Development of a Multitarget Real-Time TaqMan PCR Assay for Enhanced Detection of Francisella tularensis in Complex Specimens

Abstract

Tularemia is the zoonotic disease caused by the gram-negative coccobacillus Francisella tularensis. Its wide distribution in the environment poses a challenge for understanding the transmission, ecology, and epidemiology of the disease. F. tularensis is also considered a potential biological weapon due to its extreme infectivity. We have developed a multitarget real-time TaqMan PCR assay capable of rapidly and accurately detecting F. tularensis in complex specimens. Targeted regions included the ISFtu2 element and the 23kDa, fopA, and tul4 genes. Analysis of the four TaqMan assays demonstrated that three (ISFtu2, 23kDa, and tul4) performed within our established criterion of a detection limit of one organism. The combined use of the three assays was highly specific, displaying no cross-reactivity with the non-Francisella bacteria tested and capable of differentially diagnosing both F. tularensis and Francisella philomiragia. When the multitarget TaqMan assay (ISFtu2, 23kDa, and tul4) was compared to culturing, using environmentally contaminated specimens, the TaqMan PCR assay was significantly more sensitive than culturing (P ≤ 0.05). The sensitive and specific nature of this rapid multitarget TaqMan assay provides a valuable new tool that with future evaluations can be used for analyzing clinical specimens, field samples during bioterrorism threat assessment, and samples from outbreaks and for improving our understanding of the ecology and environmental prevalence of F. tularensis.

Francisella tularensis is a gram-negative coccobacillus that is the etiologic agent of the disease tularemia (5, 6, 9, 31). It was first described in 1911 in Tulare County, Calif., and since has been reported throughout North America and Eurasia. F. tularensis naturally occurs in ∼150 species of vertebrates, including lagomorphs and rodents, in ∼100 species of invertebrates, and in contaminated soil, water, and vegetation (18, 25). It has a low level of natural transmission to humans, with an average of 124 reported cases per year in the United States from 1990 to 2000 (5). Humans become infected through arthropod bites (ticks and deerflies), contact with infected tissues, ingestion of contaminated water or food, and inhalation of infective aerosols. The clinical presentation of tularemia varies depending upon the route of infection. While the most common manifestation is ulceroglandular tularemia, pneumonic tularemia is the most severe form of the disease. F. tularensis is considered a biological weapon due to its extreme infectivity, as <10 organisms can cause severe disease (7).

The genus Francisella consists of two species, F. tularensis and F. philomiragia (6, 31). Species and subspecies of Francisella differ with regards to their geographic distribution and virulence in humans. Of the four F. tularensis subspecies, F. tularensis subsp. tularensis, also known as type A, has the highest mortality rate in humans (31). Human cases have been limited to North America, though type A organisms have been isolated from arthropods in Europe (16). F. tularensis subsp. holarctica, also known as type B, is widely distributed throughout the Northern Hemisphere and rarely causes fatal disease. F. tularensis subsp. mediaasiatica is isolated in focal regions of Central Asia and has not been associated with human disease. F. tularensis subsp. novicida is infrequently associated with human disease in North America and more recently in Australia (33). F. philomiragia appears to be an opportunistic pathogen, primarily causing serious disease in immunocompromised patients or near-drowning victims. F. philomiragia has been isolated from patients in Europe, North America, and Australia (33).

Human outbreaks of tularemia caused by both F. tularensis subsp. tularensis and holarctica have been reported in the United States, Spain, Sweden, Norway, Russia, and Kosovo (1, 2, 3, 8, 11, 27, 28). Infections were associated with contaminated food and water, aerosolized bacteria produced during landscaping and farming, handling or skinning of lagomorphs, arthropod bites, and crayfish fishing. Since F. tularensis is present in a diverse number of animals and environmental habitats, a large variety of field and environmental samples, from carcasses, fish, vectors, water, soil, lawn clippings, hay, feces, and urine, need to be tested. Thus, for many of these outbreaks, transmission sources are not accurately understood due to laboratory challenges associated with testing, detecting, and recovering F. tularensis from complex and varied specimens.

