Tracing genomic variations in two highly virulent Yersinia enterocolitica strains with unequal ability to compete for host colonization

Yersinia enterocolitica is a globally disseminated gastrointestinal pathogen which is transmitted by the fecal-oral route, through ingestion of contaminated food or water [1]. Human clinical infections most commonly occur in young individuals and are associated with acute diarrhoea, terminal ileitis, mesenteric lymphadenitis and pseudo-appendicitis [2]. Critical elements for pathogenesis are the high pathogenicity island (HPI), carrying the siderophore-mediated iron uptake system named yersiniabactin, and the virulence plasmid pYV. While the HPI and pYV are absent in avirulent strains, they are both conserved among highly virulent strains of the three pathogenic Yersinia species, Y. pestis, Y. pseudotuberculosis and Y. enterocolitica. The 70-kb plasmid encodes a type III secretion system (T3SS) and a set of Yop effector proteins which, after injection by the T3SS into host cells, inhibit several host immune mechanisms which enable the bacteria to survive in the host environment [3].

Y. enterocolitica strains are heterogeneous and are classified into 6 biotypes (1A, 1B, 2, 3, 4 and 5) according to biochemical properties [4]. Biotype 1A strains, lacking both HPI and pYV plasmid, are considered as non-virulent in mice, whereas biotypes 2 to 5, which lack HPI, are low virulence (unable to kill mice) [5]. These five biotypes belong to the Y. enterocolitica subsp. palearctica and are generally isolated in Europe and Japan (termed “Old World” strains). Biotype 1B (subsp. enterocolitica), harboring both HPI and pYV plasmid, is highly virulent (mouse-lethal) and predominant in North America (the so called “New World” strains) [6]. More than 70 serotypes of Y. enterocolitica have been described; however only few of them are virulent with serotypes O:3, O:5,27, O:8, O:9, O:20 and O:13 being the most pathogenic to humans [2]. In the past, Y. enterocolitica bioserotype 1B/O:8 strains were predominant in the United States [7]. Nowadays these strains are also isolated in other countries; nevertheless, bioserotype 4/O:3 strains are the most commonly Y. enterocolitica strains found over the world [8], [9].

Many Yersinia research laboratories use in their studies two mouse-virulent Y. enterocolitica 1B/O:8 strains, named 8081 [10] and WA-314 [5]. Y. enterocolitica strain 8081 is an American isolate from a fatal-septicemia patient [10] and has been widely used in murine infection models. Y. enterocolitica strain WA-314 was isolated from the blood of a human patient and proved to be highly virulent for mice and rats [11]. Recently, in an attempt to identify chromosomal variations underlying the different properties of high versus low virulence strains, the genome of the highly virulent strain 8081 [12] has been compared to genomes of low virulence strains [13, 14]. However, a complete comparative characterization of highly virulent 1B/O:8 strains with each other had not been performed. In this context, the sequence of the pYV_WA-314 has been recently determined, showing global similarity to pYV plasmids of other Y. enterocolitica strains but with noticeable differences in the amino acid sequences of the T3SS proteins SycH, YopM, LcrV and YscP [15]. Phenotypic differences in autoagglutination experiments [15] and results from our pilot experiments with strains 8081 and WA-314, displaying differences in bacterial growth curves and optical density-cellular mass correlation in vitro, suggested that these two strains possessed distinct virulence and colonization properties.

In this work, we demonstrated that strain WA-314 overcame strain 8081 in the colonization of spleen and liver during in vivo co-infection experiments. Genome comparison between the genomic sequence of strain 8081 [12] and a high-quality whole-genome sequence of strain WA-314, presented in this study, allowed the identification of putative virulence factors which may account for the different in vivo phenotypic behavior of these two 1B/O:8 strains.

Highly virulent Yersinia enterocolitica strains have been extensively used to clarify the virulence/fitness mechanisms of this heterogeneous gastrointestinal pathogen. The use of different bioserotype 1B/O:8 strains is normally not considered a variable factor in experimental animal infection procedures. In this study, however, two Y. enterocolitica 1B/O:8 strains, 8081 and WA-314, demonstrated different virulence behaviors in mice, both in single strain infection and in competition assays, with strain WA-314 showing a higher virulence/fitness level. A further characterization of these two strains was therefore necessary to uncover the genetic background behind the phenotypic differences.

