Genomic Comparison of the Closely Related Salmonella enterica Serovars Enteritidis and Dublin

1. INTRODUCTION

Infection with non-typhoidal Salmonella enterica is a major cause of food-borne disease in humans worldwide [1-3]. Animals and their products are regarded as the main sources of this pathogen, although it may also be present in other potential sources, such as fresh vegetables [4-6]. From over 2500 different serovars of Salmonella enterica (defined by their surface antigenic properties, both somatic O antigen and flagellar H antigens) about 50 are significant pathogens of animals and humans. Acute infections in humans can develop in one of four ways: enteric fever, gastroenteritis, bacteremia, or extraintestinal focal infection [7]. As with other infectious diseases, the course and outcome of the infection depend on a variety of factors, including the inoculating dose, the immune status of the host, and the genetic background of both the host and the infecting organism.

Although S. enterica serovars are genetically very similar, they differ significantly in host range and disease spectrum. S. enterica serovars may be classified as ubiquitous, host-restricted or host-specific. Ubiquitous serovars, which include Typhimurium and Enteritidis, most commonly produce self-limiting gastrointestinal infections in a wide range of hosts. Host-specific serovars, such as Typhi in humans or Gallinarum in fowl, cause severe systemic diseases in their specific hosts. A few Salmonella serovars, such as Choleraesuis and Dublin, have a narrow host range and are classified as host-restricted [8].

Host-restricted and host-specific serovars are generally more prone to cause invasive disease than ubiquitous serovars [9, 10]. Globally, human extra-intestinal salmonellosis is generally associated with those serovars that are also associated with gastroenteritis, as is the case with S. Enteritidis and S. Typhimurium. However, certain serovars are more prone to cause invasive infections than others, as is clear when the percentage of isolates from bacteremia related to total cases (invasive index) is calculated [7, 11]. For S. Typhimurium and S. Enteritidis, the invasive index ranges from 1 to 7% [11, 12], while for S. Dublin different reports indicate that the invasive index ranges from 50% to 70% [7, 11, 13-15]. Loss of gene function through pseudogene accumulation has been indicated as a hallmark of host-specific pathogenic bacteria as compared to their host-generalist relatives [16-22].

The Enteritidis (O: 1, 9, 12: gm: -) and Dublin (O:1, 9, 12: gp: -) serovars share antigenic properties and are phylogenetically closely related, yet they seem to differ significantly in pathogenic potential [23, 24]. S. Enteritidis commonly causes gastroenteritis but rarely causes invasive disease in humans. S. Dublin usually infects cattle causing abortion and systemic infection, but occasionally can be found infecting other hosts such as pigs and humans. On the rare occasions when it infects humans it often results in bacteraemia with severe disease and high mortality [25-27]. Characterization of the mechanisms underlying these differences is central to a more general understanding of the invasiveness of salmonellae. To date only one complete genome of a S. Enteritidis strain (P125109, hereafter referred as PT4) and two S. Dublin isolates (CT_02021853 and 3246) have been sequenced and annotated and are publicly available [28], [http://www.ncbi.nlm.nih.gov/genomeprj/19467] [29].

To gain new insights into genetic differences that could help to understand the basis of such marked different pathogenic behaviors, here we describe a comparative study between S. Enteritidis and S. Dublin. We conducted microarray-based comparative genomics between four S. Dublin clinical isolates and the core genome resulting of the comparative genome analysis of 29 S. Enteritidis isolates previously reported by us [30]. Further the pseudogene content of each serovar was also evaluated using the available genome sequences.

2. MATERIALS AND METHODS

3. RESULTS

DISCUSSION

S. Enteritidis and S. Dublin are two closely related serovar with marked differences in pathogenic traits and epidemiological behavior, thus it is reasonable to assume that genomic comparison between them could shed some light on the molecular basis of these differences. A single previous report described a microarray-based genome comparison [38], and here we conducted a similar analysis using a different set of field isolates and microarray chip. Further, we now report a comparison of the full genome sequences of S. Enteritidis and S. Dublin particularly looking at differences in pseudogene composition between them.

Our comparative genome hybridization study predicted 33 genes specific to S. Enteritidis and 87 specific to S. Dublin. The analysis revealed four genetic regions and seven single genes that seem to be exclusive of S. Enteritidis core genome, as well as six regions and six single genes specific for S. Dublin. These results corroborate and extend the previous report where 3 S. Dublin and 24 S. Enteritidis strains where compared [38]. This report described the same four regions specific for Enteritidis but only one of the six S. Dublin regions found by us. This particular region, that we denominated Du3, corresponds to regions B24, B25_a and B25_b as of the earlier report. Region Du3 corresponds to genes found in SPI-6 from S. Typhi CT18. This region encodes a ClpB heat-shock protease-like protein, as well as different membrane proteins and lipoproteins that belong to the T6SS encoded in this island. Interestingly, this region includes a gene in the rhs family (STY0321) that has no homologue in the CT_02021853 genome sequence.