Diagnostic methods for presumptive and confirmatory identification of F. tularensis include a direct fluorescently labeled antibody (DFA) test, PCR, and culturing. The DFA test is a rapid, specific test (6, 20), but it has a sensitivity limitation of ∼10⁶ cells/ml. Gel-based PCR assays targeting the tul4 and fopA genes have been developed, but they lack sensitivity and rapidity in comparison to real-time PCR assays (13, 23, 29). Culturing is the Centers for Disease Control and Prevention (CDC) “gold standard” for confirmatory diagnosis of infection. However, F. tularensis is a slow-growing, fastidious organism that requires 24 to 72 h for growth on cysteine-enriched agar (6, 31) and its growth is often out-competed by contaminating bacteria, particularly when environmental samples are being tested. F. tularensis is also notorious for causing laboratory-acquired infections and must be handled under biosafety level 3 (BSL-3) conditions (6).

Real-time TaqMan PCR is a rapid assay with high sensitivity (21, 22). The utilization of a fluorogenically labeled probe enhances sensitivity by at least 7 orders of magnitude compared to conventional PCR (21). In addition, the fluorogenic probe allows for real-time data collection due to the production of a fluorescent signal during amplification, eliminating the need for gel electrophoresis.

For this study, four individual real-time TaqMan PCR assays were developed for combined use for the detection of F. tularensis in a LightCycler instrument (Roche Applied Science, Indianapolis, Ind.). We chose to develop TaqMan PCR assays, as opposed to other real-time assays, since TaqMan assays can provide high specificity due to binding of two primers as well as a probe. Currently, a real-time TaqMan PCR assay for F. tularensis is available through the Laboratory Response Network (CDC, Atlanta, Ga.). However, due to national security concerns regarding F. tularensis as a potential biological weapon, these reagents have limited distribution and use. Therefore, we developed a real-time TaqMan PCR assay that would be available for public use for research, clinical, or surveillance purposes. Our assay development included two goals: (i) the identification of multiple F. tularensis-specific genomic targets that could be utilized in combination (multitarget assay) and (ii) the reliable detection of one organism of F. tularensis. Multiple targets would enhance specificity by reducing the likelihood of obtaining false positives due to reliance on a single target. Additionally, a detection limit of one organism increases the likelihood of detecting F. tularensis in environmental samples (e.g., water, hay, and grass) in which the number of organisms is expected to be quite low.

Regions targeted for TaqMan assay development were specific for Francisella and included the ISFtu2 element, a newly described insertion element-like sequence present in multiple copies (32; Y. Zhou, S. W. Bearden, Z. L. Berrada, L. G. Carter, B. M. Yockey, S. K. Urich, and M. C. Chu, unpublished data), the 23kDa gene, which encodes a protein that is expressed upon macrophage infection (14), and the tul4 and fopA genes, which encode outer membrane proteins (26, 30). The ISFtu2, 23kDa, and tul4 assays were each capable of detecting one organism and therefore were combined into a single multitarget assay for further analysis. This multitarget assay was species specific, able to differentially diagnose both F. tularensis and F. philomiragia. Comparison of the F. tularensis TaqMan assay with culturing, using contaminated specimens, demonstrated that the combined TaqMan PCR assay was significantly more sensitive than culturing (P ≤ 0.05). Thus, the F. tularensis multitarget TaqMan PCR assay described here should be of valuable use for detecting F. tularensis during future outbreaks and field studies.

MATERIALS AND METHODS

TaqMan primers and probes.

The GenBank sequences for F. tularensis ISFtu2 (accession no. AY06), 23kDa (accession no. Y08861), tul4 (accession no. M32059), and fopA (accession no. AF097542) were used to design four sets of TaqMan primers and probes (Table 1). For primer and probe design, ABI Primer Express software was used (Applied Biosystems, Foster City, Calif.). Prior to synthesis of primer-probe sets by the CDC Biotechnology Core Facility (Atlanta, Ga.), the sequences were subjected to a BLAST search to verify their specificity for Francisella. All fluorogenic probes were synthesized with a 6-carboxy-fluorescein reporter molecule attached to the 5′ end and a Black Hole Quencher (BioSearch Technologies, Inc., Novato, Calif.) attached to the 3′ end.

TABLE 1.

