High heterogeneity of Y. enterocolitica subsp. enterocolitica

Highly-virulent Y. enterocolitica strains of bioserotype 1B/O:8 have been largely and interchangeably used to elucidate the virulence/fitness mechanisms of this gastrointestinal pathogen. According to the results obtained in this study, YE strains 8081 and WA-314 demonstrated not only high genomic diversity, but also different behaviors during in vitro and in vivo growth.

5.1.1. Genomic diversity of Y. enterocolitica subsp. enterocolitica

The newly sequenced genome of Y. enterocolitica strain WA-314 was analyzed and compared to the reference genome of strain 8081 (Thomson et al., 2006). Overall, the two highly-virulent strains have similar genome properties (Table 21), with strain WA-314 having smaller chromosome and plasmid sizes because of the incompleteness of the sequences. In fact, shotgun sequencing approaches are based on libraries created by the fragmented target genome, producing sequence reads which have to be ordered and assembled. When the sequence contains repeats, correct assemblies are not easily obtained, since algorithms are not able to place the read in the different alternative positions through the genome. The resulting assemblies are therefore formed by scaffolds, with gaps usually representing the un-assembled repetitive regions (e.g. transposases, insertion sequences and tRNAs). Since these gaps have not been closed in the genome of YE strain WA-314, the resulting sequence sizes and the number of tRNAs are underestimated and smaller than the related strain 8081.

Approximately 91% of genes in strain WA-314 have orthologs in the genome of strain 8081, with the remaining 9% mostly encoding hypothetical and mobile element-related proteins. In fact, important differences have been detected in prophages and genomic islands. Strain WA-314 is devoid of the islands YGI-3, YGI-4 and YAPI, which is involved in Y. pseudotuberculosis virulence (Collyn et al., 2004). Additionally, strain WA-314 harbors seven prophages, which show low sequence similarity with prophage regions in strain 8081. In particular, a PilV-like adhesin, carried by the prophage YWA-4, may increase intestinal colonization of strain WA-314, as it has been shown for other Enterobacteriaceae (Bieber et al., 1998; Zhang et al., 2000).

Curiously, the YWA-4 prophage occupies the same genomic region as the YGI-3 in strain 8081, between two tRNA genes. As demonstrated also by genome comparison of the whole YE species (section 4.2.4 and Figure 20), this region is highly variable and may be considered a

“hot-spot” for the integration of genetic material acquired by horizontal gene transfer (HGT) in YE biotypes 1A and 1B. Genomic islands and prophages greatly contribute to inter- and intra-species genetic variability and evolution of a variety of pathogenic bacteria (Dobrindt and Hacker, 2001), stressing the main role of horizontal gene transfer and mobile elements in shaping the genomic versatility of virulent YE. Besides mobile elements, YE highly-virulent strains 8081 and WA-314 differ in the presence and sequence similarity of certain gene clusters and protein-encoding genes. Strain WA-314 specifically possesses a xenobiotic-acetyltransferase-encoding gene, a restriction modification system and a putative colicin cluster.

Whether these proteins are efficiently expressed and functional in vitro and/or in vivo needs to be clarify. From an evolutionary perspective, being isolated from human, animal and environmental sources (Fredriksson-Ahomaa et al., 2006), YE serotype O:8 strains undergo numerous opportunities for exchanging genetic material. These flexible lifestyles make these bacteria be in contact with many microbial communities, and may therefore explain the genome plasticity as a result of continue adaptation for survival in different niches.

5.1.2. New virulence factors for the highly-virulent phenotype

Comparing and analyzing CDSs with low-sequence similarity between strains 8081 and WA-314 allowed identification of new factors putatively involved in Y. enterocolitica pathogenesis.

