in infection efficiency compared to the respective wild types. To evaluate the role of different gene allelic variants in virulence, non-synonymous SNPs were selected which discriminate V from NV isolates. Two allelic variants of hilD were detected: HilD is the activator of Salmonella Pathogenic Island (SPI) 1, which encodes for the Type 3 Secretion System (T3SS), responsible for invasion in epithelial cells. In human cells a V isolate (ER1175 WT) carrying the hilD NV allele (hilD_nv) has the same low infection efficiency of NV isolates. Accordingly, a NV isolate (ER278 WT) carrying the V hilD allele (ER278::hilD_v) shows the same high infection efficiency of V isolates. Both strains deleted individually for hilD and invA (encoding for an essential component of T3SS) infect human cells at the same level of NV isolates. In swine cells the introduction of hilD_nv as well as the hilD deletion in ER1175 WT does not cause any changes in the virulence phenotype, whereas the invA deletion triggers a decrease in virulence. No differences were found among ER278 WT, ER278ΔhilD and ER278ΔinvA in swine cells, but hilD_v insertion causes an increase in both invasion and replication compared to ER278 WT. The two hilD allelic variants thus explain virulence differences between V and NV isolates in human cells, but not in swine cells. To assess differences in SPI-1 genes expression caused by hilD allelic variants, a RNA-seq-based transcriptomic analysis was performed extracting RNA in in vitro conditions causing SPI-1 upregulation. There are no differences in genes expression between strains carrying hilD_nv and the respective hilD-deleted strains, whereas SPI-1 genes are upregulated in both ER1175 and ER278 carrying hilD_v. These results prove that hilD_nv produces a non-functional protein. The same results were obtained analyzing the expression of SPI-1 genes in ER1175 WT, ER1175::hilD_nv and ER1175ΔhilD during human cells infection. These 3 strains and ER278 WT were tested in a mouse model of Salmonella infection to assess if hilD variants cause differences in in vivo host-bacterial interaction. No differences among strains were found in cecum bacterial loads as well as in the expression of host genes correlated with inflammation (NOS2 and KC), not significantly different from uninfected mice. Strains seem thus to not induce gut inflammation in mice. No differences in dissemination extent were observed among ER278 WT, ER1175 WT and ER1175::hilD_nv. ER1175ΔhilD accumulates in liver and spleen at higher level than ER1175 WT, suggesting a negative effect of HilD on S. Derby dissemination. Together, this study has provided new information about mechanisms of host adaptation adopted by S. Derby.