F. tularensis TaqMan primers and probes

Gene target	Primer or probea	Sequence (5′→ 3′)	Amplicon size (bp)
ISFtu2	ISFtu2F	TTGGTAGATCAGTTGGTGGGATAAC	97
	ISFtu2R	TGAGTTTTACCTTCTGACAACAATATTTC
	ISFtu2P	AAAATCCATGCTATGACTGATGCTTTAGGTAATCCA
23kDa	23kDaF	TGAGATGATAACAAGACAACAGGTAACA	84
	23kDaR	GGATGAGATCCTATACATGCAGTAGG
	23kDaP	TCAGTTCTCACATGAATGGTCTCGCCA
tul4	Tul4F	ATTACAATGGCAGGCTCCAGA	91
	Tul4R	TGCCCAAGTTTTATCGTTCTTCT
	Tul4P	TTCTAAGTGCCATGATACAAGCTTCCCAATTACTAAG
fopA	FopAF	ATCTAGCAGGTCAAGCAACAGGT	87
	FopAR	GTCAACACTTGCTTGAACATTTCTAGATA
	FopAP	CAAACTTAAGACCACCACCCACATCCCAA

^aF, forward primer; R, reverse primer; P, probe.

PCR optimization.

TaqMan PCR conditions were optimized with 10 ng of DNA from F. tularensis subsp. tularensis (strain SchuS4). All reactions were performed in a final volume of 20 μl and contained LightCycler Fast Start DNA master hybridization probe mix (Roche) at a 1× final concentration, primers (500 nM), probes (100 nM), MgCl₂, and 0.5 U of uracil-DNA glycosylase per reaction. For the 23kDa, tul4, and fopA TaqMan assays, the final MgCl₂ concentration was 4 mM, and for the ISFtu2 assay, the final concentration was 5 mM. The optimum annealing temperature for all four TaqMan assays was 60°C. Thermal cycling conditions were as follows: 50°C for 2 min, 95°C for 10 min, 45 cycles at 95°C for 10 s and 60°C for 30 s, and then 45°C for 5 min.

Bacterial isolates.

The specificities of the four TaqMan assays were evaluated with DNA from 87 non-Francisella species isolated from both animals and humans (Table 2). Bacterial cultures were obtained from the Poudre Valley Hospital Clinical Laboratory (Ft. Collins, Colo.), Denver Children's Hospital (John James), and Colorado State University Veterinary Diagnostic Laboratory (Doreene Hyatt). All isolates were stored in brain heart infusion broth with 10% glycerol at −70°C. DNA from Brucella species was kindly provided by R. Martin Roop II (East Carolina University, Greenville, N.C.). Barry Fields (CDC) kindly provided DNA from Legionella species.

TABLE 2.

Evaluation of cross-reactivity of ISFtu2, 23kDa, tul4, and fopA TaqMan PCR assays using non-Francisella bacterial isolates (n = 87)^a

Genus	Species	No. tested
Acinetobacter	A. calcoaceticus	2
Bacillus	Bacillus spp.	3
Brucella	B. abortus (2308, RB51), B. melitensis 16M, B. suis biovar II	4
Corynebacterium	C. diphtheriae	3
Enterobacter	E. cloacae	4
Escherichia	E. coli	9
Hemophilus	H. influenzae	3
Klebsiella	K. oxytoca, K. pneumoniae	11
Legionella	L. anisa, L. bozemani, L. feeleii, L. longbeachae, L. pneumoniae	5
Moraxella	M. catarrhalis	1
Proteus	P. mirabilis	4
Pseudomonas	P. aeruginosa	8
Serratia	S. marcesans	2
Staphylococcus	S. aureus, coagulase-negative staphylococci	9
Streptococcus	S. faecalis, S. pneumoniae	7
Yersinia	Y. enterocolitica, Y. pestis, Y. pseudotuberculosis	12

^aResults for all isolates for all four assays were negative, with a negative result being no DNA amplification after 45 cycles.

DNAs were also analyzed from a total of 55 wild-type Francisella isolates which originated from a variety of sources, including humans, animals, and water (Table 3). The 55 isolates included F. tularensis subsp. tularensis (n = 18), F. tularensis subsp. holarctica (n = 18), F. tularensis subsp. novicida (n = 4), and F. philomiragia (n = 15) and were selected from the CDC reference strain collection (Ft. Collins, Colo.). All Francisella isolates were confirmed by characteristic growth on cysteine heart agar with 9% chocolatized blood (CHAB) at 37°C, by DFA test, and by biochemical profile (Biolog GN2 identification system; Biolog Inc., Hayward, Calif.).

TABLE 3.