Among the other regions specific for S. Dublin described here, Region Du1 was recently proposed to be a pathogenicity island (SPI-19) identified in S. Gallinarum, S. Dublin, S. Weltevreden and S. Agona that encodes a type-6 secretion system (T6SS) [39]. In S. Enteritidis, an internal deletion has eliminated most of the island. Region Du2, includes various bacteriophage regulatory proteins, recombinases, transposases, and structural proteins. It also includes one gene (SG1186) previously annotated as encoding a putative phage-encoded cell division inhibitor protein belonging to the kil super-family and associated with the capacity to inhibit the essential ftsZ cell-division gene [40]. ftsZ expression is altered during the intracellular phase of infection with S. enterica, a process that is independent of sulA, a known inhibitor of ftsZ [41]. Genes encoding proteins belonging to the same super-family are also present in several S. Typhi genome sequences, as well as in other enterobacteria (e.g. different STEC strains, Shigella flexneri, Shigella dysenteriae and others) as revealed by Blast-p analysis, suggesting a possible role for these proteins in pathogenesis. Regions Du1 and Du2 were not represented in the microarray used by Porwollik and collaborators [38], thus we cannot exclude that these regions were also present in the strains studied there, but simply not found because of the particular microarray used. Regions Du4, Du5 and Du6, correspond to prophages found in the genome sequence of S. Typhi CT18 [16]. Region Du4 comprises 17 genes from a lambdoid bacteriophage that include several CDS encoding for DNA binding proteins. Region Du5 includes 3 genes that are part of a degenerate bacteriophage; one of these (STY2044) encodes a putative endolysin similar to several lysozymes from E. coli and Shigella strains. Region Du6 spans 10 genes including a DNA adenine methylase (STY3667), regulatory proteins and endonucleases. These 3 regions of difference were not found in the earlier report, despite the CDS been present in the microarray. Instead, that work reported differences in other prophage-derived genes. Thus, it could that the genomes of the particular set of strains used in both studies posses different prophage composition. The analysis of Du4-Du5-Du6 in both S. Dublin sequenced isolates, revealed that regions Du5 and Du6 are very conserved in both strains whereas region Du4 is almost complete in CT_ 02021853 but incomplete and less conserved in strain 3246, supporting the hypothesis of different content in phage genes among S. Dublin isolates.

Among the seven single genes that are for the first time described here as absent in S. Dublin strains, safA (SEN0281) and dcp (SEN1539) are of special interest. safA is the first gene of the saf fimbrial operon and encodes a lipoprotein. The operon forms part of the degraded pathogenicity island SPI-6 in the S. Enteritidis chromosome. This operon is not annotated in the S. Dublin genome sequences available. However, Blast analysis revealed that this is a region highly conserved at a nucleotide level between PT4 and both S. Dublin sequenced isolates. There are several stop codons in the S. Dublin sequence homologous to safA, suggesting that this gene is in process of degradation. The fact that we cannot detect safA by CGH in the S. Dublin Uruguayan isolates may be related with this. The dcp gene encodes for dipeptidyl-carboxypeptidase II, which is highly conserved among the Enterobacteriaceae. This gene has been described previously as a frequent site for SNPs in S. Enteritidis [42], and it is absent from the CT_02021853 sequence.

Overall, the CGH analyses did not detect clear differences in genes that have been previously reported as required for virulence to explain the differences in pathogenicity of both serovars. However, the presence/absence of a gene, as detected by this methodology, does not inform about its expression, thus these results should be interpreted with caution.

The high number of pseudogenes detected in CT_02021853 suggests that this mechanism might be relevant in the process of host adaptation of this serovar, as well as in the different epidemiological and pathogenic behavior of S. Dublin when compared with S. Enteritidis. As we describe in Table 5, we observed a differential distribution of functionality amongst the CDS inactive in S. Enteritidis and S. Dublin. More than 40% of the pseudogenes specific for S. Enteritidis correspond to CDS related to phages or transposases but only 12% with those involved in metabolism and regulatory proteins. Conversely, among the pseudogenes specific for S. Dublin 33% correspond to CDS encoding proteins involved in central metabolism or regulatory proteins and 37% to CDS related to surface structures but only 3% to phages and transposases. These observations may be relevant to understand the host restriction of S. Dublin.

We found 21 CDS that appear to be active genes in the broad host-range S. Enteritidis but pseudogenes in the host-restricted S. Dublin and in the host-specific S. Gallinarum. From this set of CDS, 9 are pseudogenes as well in the other host-restricted serovar S. Cholerasuis suggesting that their inactivation could be relevant as genetic determinants of host adaptation. These nine CDS correspond to two hypothetical proteins (SEN0784 and SEN2783), one putative transport protein (SEN0042), the gene encoding the outer membrane usher protein LpfC (SEN3461), one probable phosphotransferase system permease (SEN3672), one gene encoding a putative Type I restriction modification system protein (SEN4290), and the gene encoding a probable glucarate dehydratase 2 (SEN2806 or ygcY). The other two genes that complete this list are mglA (SEN2182) and shdA (SEN2493), which are pseudogenes in S. Typhi CT18 and Ty2 as well as in S. Paratyphi A ATCC 9150 and S. Paratyphi A AKU_12601 [22]. ShdA is involved in colonization of Peyer’s patches by S. Typhimurium and in shedding of the bacteria after infection [43-45]. MglA is a galactoside transport ATP binding protein. The roles of these genes in the broad host-range of S. Enteritidis remain to be established.

All these nine CDS are pseudogenes in the seven S. Dublin clinical isolates evaluated in this work, as well as in the other fully sequenced isolate S. Dublin 3246, suggesting that the lost of their functionality is not a consequence of random mutation. Two of these 9 pseudogenes in the Uruguayan isolates have lost their functionality by mutations that are different from those seen in the sequenced strain CT_02021853 suggesting that this loose of functionality involves a process of convergent evolution.

In conclusion, our results show several genetic differences that may help to explain why such close related organisms can nevertheless behave with such marked differences. Comparison of larger numbers of field strains at full genome scale is becoming increasingly feasible, and may provide new insights into the genetic basis of host adaptation.

CONFLICT OF INTEREST

None declared.

REFERENCES