Autotransporters represent the largest protein family in pathogenic Gram-negative bacteria and perform heterogeneous functions, such as mediation of biofilm formation, bacterial aggregation and adhesion to epithelial cells. This functional diversity is due to the variable N-ter passenger domain, which is responsible for the pathogenesis of bacteria (Benz and Schmidt, 2011). Two orthologous autotransporters, showing only 32% sequence identity between strains 8081 and WA-314 (respective locus tags: YE3700 and YWA314_14949), contain dissimilar passenger

domains, which may confer strain-specific adhesion properties. While this gene is conserved among all YE, even if with low sequence similarity (section 4.2.2), another adhesin is specific for YEE (locus tags: YE0694 and YWA314_00878), and may contribute to the highly-virulent phenotype of strains 8081 and WA-314. In addition, a biotype 1B-specific fimbrial operon with differences between strains 8081 and WA-314 (locus tags: YE1111-1114 and YWA314_11901-11886) has been identified, and may also be considered a specific virulence factor for highly-virulent strains. In fact, fimbriae (or pili) have demonstrated adhesion properties in different Gram-negative bacteria, leading to colonization of host tissues in the urinary, genital and gastrointestinal tracts (Proft and Baker, 2009). Genome comparison of two YE biotype 1B strains, therefore, allowed identification of yet unknown genes which may be involved in pathogenesis of this highly-virulent group of bacteria.

5.1.3. Differences in the virulence plasmid sequences

Considering the established virulence determinants of Y. enterocolitica (Table 3 and section 1.2), significant differences between strains 8081 and WA-314 were found in certain pYV-encoded proteins. YopM is a T3SS-dependent effector protein indispensable for full virulence of pathogenic Yersinia, even though its molecular functions still remain largely obscure. In Y. pseudotuberculosis YopM modulates virulence by interaction with two intracellular serine/threonine kinases, PRK2 and RSK1, involving the LRR6-LRR15 and the LRR12-C-ter regions, respectively (McCoy et al., 2010; McDonald et al., 2003; McPhee et al., 2010). Notably, YopM8081 is 367-aa long with 13 LRRs, while YopMWA-314 has 505 residues and 24 LRRs (Figure 7). However, it is not clear whether the different number of LRRs in YopM proteins of strain 8081 and WA-314 has different consequences on the virulence of YE in the mouse model.

LcrV is one of the three translocator proteins which form the pore complex connecting the needle tip of the T3SS injectisome to the target host cell (Dewoody et al., 2013b). The 9-aa insertion identified in LcrVWA-314 has been previously observed in strain WA-314 and in other biotype 1B strains but is absent from strain 8081 and from weakly-virulent YE (Foultier and Cornelis, 2003; Roggenkamp et al., 1997). The role of this polymorphic region seems to be related to a humoral immunosuppression function, in that antibodies directed against Y. pestis LcrV are unable to block the T3SS-based virulence of YE strain WA-314 and strain W22703 (serotype O:9) (Miller et al., 2012; Motin et al., 1994). Therefore, antibodies produced against LcrV8081 might not protect from YE strain WA-314 infections. However, the observed dissimilarities are independent from the amino acid polymorphisms in the LcrV sequence, but may be due to differences in injectisome assembly between Y. pestis and YE (Ligtenberg et al.,

2013), suggesting that the different LcrV sequences of strains 8081 and WA-314 might not be involved in different virulence properties.

YadA is an important adhesin involved in serum resistance and in mediation of adherence to epithelial cells and to extracellular matrix (ECM) proteins (Balligand et al., 1985; Heesemann and Grüter, 1987). The fundamental role of YadA for YE virulence has been demonstrated in the mouse infection model (Roggenkamp et al., 1995). The domains involved in these virulence-associated functions are located in the N-terminal head region of the YadA protein (El Tahir and Skurnik, 2001); in particular, the amino acids 29 to 81 are important for neutrophil interaction and virulence in mice (Roggenkamp et al., 1996). Interestingly, YadA8081 and YadAWA-314 sequences possess six different amino acids at positions 53 to 67 (section 4.1.1, page 61). Whether the detected mutations alter neutrophil binding and/or virulence of YE strains 8081 and WA-314 needs further investigations.