Panel of wild-type Francisella strains used for evaluation of TaqMan assays

Francisella species (n)	Origin	Host or Source	CDC no.
F. tularensis subsp. tularensis (18)	Ohio	Human	SchuS4
	Massachusetts	Human	MA002987
	Colorado	Cat	CO012364
	Colorado	Rabbit	CO013713
	North Dakota	Human	ND000952
	Kansas	Cat	KS000948
	Oklahoma	Human	OK002731
	North Carolina	Rabbit	NC993990
	Arkansas	Human	AR000028
	Utah	Human	UT983134
	New Mexico	Rabbit	NM990295
	North Carolina	Rabbit	NC973057
	North Carolina	Cat	NC015379
	Arkansas	Rabbit	AR982147
	Oklahoma	Human	OK004337
	Arkansas	Human	AR011117
	Missouri	Human	MO011907
	California	Human	CA020099
F. tularensis subsp. holarctica (18)	Russia	Rat	LVS
	Oregon	Monkey	OR960246
	Canada	Human	CN985979
	Arizona	Rat	AZ001325
	Illinois	Human	IL003633
	Missouri	Human	MO011673
	Kentucky	Human	KY001708
	Ohio	Prairie dog	OH013209
	Utah	Human	UT002098
	Spain	Rabbit	SP986120
	Indiana	Rat	IN983055
	Colorado	Vole	CO961243
	California	Human	CA990837
	Indiana	Human	IN002758
	Arizona	Squirrel	AZ001324
	California	Monkey	CA993992
	Spain	Human	SP982108
	New Mexico	Human	NM002642
F. tularensis subsp. novicida (4)	Louisiana	Human	GA993548
	California	Human	GA993549
	Utah	Water	GA993550
	Utah	Human	UT014992
F. philomiragia (15)	California	Human	GA012793
	Colorado	Human	GA012794
	New York	Human	GA012795
	California	Human	GA012796
	Pennsylvania	Human	GA012797
	Connecticut	Human	GA012799
	Connecticut	Human	GA012800
	New York	Human	GA012801
	California	Human	GA012802
	New Mexico	Human	GA012803
	Virginia	Human	GA012804
	Massachusetts	Human	GA012806
	Unknown	Human	GA012807
	Utah	Water	GA012810
	Utah	Water	GA012811

DNA extraction.

For isolation of genomic DNA from non-Francisella species, bacteria were freshly cultured on 5% sheep blood agar at 37°C. All Francisella isolates were freshly cultured on CHAB at 37°C in a BSL-3 laboratory, with use of appropriate protective gear, including closed front gown, N95 mask, and gloves. Template DNA was isolated from gram-negative bacteria by use of the QiaAmp DNA mini kit (Qiagen Inc., Valencia, Calif.) and from gram-positive bacteria by use of the MasterPure DNA purification kit (Epicentre, Madison, Wis.). For purification of DNA from Francisella isolates, lysis was performed under BSL-3 conditions. Once DNA was lysed, purification reactions were performed on the benchtop in a BSL-2 laboratory. Concentrations of DNA were approximated by electrophoresis on a 1% agarose gel and visual comparison to known concentrations of DNA (1-kb EZ Load molecular ruler; Bio-Rad Laboratories, Hercules, Calif.). All samples were stored at −20°C until use. For preservation of the integrity of the genomic DNAs, aliquots were prepared and freeze-thawing was minimized.

Real-time TaqMan PCR assays.

PCR amplification and product detection were performed in a LightCycler instrument (Roche). For each TaqMan run, both negative (no template) and positive (1 ng of purified F. tularensis strain SchuS4 DNA) controls were included (one of each for every five samples tested). If any of the controls failed, the run was repeated. For evaluation of genomic DNA purified from bacterial cultures, 1 ng from both Francisella and non-Francisella species was tested; 1 ng of F. tularensis DNA is ≈500,000 genome equivalents (GEs) (genome size, 2 MB) and provides a strong fluorescent TaqMan signal (cycle threshold [C_t] value, ≤18). The C_t values were determined by the LightCycler instrument for each reaction by setting the y axis to F1/F3 and performing automatic quantification using the second derivative maximum method. To verify that amplified products were the correct size, amplicons were electrophoresed on 3.5% agarose gels and visualized by staining with ethidium bromide.

Sensitivity determinations.

