To the Editor: Francisella tularensis, the etiologic agent of tularemia, is highly infectious and considered a potential bioweapon (1–3). Although 4 subspecies of F. tularensis are recognized, most cases of tularemia are due to infection by subsp. tularensis (type A) or holarctica (type B). North America is the only region where both type A and type B cause human disease. Subspecies novicida is also found in North America, but it is of reduced virulence. Disease incidence attributable to either type A or type B is essentially unknown because the traditional method for classification of these subspecies is glycerol fermentation, which requires culture recovery (4). F. tularensis is fastidious and slow growing, with isolates recovered in a small percentage of cases.

We developed real-time TaqMan PCR assays for classification of F. tularensis type A and type B after F. tularensis is identified by culture or, in the absence of culture, by a PCR method such as the F. tularensis multitarget TaqMan assay (5). The type A TaqMan assay targets pdpD, which is present in type A, almost entirely absent from type B, and contains a 144-bp insert in novicida (6,7) (F: 5´-GAGACATCAATTAAAAGAAGCAATACCTT-3´; R: 5´-CCAAGAGTACTATTTCCGGTTGGT-3´; probe: 5´-AAAATTCTGC"T"CAGCAGGATTTTGATTTGGTT-3´). The type B assay targets a junction between ISFtu2 and a flanking 3´ region (GenBank AY06) (F: 5´- CTTGTACTTTTATTTGGCTACTGAGAAACT-3´; R: 5´- CTTGCTTGGTTTGTAAATATAGTGGAA-3´; probe: 5´- ACCTAGTTCAACC"T"CAAGACTTTTAGTAATGGGAATGTCA-3´). In type A and novicida, ISFtu2 is absent from this position (8). Oligonucleotides were designed with Primer Express version 2.0 (Applied Biosystems, Foster City, CA, USA). Probes were synthesized with a 5´ 6-carboxy-fluorescein reporter and an internal quencher (either BHQ1 [type A] or QSY-7[type B]) at the nucleotide position indicated by the quotation marks.

Assays were optimized by using 1 ng of type A (strain SchuS4) or type B (strain LVS) DNA on the LightCycler 1.2 (Roche Applied Science, Indianapolis, IN, USA). Optimized concentrations (20 μL final volume) were 1× LightCycler Fast Start DNA Master Hybridization Probe mix (Roche), 750 nmol/L primers, 200 nmol/L probe, 5 mmol/L MgCl₂ and 0.5 U uracil-DNA glycosylase. PCR conditions were 50°C for 2 min, 95°C for 10 min, 45 cycles of 95°C for 10 s and 65°C for 30 s, then 45°C for 5 min. Cycle threshold (C_t) values were calculated by using the second derivative maximum method with the y-axis at F1/F3 (LightCycler software version 3.5).

Sensitivity of each assay was assessed by using 10-fold serial dilutions (100,000 to 1 genomic equivalents [GE]) of SchuS4 or LVS DNA. Testing was performed in triplicate, with a reproducible detection limit of 10 GE for both assays. Specificity of each assay was tested with 1 ng of DNA from a panel of 62 Francisella isolates (Table A1) and 22 non-Francisella isolates (Acinetobacter, Bacillus, Brucella, Corynebacterium, Enterobacter, Enterococcus, Escherichia, Haemophilus, Klebsiella, Legionella, Proteus, Pseudomonas, Serratia, Staphylococcus, Streptococcus, and Yersinia species). Isolates were grown, DNA purified, and quantified as previously described (5). Specificity was also evaluated with DNA (2 μL) extracted as previously described from Francisella-like tick endosymbionts of Dermacentor variabilis and Francisella-like soil bacteria (Table A1) (9,10). The type A assay recognized all type A isolates with an average C_t value of 17.9 (n = 19). The type B assay detected all type B strains with an average C_t value of 17.1 (n = 21). Neither assay displayed cross-reactivity with F. tularensis subsp. novicida (n = 7), F. philomiragia (n = 15), Francisella-like tick endosymbionts (n = 3), Francisella-like soil bacteria (n = 7) (Table A1), or non-Francisella spp. (n = 22).

To evaluate the ability of the type A and type B TaqMan assays, in conjunction with the multitarget assay, to identify F. tularensis and classify subspecies in primary specimens, human, animal, and tick samples were tested (Table ). DNA was extracted from 200 μL fluid, 25 mg liver, and 10 mg spleen or lung by using the QIAamp DNA MiniKit (Qiagen, Valencia, CA, USA) and 1 μL tested. Multitarget PCR conditions were as described (5).

The multitarget and subspecies-specific PCR assays accurately identified and classified F. tularensis in all specimens positive by standard diagnostic methods (Table). In addition, the type A and type B assays provided subspecies information for positive specimens in which an isolate was not recovered for glycerol fermentation testing (Table ). All specimens negative by standard diagnostic methods tested negative by PCR. These preliminary results suggest that a F. tularensis PCR identification method, in combination with the type A and type B assays, provides the capability to identify F. tularensis and determine subspecies in the absence of culture.

We describe real-time PCR assays capable of classifying F. tularensis type A and type B and distinguishing these subspecies from the less virulent subsp. novicida. These assays are designed for use after F. tularensis has been identified by culture or by PCR. Supplemental use of these assays will allow laboratories to actively subtype F. tularensis isolates and primary specimens, thus providing subspecies information for a higher percentage of tularemia cases. Improved subspecies information will further understanding of the disease incidence and geographic distribution of F. tularensis type A and type B in North America.