Genome comparison between a high-quality sequence of Y. enterocolitica strain WA-314 and the published sequence of Y. enterocolitica strain 8081 [12] revealed, besides an overall similarity in gene composition, important differences in genomic islands and prophages. In pathogenic bacteria, genomic islands and prophages represent a large source of inter- and intra-species genetic variation [30]. Pathogenicity islands also play an essential role in spreading virulence genes through lateral gene transfer, an evolutionary process by which bacterial pathogens acquire new virulence factors en bloc. Our analysis revealed that the YGI-4 and the pathogenicity island YAPI, carrying a type IV secretion system known to be involved in Y. pseudotuberculosis virulence [31], are absent from strain WA-314. Moreover, both 8081 and WA-314 strains harbor prophages that have no or low similarity. We observed that the mobile genetic elements tend to occupy the same positions in both 1B/O:8 genome backbones. The P4-like prophage YWA-4, for example, is inserted in strain WA-314 in the same genomic region as the YGI-3 plasmid-like element in the chromosome of strain 8081. This speaks in favor of the presence of “hot-spots” for the integration of the acquired genetic material. Such hot spots, besides being integration sites, might represent genome regions with high gene expression potential, an important factor for the homing of laterally acquired genetic clusters. Taken together, these data emphasize the important role of horizontal gene transfer and mobile genetic elements in the evolution and genetic diversification among pathogenic Yersinia.

Besides mobile elements, the genomes of Y. enterocolitica strains 8081 and WA-314 differ in a number of gene clusters and single protein-encoding genes. Strain WA-314 specific acquisitions include a XAT-encoding gene, a RM system and a putative colicin cluster. Strain 8081, on the other hand, harbors a specific toxin/antitoxin system, similar to the HigBA family. Nucleotide polymorphisms in homologous genes also contribute to the genetic variation between strains 8081 and WA-314. The plasticity of Y. enterocolitica 1B/O:8 genomes mirror the versatile lifestyles of this heterogeneous bacterial species, found in human, animal and environmental sources, and results from the on-going process of adaptation. By living in contact with various microbial communities in different niches, Y. enterocolitica experiences frequent opportunities for exchanging genetic material.

One of the main challenges of comparative genomics is to identify genes involved in pathogenesis. Virulence factors are generally involved in adherence, invasion, colonization of the host, interference with host defense mechanisms and damage to the host [32]. A putative autotransporter, an adhesin and two fimbrial proteins show low sequence similarity in the genomes of strains 8081 and WA-314. Together with a type IV pilus, present only in the genome of strain WA-314, they may contribute to adhesion and colonization of host tissues. Such new potential virulence determinants may be able to explain the phenotypic differences observed in vivo between Y. enterocolitica strains 8081 and WA-314. The real involvement of these proteins in virulence, however, needs to be elucidated.

Established virulence-associated determinants of Y. enterocolitica have been extensively studied and reviewed [33]. One group is pYV plasmid-encoded, such as YadA, the T3SS secretion machinery Ysc and the T3SS effectors Yops, while the other group is encoded within the chromosome, for example Inv, Ail and the HPI. The genomes of Y. enterocolitica strains 8081 and WA-314 show no significant differences in the sequences of these classical virulence markers, except for the plasmid-encoded effector protein YopM and YscP, the “needle ruler” of the Yersinia injectisome. The function of the YopM protein is not completely understood. However, it has been shown to form a complex with two intracellular serine/threonine kinases, protein kinase C-like 2 (PRK2) and ribosomal S6 protein kinase 1 (RSK1) [34]. In Y. pseudotuberculosis, the interaction with RSK1 requires the region from LRR12 to C-terminus of YopM [35], whereas PRK2 binding involves the LRR6 to LRR15 region of YopM [36]. Both RSK1 and PRK2 interaction domains of YopM are critical for virulence, e.g. by inducing production of IL-10, as demonstrated by different YopM mutant proteins [36]. Interestingly, Y. enterocolitica strain 8081 encodes a YopM of 367 residues with 13 LRRs, while YopM of strain WA-314 has 505 residues and 24 LRRs. Therefore, further studies are required to elucidate whether the different number of LRRs in YopM proteins of strain 8081 and WA-314: i) generally results in the interaction of the corresponding YopM proteins with different targets in the host, ii) especially causes distinct interaction or binding affinity with PRK2 and RSK1 and iii) has different consequences on the virulence of Y. enterocolitica in the mouse infection model. YscP, a protein highly variable within Y. enterocolitica species, determines the needle length of the Yersinia spp. injectisome, with a linear correlation between the size of YscP and the needle length [28]. It has been shown that the Y. enterocolitica needle needed to have a minimal length to be fully functional [37]. Such a minimal needle length, which also correlated with the length of the YadA adhesin, provided optimal contact between the needle and the host cell membrane. Thus shorter YscP proteins or longer YadA proteins led to suboptimal Yop translocation [37]. Curiously, YscP of Y. enterocolitica strain 8081 contains 452-aa, while in strain WA-314 YscP is 538-aa long. As YadA length is unchanged between Y. enterocolitica strains WA-314 and 8081 (as predicted by gene sequence comparison), we propose that the longer YscP protein in strain WA-314 would allow higher Yop translocation efficiency than strain 8081 and, therefore, improved virulence activity. This would partly explain the different phenotypes of strains 8081 and WA-314 observed in the co-infection experiment.