TaqMan assays were evaluated by use of F. tularensis subsp. tularensis (strain SchuS4) and subsp. holarctica (strain LVS) and two independent methods for determining sensitivity (CFU and GEs). For determination of CFU, single colonies were picked from a fresh culture and suspended in phosphate-buffered saline (pH 7.2) containing 5 mM MgCl₂ to a final density of 56% transmittance at 590 nm. Tenfold serial dilutions, to 10⁻⁹, were made and 50 μl of each dilution was spread evenly in duplicate on CHAB. Plates were incubated for 48 h at 37°C, at which time colonies were counted and the results from the two plates were averaged. For standard curves based on CFU, 50 μl was removed from the 10⁻¹ to 10⁻⁹ dilutions and added to 150 μl of Tris-EDTA buffer (pH 8.0). DNA was extracted by boiling lysis at 95°C for 10 min, and 1 μl from each dilution was tested. For standard curves based on GEs, DNA was extracted from the 10⁻¹ dilutions by use of the MasterPure DNA purification kit (Epicentre). The DNA concentration was approximated by electrophoresis in a 1% agarose gel and visual comparison to known concentrations of DNA (1-kb EZ Load molecular ruler; Bio-Rad). Tenfold serial dilutions (to 10⁻⁹) of DNA were then made and 1 μl from each dilution was tested. GE calculations were based on assuming a 2-MB genome size for F. tularensis. Standard curves based on both CFU and GEs were plotted as C_t versus log input. Standard deviations were calculated based on the averages of three independent experiments performed by the same operator. To assess the log-linear relationship of the assays, the linear regression and square regression coefficients (R²) were calculated.

Animal tissues.

A total of 10 different F. tularensis strains, 5 F. tularensis subsp. tularensis and 5 F. tularensis subsp. holarctica (Table 4), were inoculated into 10 pathogen-free Swiss Webster mice (IACUC protocol 00-06-018-MUS). In all cases, 100 CFU were administered by subcutaneous injection. All inoculations were performed in a BSL-3 animal facility, and appropriate biosafety measures were followed. Animals were sacrificed when signs and symptoms of tularemia were evident (3 to 5 days after inoculation). Liver and spleen tissues were excised from each mouse (total of 20 tissue samples) and DNA was extracted from 25 mg of liver and 10 mg of spleen by the QiaAmp DNA mini kit (Qiagen Inc.) tissue protocol. DNA samples were divided into aliquots and stored frozen (−20°C) until use. For real-time PCR analysis of total DNA (animal plus bacterial) from tissues, 1 μl of purified DNA was used.

TABLE 4.

Evaluation of F. tularensis multitarget TaqMan PCR assay using mouse tissues^a

Subjects (n)	CDC strain no.	Result by multitarget TaqMan assayb (ISFtu2, 23kDa, and tul4)
Mice infected with F. tularensis subsp. tularensis (5)	Schu4	+
	ND000952	+
	CO013713	+
	MA002987	+
	CO012364	+
Mice infected with F. tularensis subsp. holarctica (5)	LVS	+
	OR960246	+
	AZ001325	+
	IL003633	+
	CN985979	+
Healthy uninfected mice (5)		−

^aFor each mouse (15), the liver and spleen were tested.

^b+, positive result, 9 ≤ C_t ≤ 25; −, negative result, no DNA amplification after 45 cycles.

Comparison of the multitarget TaqMan PCR assay to culturing.

A total of 86 tissues from a naturally occurring tularemia prairie dog outbreak were evaluated to compare multitarget TagMan PCR with culturing: 46 tissues were from animals succumbing to tularemia and 40 tissues were from healthy uninfected animals (4). Tissues were inoculated onto CHAB or CHAB containing antimicrobials (19), incubated for 7 days at 37°C, and checked daily for growth. For DNA extraction, liver and spleen tissues were processed and stored by the animal tissue protocol described above. For real-time PCR analysis, 1 μl of purified DNA was tested under blind coding.

Statistical analysis.

The differences between test sensitivities were evaluated by McNemar's test.

RESULTS

Initial evaluation.

To demonstrate that the four TaqMan assays amplified their respective genomic targets from F. tularensis, we performed an initial assessment of the assays using F. tularensis subsp. tularensis and holarctica. Logarithmic amplification was observed for all four genomic targets (12 ≤ C_t ≤ 19) and DNA fragments of the expected sizes were produced (Table 1; data not shown).

Specificity.

Specificity of the TaqMan assays was determined by use of DNAs from 87 bacterial strains originating from both animal and human sources (Table 2). These bacterial species were chosen for comparison because they represent environmental bacteria, organisms found in the same infection sites as F. tularensis (respiratory tract and wounds), vector-borne organisms, and potentially cross-reactive bacteria (Brucella and Legionella). No evidence of cross-reactivity was detected with the four TaqMan assays, as evidenced by the absence of DNA amplification after 45 cycles (Table 2).

