Pigeon-Associated Strains of Salmonella enterica Serovar Typhimurium Phage Type DT2 Have Genomic Rearrangements at rRNA Operons

The species Salmonella enterica consists of 2,487 serovars (16). Most can infect multiple hosts, while a few can infect only a single host species. Some serovars have host preferences that lie between those of the strict host-specific and generalist pathogens. They are usually found only in a primary host but can occasionally be found in secondary hosts, particularly if the secondary host frequently interacts with the primary host (3). Examples of these host-adapted serovars include S. enterica serovars Dublin (cattle adapted) and Choleraesuis (swine adapted) (2, 3).

Strict host-specific serovars of S. enterica frequently have large-scale genomic rearrangements due to recombination between the seven homologous rRNA (rrn) operons (4, 6-8), while generalist serovars have stable genomes at the rrn level (Fig. 1). More than 50 independently isolated strains of generalist serovars from natural sources worldwide have been tested, and no rrn rearrangements were detected (4, 6, 9; also data not shown). In host-adapted serovars, however, rrn rearrangements are detected, albeit at a lower frequency than those of the strict host-specific serovars (4, 8).

While most S. enterica serovar Typhimurium strains are generalists, serovar Typhimurium phage types DT2 and DT99 are highly associated with systemic disease in pigeons, but not other animals. Researchers have suggested that these pigeon-associated strains may be a separate lineage branching from the more common strains of serovar Typhimurium (15, 19). We tested 35 independently isolated pigeon-associated strains for rrn rearrangements using a previously described PCR method (6). DNA-DNA hybridization was then used to compare some of these strains with the sequenced serovar Typhimurium strain LT2. The pathogenicity of the pigeon-associated strains was assessed in mice relative to host-specific serovars adapted to nonmurine hosts.

Pigeon-associated strains were gathered from various regions in Germany from 1997 through 2003, phage typed, and stored at the Robert Koch Institute in Wernigerode, Germany. Thirty-five of these strains were randomly chosen for this study (Table 1) and grown on Luria-Bertani medium at 37°C (10). PCR was performed, and results were analyzed as described previously (6). Microarray hybridization and data acquisition and analysis were performed as described previously (11, 12, 18). Six- to eight-week-old female BALB/c mice were obtained from Harlan Sprague Dawley (Indianapolis, Ind.) and were handled in compliance with United States federal guidelines and institutional policies. Bacterial strains were diluted in 0.85% NaCl, and 100 μl of diluted culture was inoculated into the mice by intraperitoneal injection. A sample of the inoculum was spread on Luria-Bertani plates to determine the number of bacteria per infection.

PCR showed that 32 of the strains have the same rrn arrangement as S. enterica strain LT2 and all other natural serovar Typhimurium and S. enterica serovar Enteritidis strains isolated thus far (Fig. 1). Three strains (2248, 2289, and 2290) have an inversion between rrnD and rrnE (Fig. 2).

DNA-DNA microarray analysis comparing four pigeon-associated strains (the three strains with the inversion and strain 2291, which does not have the inversion) with laboratory strain LT2 showed no differences except for the absence of two prophages (Fels-1 and Fels-2) from the genomes of the pigeon-associated strains. The observation that Fels-1 and Fels-2 are missing from these strains is not surprising, as similar results have been found in other serovar Typhimurium strains recently (1, 17). No genomic differences between strains with the inversion and the strain lacking the inversion were detected.

Virulence of two pigeon-associated strains was assayed in BALB/c mice. Four mice were infected with strain 2250 (a pigeon-associated strain with the inversion), three mice were infected with strain 2248 (a pigeon-associated strain with the inversion), and three mice each were infected with fowl-specific S. enterica serovar Gallinarum and human-specific S. enterica serovar Typhi. Each mouse received approximately 650 CFU. After 6 days, one of the three mice infected with strain 2248 and two of the four mice infected with strain 2250 died of infection. A recent report shows pigeon-associated serovar Typhimurium variants isolated from Belgium also cause disease in mice, although the pathogenicity of strains with rrn arrangements was not assessed in that study (14). As expected, none of the mice infected with serovar Gallinarum or Typhi showed any symptoms of disease.