Small reproducible differences between the in vitro growth rates of Y. enterocolitica strains 8081 and WA-314 have been documented, with strain WA-314 growing slightly faster than strain 8081 (data not shown). Thus the lower in vivo colonization ability of strain 8081 might be related to its growth behavior and to metabolic and regulatory factors, without regard to virulence determinants. However, no obvious differences in metabolic and nutrient acquisition systems have been found between strain 8081 and WA-314 genomes. In vitro growth conditions for Yersiniae are also extremely different from those in vivo: for example, Yersinia optimal growth temperature is 27°C, in contrast to the in vivo temperature of 37°C, and most virulence factors are only expressed at 37°C. Thus a correlation between the in vitro growth rates of Y. enterocolitica strains 8081 and WA-314 and their in vivo colonization properties is unlikely, but such a possibility cannot be completely excluded.

It is well known that pathogenesis induced by bacteria is dependent on bacterial proliferation in the host. Successful invasion of and growth in host niches requires adaptation of the pathogen to the host micro-environment. Micro-environmental factors influencing bacterial colonization of the host include nutrient composition and availability in the host niches, on one hand, and the host defense mechanisms, on the other hand. In the case of co-infection or co-existence of different pathogens in the same micro-environment, a further important factor is represented by the ability of each pathogen to compete against each other for host niches and nutrient sources. Pathogens are adapted to those micro-environmental factors. They carry adhesion molecules, which enable invasion of specific tissue and cell types, and metabolic systems, such as iron-binding proteins. They also express virulence factors, which counteract immune responses of the host, and molecules, like bacteriocins, which inhibit the growth of closely related bacterial strains, thus enabling competition against other pathogens. In our co-infection model (intra-peritoneal route) with two highly virulent Y. enterocolitica strains, strain WA-314 seemed to be better adapted than strain 8081 to the micro-environmental factors encountered in host tissues like spleen and liver (Figure 7). Thus the present data suggest that Y. enterocolitica strain WA-314 is a hyper-virulent strain. In fact, mice infected with strain 8081 showed lower bacterial loads than mice infected with strain WA-314 and strain 8081 could not colonize the host as efficiently as strain WA-314, even in the absence of a competitor pathogen. The colicin operon found in strain WA-314 showed no in vitro activity, but we cannot exclude that this colicin could be effectively expressed in vivo at the host’s environmental conditions, thus probably conferring additional advantage to strain WA-314 against strain 8081. Finally, the identified putative adhesion proteins, that have been shown here to be specifically acquired by strain WA-314, may also contribute to the higher capacity of this strain to invade the host.

This study demonstrated that Y. enterocolitica highly virulent strains exhibit significant strain-to-strain genotypic and phenotypic differences, resulting in differences in their pathogenicity. Accordingly, virulence of pathogens can be increased, e.g. by acquiring new genes and/or improving the function of essential virulence proteins. Thus, understanding the genetic factors which allow bacterial hyper-virulence would enable the design of improved therapeutic strategies against such strains.

DG designed and executed all experiments, carried out the genome comparison and bioinformatics analysis and drafted the manuscript. HB performed the major part of the mouse infection experiments, provided ideas for the study and participated in the composition of the manuscript. JH and AR supervised the study and helped to draft the manuscript. AR also conceived of the study and participated in its design. All authors read and approved the final manuscript.