In comparison, the ISFtu2 element and 23kDa, tul4, and fopA genes were all detected when template DNAs from 18 F. tularensis subsp. tularensis, 18 F. tularensis subsp. holarctica, and 4 F. tularensis subsp. novicida isolates were tested (Table 3; Fig. 1). When the genomic targets were amplified from F. tularensis subsp. tularensis, C_t values averaged 14, 15, 16, and 16 for ISFtu2, 23kDa, tul4, and fopA, respectively (Fig. 1). For the amplification of ISFtu2, 23kDa, tul4, and fopA from F. tularensis subsp. holarctica, C_t values averaged 14, 18, 19, and 19 (Fig. 1). The average C_t value for F. tularensis subsp. novicida for the 23kDa assay was 27, while it was 11, 16, and 17, respectively, for the ISFtu2, tul4, and fopA assays (Fig. 1). Appropriately sized DNA products were produced from all F. tularensis subspecies tested.

FIG. 1.

Average C_t values for the ISFtu2 (A), 23kDa (B), tul4 (C), and fopA (D) assays. C_t values for the 18 F. tularensis subsp. tularensis strains (F.t.t.), 18 F. tularensis subsp. holarctica strains (F.t.h.), 4 F. tularensis subsp. novicida strains (F.t.n.), and 15 F. philomiragia stains (F.p.) and their respective standard deviations are shown. *, for F. philomiragia, the 23kDa C_t value represents the average for four isolates in which 23kDa was detected.

To determine if the TaqMan assays were species specific for F. tularensis, DNAs from 15 F. philomiragia isolates were tested (Table 3). The ISFtu2 and fopA TaqMan assays amplified products of the appropriate sizes from the 15 F. philomiragia DNAs; however, the average C_t values, 33 and 36, respectively, were consistently higher than those observed for F. tularensis (Fig. 1). The region of the 23kDa gene targeted by the TaqMan assay was only amplified from 4 of the 15 F. philomiragia isolates tested. The region of tul4 targeted in our assay could not be amplified from any of the F. philomiragia DNAs, demonstrating that the tul4 TaqMan assay is specific for F. tularensis. Thus, the combined use of the tul4 assay with either the ISFtu2 or fopA TaqMan assay is discriminatory for the two Francisella species, F. tularensis and F. philomiragia.

Sensitivity.

We next generated standard curves to determine the detection limits of the four TaqMan assays. For cell standard curves, DNA was amplified from 10-fold dilutions of F. tularensis suspensions (subsp. tularensis [strain SchuS4]) and holarctica [strain LVS]) with known CFU. Analysis of amplification curves for all four TaqMan assays showed that each assay displayed similarly shaped curves over the 10-fold dilution series. For both F. tularensis subsp. tularensis and subsp. holarctica, the detection limit for the tul4 TaqMan assay was ∼1 CFU, while it was <1 CFU for the ISFtu2 and 23kDa assays (Fig. 2A to C; data not shown). All three assays showed a linear log correlation. In contrast, the fopA TaqMan assay had a detection limit of ∼45 CFU (data not shown).

FIG. 2.

Cell (top) and DNA (bottom) standard curves for F. tularensis subsp. tularensis (strain SchuS4). The relationship between 10-fold serial dilutions of CFU versus C_t for the ISFtu2 (A), 23kDa (B), and tul4 (C) TaqMan assays is shown in the top panels; the calculated square regression coefficients were 0.98, 0.9971, and 0.9967, respectively. The relationship between 10-fold serial dilutions of GEs versus C_t for the ISFtu2 (D), 23kDa (E), and tul4 (F) TaqMan assays is shown in the bottom panels; the calculated square regression coefficients were 0.9853, 0.9997, and 0.9632, respectively. The log-linear regression and standard deviations are indicated.

Since growth of live bacteria on culture plates (CFU) may not be an accurate estimate of the total number of colonies present (both live and dead), especially for a fastidious organism like F. tularensis, the detection limits were also calculated based on DNA concentrations. For the 23kDa, tul4, and fopA assays, we assumed that 1 CFU was equal to 1 GE, and for the ISFtu2 assay, we assumed that 1 CFU was equal to ∼15 and 28 GEs for F. tularensis subsp. tularensis and holarctica, respectively. When logarithmic dilutions of both F. tularensis subsp. tularensis and holarctica were tested, the detection limit for the 23kDa and tul4 assays was ∼1 GE and for the ISFtu2 assay was <1 GE (Fig. 2D to F; data not shown). The ISFtu2, 23kDa, and tul4 TaqMan assays had a linear log correlation. As observed for cell standard curves, the fopA assay had the highest limit of detection (∼44 GEs) based on DNA concentrations (data not shown). Therefore, further analysis was only performed with the ISFtu2, 23kDa, and tul4 TaqMan assays, all of which met the established criterion of detection of 1 CFU or GE.