YscP regulates the Yersinia injectisome needle length, which is directly correlated to the size of the YscP protein (Wagner et al., 2009). A minimal needle length is required for a fully functional injectisome, also in relation to the length of the YadA adhesin, providing optimal contact with the host cell (Mota et al., 2005). Yop translocation efficiency might be therefore reduced by shorter YscP proteins or longer YadA proteins. Sequence comparison of pYV8081 and pYVWA-314

discovered a longer YscP protein in YE strain WA-314 compared to strain 8081 (Figure 8), due to three additional amino acid repeats. Since the length of YadA is the same, the longer YscP protein in strain WA-314 might improve the Yop translocation and, therefore, provide better virulence properties than strain 8081.

Apart from differences in the YopM, LcrV, YscP and YadA sequences, YE WA-314 carries the ylpA gene, which is completely missing in pYV8081. This gene is conserved among the three pathogenic Yersinia species and encodes a lipoprotein related to the TraT protein. TraT is a cell-surface-exposed lipoprotein carried by plasmids of Gram-negative bacteria. Probable roles in pathogenesis are supported by the observation that TraT confers serum-resistance in E. coli (Sukupolvi and O'Connor, 1990). However, no differences were found between YE strain W22703 and the corresponding ylpA mutant in terms of resistance to human serum (Balligand et al., 1985), nor in bacterial counts from organs of mice infected with these strains (China et al., 1990). An implication of YlpA in YE strain WA-314 virulence is therefore unlikely.

5.1.4. Different colonization properties of highly-virulent Y. enterocolitica

Intra-species variation among Y. enterocolitica biotype 1B strains is evident from a genomic perspective, but is also confirmed by phenotypical diversification of strains 8081 and WA-314

during in vivo growth. In fact, these two highly-virulent strains demonstrated different virulence behaviors in a mouse model, especially in competition assays (section 4.1.3). A statistically significant number of mice were used for single strain and co-infection experiments via the intra-peritoneal route, which was preferred to oral infection to avoid different dissemination efficiency from the intestine into spleen and liver, and to enable an accurate and controlled injection of the bacterial suspension. Nevertheless, two mice were not successfully injected with strain WA-314, while another mouse showed no bacteria in the spleen and low counts in the liver, probably because of a faster splenic clearance of bacteria. Overall, bacterial growth was more efficient for strain WA-314 than strain 8081, even though the loads were statistically significantly higher in the spleens and not in the livers. In co-infected mice, strain WA-314 outcompeted strain 8081, demonstrating improved colonization efficiency and adaptation to the host environment. YE mouse-virulent strains, therefore, seem to differ in invasion and proliferation in host niches, affecting their pathogenicity level.

The genetic differences detected in strain 8081 and WA-314 may partly explain the higher virulence/fitness of strain WA-314. Apart from strain-specific prophages, which may carry hypothetical proteins with virulence-associated functions, adhesion proteins and known pYV-encoded virulence factors showed significant differences between the two analyzed strains, as discussed in the previous sections. Moreover, a colicin operon has been identified in strain WA-314. Although its activity has not been proven, it can be postulated that this colicin confers additional advantage to strain WA-314 against strain 8081, by inhibiting growth of competitor organisms. Colicins from Yersinia species are known and activity against closely-related bacteria have been demonstrated (Bosak et al., 2012; Bosak et al., 2013; Ferber and Brubaker, 1979).

The small differences observed in in vitro growth phenotypes (section 4.1.2) could also account for the in vivo behavior of YE strains 8081 and WA-314, independently of their virulence properties. However, the bacterial growth experiment was done in LB medium, at 27 °C and for some hours, in contrast to the in vivo growth rate, which was measured at the host temperature (37 °C) during a period of days. Most Yersinia virulence factors are not expressed at 27 °C but only at 37 °C, and, in general, the conditions of the in vitro growth experiment are totally different than the environmental conditions of mouse body-tissues and -fluids. Accordingly, differential expression profiles of metabolic pathways and virulence factors have been shown in Yersinia cultured in human plasma and in LB medium (Rosso et al., 2008). In conclusion, there are many factors that can determine the in vivo survival and growth of Yersinia, including nutrient composition and availability in the host niches, ability of the bacteria to gain access to the necessary nutrients, virulence factors and host defense mechanisms.