There is a strong correlation between host restriction and rrn rearrangements in S. enterica (4-9). Previous work suggests that this is not due to the ability of a given strain to undergo recombination but instead is probably a result of the host-specific lifestyle (4, 5). Since most S. enterica serovars are generalists and since these serovars have the same conserved rrn arrangement as other enteric bacteria, it is likely that when S. enterica and Escherichia coli diverged from a common ancestor 140 million years ago (13), these rearrangements were not present, and the organism was a generalist pathogen. As time, mutations, and natural selection progressed, some strains tended to favor particular hosts. These strains eventually evolved into independent serovars, optimizing particular traits to suit infection of a specific host. This scenario suggests that the less strict host-adapted serovars may become truly host-specific serovars in the future. As they evolve toward host specificity, more genomic rearrangements will be observed. Elegant studies previously performed on pigeon-associated serovar Typhimurium indicate that these strains are beginning to pursue a divergent evolutionary path relative to the more common generalist serovar Typhimurium strains (1, 15, 19). Results obtained in this study confirm this.

These results indicate that these pigeon-associated strains of serovar Typhimurium share many of the qualities found in host-adapted strains. This is the first time rrn rearrangements have been observed in natural isolates of serovar Typhimurium. The finding that only 3 of 35 independently isolated strains had an inversion suggests that the genomes of the pigeon-associated strains are much more stable than host-specific serovar Typhi. One study showed that 125 of 127 independently isolated serovar Typhi strains had a different rrn arrangement than the standard arrangement (7). Unlike strict host-specific strains, the pigeon-associated strains cause disease in BALB/c mice. Furthermore, disease progressed regardless of whether the rrn inversion was present. The fact that there are only two phage types among the 35 pigeon-associated strains (and 33 of the strains are within the same phage type) coupled with the microarray data indicates a clonal relationship among these strains. This clonality fits the model of a newly emerging niche.

The data obtained from this study, together with observations from previous studies, suggest that these strains may represent a snapshot of an intermediate phase of evolution in which an organism evolves from a generalist strain to a host-specific strain.

• Open in new tab
• Download powerpoint

FIG. 1.

rrn arrangements of S. enterica serovars Typhimurium and Typhi. (A) rrn arrangement of serovar Typhimurium strain LT2. This arrangement is found in most other generalist Salmonella strains, E. coli, and other enteric bacteria. (B) rrn arrangement of serovar Typhi strain TY2, which at some point in time has undergone inversions due to recombination between rrnG and rrnH and between rrnD and rrnE, in addition to a translocation of genome fragment D. While a given rrn arrangement is stable in standard laboratory growth conditions, different strains of serovar Typhi have very different rrn arrangements.

• Open in new tab
• Download powerpoint

FIG. 2.

Results of rrn PCR and the corresponding rrn arrangements of two representative strains from this study. Forty-nine PCRs were performed for each strain in this study. Each lane corresponds to a different hybrid rrn possibility. Each 7-kb product represents a single rrn operon. The leftmost lane in the top and bottom rows contain a size standard. (A) Strain 2291, exhibiting the standard rearrangement. The 7-kb product in the top row represents rrnG. The products in the middle row represent, from left to right, rrnA, rrnB, rrnE, and rrnH. The products in the bottom row, from left to right, represent rrnC and rrnD. All but three strains in this study had this arrangement. (B) rrnD/E inversion of strain 2248. The products for rrnD and rrnE are missing and rrnD/E (top row, six lanes from the left) and rrnE/D (bottom row, second lane from the right) are present. Strains 2289 and 2290 also have this rrn arrangement.

• American Society for Microbiology