Analysis of tissue specimens.

To test the ability of the ISFtu2, tul4, and 23kDa TaqMan assays to identify F. tularensis in biological specimens, infected mouse tissues were analyzed. Whereas DNAs isolated from spleens and livers of control mice were negative in the assays, DNAs isolated from tissues of F. tularensis-infected mice were consistently positive by all three assays (Table 4). In addition, tissue samples spiked with known concentrations of F. tularensis DNA showed no evidence of inhibition (data not shown). Thus, the ISFtu2, 23kDa, and tul4 assays can be used in combination to accurately detect F. tularensis in sterile tissue specimens.

Comparison of the F. tularensis multitarget TaqMan PCR assay to culturing.

We next compared the sensitivity of the combined ISFtu2, 23kDa, and tul4 TaqMan assay to culturing, since culturing is the CDC gold standard for the diagnosis of F. tularensis. Because our goal was to develop an enhanced assay capable of accurately identifying F. tularensis in the presence of contaminating bacteria, our comparison focused on F. tularensis-infected animal carcasses. Culturing of these tissues on CHAB medium showed them to be highly contaminated with environmental bacteria, including Proteus, Pseudomonas, and Staphylococcus (J. M. Petersen, unpublished data).

F. tularensis was isolated from 40 of the 46 carcasses by culturing, yielding a recovery rate of 86.9% (Table 5). In comparison, the combined ISFtu2, 23kDa, and tul4 TaqMan assay accurately detected F. tularensis in 46 of the 46 carcasses (100%), with average C_t values of 20, 22, and 22, respectively. None of the 40 (0%) F. tularensis-negative tissues were positive by the TaqMan assays. The difference in sensitivity between culturing and the multitarget TaqMan is statistically significant (P ≤ 0.05), with the combined TaqMan assay being more sensitive than culturing for testing of complex specimens from naturally occurring outbreaks.

TABLE 5.

Comparison of multitarget TaqMan assay with culturing for detection of F. tularensis in complex specimens

Samples (n)a	No. (%) of positive results by:
	Culture	Multitarget TaqMan assay (ISFtu2, 23kDa, and tul4)
Infected animal carcasses (46)	40 (86.9)	46 (100)
Uninfected animals (40)	0 (0)	0 (0)

^aThe tissues tested were spleens and livers. In all cases, the same tissue was tested by both culturing and real-time PCR.

DISCUSSION

For the present study, we developed a highly sensitive and specific multitarget real-time TaqMan PCR assay that exceeds the detection limit of existing real-time PCR assays for F. tularensis (10, 17). This multitarget assay is directed against three different genomic loci, the ISFtu2 element and the 23kDa and tul4 genes, and is capable of detecting 1 CFU or GE of F. tularensis. We are confident in the detection limit of this assay, as two independent methods (CFU and GE) were utilized and produced similar results. This multitarget TaqMan assay is also species specific and can differentially diagnose both F. tularensis and F. philomiragia. Identification of F. tularensis occurs when all three targets (ISFtu2, 23kDa, and tul4) are positive, whereas identification of F. philomiragia occurs when the ISFtu2 assay is positive and the tul4 assay is negative.

Multiple genomic loci (multitarget) were targeted in the real-time TaqMan PCR assay in order to enhance the specificity of the test. While the tul4 gene has been targeted in previous standard and real-time PCR assays (10, 23, 29), this is the first time that either the 23kDa gene or the ISFtu2 element have been utilized. The 23kDa gene is unique to Francisella, and research has focused on the role the gene product plays in macrophage infection and pathogenesis (14, 15). The ISFtu2 element, a recently described insertion sequence-like element, was chosen for development since it is present in multiple copy numbers (12 to 17 copies in F. tularensis subsp. tularensis, 26 to 30 copies in subsp. holarctica, 6 to 18 copies in subsp. novicida, and 1 to 2 copies in F. philomiragia), thus ensuring maximum sensitivity for the assay (32; Y. Zhou et al., unpublished data). Indeed, the ISFtu2 assay was the most sensitive of the three and had a detection limit of <1 organism for both F. tularensis subsp. tularensis and holarctica.

