Tularemia, caused by the bacterium Francisella tularensis, is a potentially severe disease that often causes unspecific symptoms; because of its low infectious dose and ease of dissemination, F. tularensis is considered a category A biothreat agent (1). Three subspecies of F. tularensis have been identified; F. tularensis subsp. tularensis (type A) has been identified only in North America. Numerous subtyping schemes have subdivided type A into 2 groups, A.I and A.II (2–8). Group A.II is found primarily in the western United States (3,4), whereas group A.I is found throughout the central and eastern regions of the country and sporadically in some western states (3,4,9).

Groups A.I and A.II differ in virulence, as do subgroups within A.I, although clinical signs and symptoms can be similar. Human infections involving A.I strains are associated with a higher fatality rate than that for infections involving A.II strains (4,10); this finding was experimentally confirmed in mice (11). Kugeler et al. (10) used pulsed-field gel electrophoresis (PFGE) to identify 2 subgroups within A.I, A1a and A1b; this study found A1b strains were associated with higher death rates and were more often isolated from human tissue types that were associated with severe disease. This difference was also experimentally confirmed in mice (11,12). However, virulence testing is not often used in clinical settings because it is slow, complicated, and expensive. Thus, molecular approaches that can rapidly assign an unknown strain to one of the recognized groups with known differences in virulence may provide valuable information to clinicians.

Figure 1

Figure 1. Neighbor-joining tree of 14 Francisella tularensis subsptularensis group A.I strains constructed on the basis of single-nucleotide polymorphisms (SNPs) discovered from whole-genome sequencingLines represent major groups within A.I: red, A.I.12; purple, A.I.8;...

Because PFGE lacks the phylogenetic resolution of some other testing methods (6), we independently identified genetic subgroups within A.I by conducting whole-genome sequencing (WGS) of 13 A.I strains (Figure 1; Table 1, Appendix). The 13 strains were selected on the basis of assignment to PFGE subgroups A1a or A1b (10) and to maximize geographic diversity; the previously sequenced A.I strain Schu S4 (13) was also included. WGS data were generated, assembled, and analyzed as described in the Technical Appendix (wwwnc.cdc.gov/EID/article/20/5/13-1559-Techapp1.pdf).

Our whole-genome phylogeny revealed 3 major subgroups within F. tularensis subsp. tularensis A.I: A.I.3, A.I.8, and A.I.12 (Figure 1). The names we assigned to these subgroups are consistent with previous phylogenetic nomenclature within F. tularensis (14). With the exception of 1 strain (ND01-1900) that was not assigned to any of the 3 subgroups, all strains previously assigned to PFGE subgroup A1a belonged to the newly designated A.I.12 subgroup (Figure 1; Table 1). In contrast, strains previously assigned to PFGE subgroup A1b were distributed among all 3 of the new subgroups (Figure 1; Table 1). We concluded that results of characterization of subgroups A1a and A1b by PFGE are not in agreement with findings of a robust whole-genome phylogeny and therefore focused the remainder of our analysis on subgroups identified by using WGS.

We observed several differences among the 3 subgroups in the whole-genome phylogeny (Figure 1). The first split separated the A.I.3 subgroup from the A.I.8 and A.I.12 subgroups; a second split separated the A.I.8 and A.I.12 subgroups. A long branch of 25 single nucleotide polymorphisms (SNPs) led to the A.I.3 subgroup, in which relatedness among the sequenced strains was moderate. A branch of 9 SNPs led to the A.I.8 subgroup, and again, relatedness among the sequenced strains was moderate. The branch leading to subgroup A.I.12 was, by comparison, much longer (37 SNPs), and the sequenced strains were separated only by 3 short branches (1–4 SNPs). This pattern of several short branches without hierarchical structuring is consistent with a recent radiation, an evolutionary process in response to adaptive change, new ecologic opportunities, or a combination of these factors.

Figure 2

Figure 2. Geographic alignment of 179 geographically diverse Francisella tularensis subsptularensis A.I strains, by subgroup, United StatesA) Canonical single-nucleotide polymorphism (canSNP) topology of 15 intervening and terminal subpopulations defined by screening of 16...

To show more comprehensive phylogenetic patterns, we developed 16 canonical SNP (canSNP) assays as described (Technical Appendix) and used them to screen 179 F. tularensis subsp. tularensis A.I strains selected from the collections of the Centers for Disease Control and Prevention (Fort Collins, CO, USA). We selected strains that were representative of all states where A.I infections occur and of all PFGE classification types (Table 1). One limitation of our study is that we did not analyze an equal number of strains from all regions of the country. However, our sample reflects the distribution of human disease caused by F. tularensis subsp. tularensis A.I strains: prevalent in the central United States, less common in the eastern United States, and rare in the western United States (4). The canSNP assays were based on 12 SNP signatures (Table 2) from the whole-genome phylogeny (Figure 1) and 4 previously described SNP signatures (6–8). Using these assays, we assigned the 179 strains to 15 F. tularensis subsp. tularensis A.I subpopulations, including 8 intervening nodes (Figure 2, panel A). We found 6 subpopulations in the A.I.12 subgroup, 4 in A.I.8, and 4 in A.I.3 (Table 1). To identify broad phylogeographic patterns, we created maps indicating specific states where strains from the 15 subpopulations were isolated (Figure 2, panel B). Within these maps, we created boundaries corresponding to 3 regions within the United States: western, central, and eastern.

Each subgroup exhibited complex yet distinct phylogeographic patterns (Figure 2, panel B). Group A.I.12 strains, assigned to 6 subpopulations (Figure 2, panel A), were isolated throughout the United States: all 6 subpopulations were found in the central region, 3 in the western region, and 5 in the eastern region (Figure 2, panel B, top). Group A.I.8 strains, assigned to 4 subpopulations, were found in the central (3 subpopulations) and western (including Alaska and British Columbia; 3 subpopulations) regions, but only 1 strain was isolated in the eastern region (Figure 2, panel B, middle). For group A.I.3 strains, assigned to 4 subpopulations, distribution differed dramatically from the other subgroups; most strains and all 4 subpopulations occurred in the eastern region and just 1 subpopulation in the central region but none in the western region (Figure 2, panel B, bottom).

Top