Enhanced specificity of the multitarget TaqMan assay (ISFtu2, 23kDa, and tul4) was shown upon evaluating a large panel of Francisella and non-Francisella isolates. First, the assay showed no evidence of cross-reactivity with non-Francisella bacteria (environmental bacteria, common respiratory tract and wound bacteria, and vector-borne organisms), as judged by the absence of DNA amplification after 45 cycles. In addition, the assay displayed no evidence of cross-reactivity when tissues known to be contaminated with environmental bacteria were tested. Second, the multitarget assay detected all Francisella species from a variety of sources, demonstrating its ability to reliably identify wild-type isolates. Third, the multitarget assay was species specific. Due to the specificity of the assay, at this point in the evaluation, we have chosen not to define a cutoff value for a positive result. Ideally, we would like to set as high a cutoff as possible to increase the likelihood of detecting positive samples in which numbers of bacteria are expected to be quite low. Defining this cutoff will require further evaluation with a variety of clinical and environmental specimens.

Because we tested a large panel of Francisella isolates for this study, we were able to make several interesting observations regarding the taxonomy of the genus. Our results are consistent with the recent classification of subsp. novicida as a subspecies of F. tularensis (31), since all subsp. novicida isolates, including a recent one from Australia, tested positive for all three genomic targets (33; data not shown) (Fig. 1). DNA from F. tularensis subsp. novicida did show a small difference from F. tularensis subsp. tularensis and holarctica. On average, higher C_t values were observed for the 23kDa assay with subsp. novicida (average C_t, 27) than with subsp. tularensis and holarctica (average C_t, 16). Since the concentrations of template DNA were the same, it is likely that the 23kDa gene sequence is somewhat divergent between subsp. novicida and subsp. tularensis and holarctica.

Our results are also consistent with the designation of F. tularensis and F. philomiragia as two distinct species within the Francisella genus (31). All F. philomiragia isolates tested positive by the ISFtu2 assay. The difference in the ISFtu2 C_t values for F. philomiragia (average C_t, 33) and F. tularensis (average C_t, 14) can be attributed to either sequence divergence or fewer copy numbers of the ISFtu2 element in F. philomiragia (Y. Zhou et al., unpublished data). The tul4 gene was not detected in any of the F. philomiragia isolates used in this study or in a single isolate tested in another study (10). DNA hybridization studies with F. tularensis have shown that the tul4 gene is present in F. philomiragia (30), suggesting that the region of the tul4 gene targeted in our assay is divergent enough in F. philomiragia to not be recognized by the tul4 assay. In addition, the majority of F. philomiragia isolates were negative by the 23kDa TaqMan assay. The 23kDa gene has never been studied in F. philomiragia, and therefore, it is unclear whether this gene is divergent or missing entirely from this species. Since the 23kDa gene is known to be important for intracellular growth of F. tularensis in macrophages (14, 15), it will be interesting to test whether F. philomiragia can also infect macrophages.

Here we show that the multitarget TaqMan assay (ISFtu2, 23kDa, and tul4) is significantly more sensitive (P ≤ 0.05) than culturing for testing of complex specimens. Tissues from dead animals are usually highly contaminated with environmental bacteria, making culture recovery a challenge since F. tularensis is a slow-growing, fastidious organism. Past studies have had limited success in isolating F. tularensis from dead animals, with culture recovery rates from carcasses only around 30% (24). More recently, F. tularensis has also been reported to persist in a viable but noncultivable state (12). Despite the challenges associated with culturing F. tularensis from carcasses, the multitarget TaqMan assay detected F. tularensis in 100% of cases, demonstrating that the multitarget TaqMan assay is advantageous for the detection of F. tularensis in complex specimens.

Current studies are under way to evaluate the ability of the TaqMan assay for use in testing environmental and field specimens associated with outbreaks of tularemia, for bioterrorism assessment and response, and to further understanding of the ecology and transmission cycles of the organism. To date, the assay has proved to accurately detect F. tularensis in animal carcasses, urine, feces, and ticks (data not shown). Future evaluations using water, soil, and grass will also be important. For translation of this assay to the clinical diagnostic laboratory, evaluations between culturing and the multitarget TaqMan assay using clinical specimens such as bronchial fluid, blood, pleural fluid, and lymph node aspirates must be performed. In humans, antibiotic treatment can prevent the recovery of a culture, making PCR an important diagnostic tool. In the case of a bioterrorism event, for which results would be needed immediately, PCR can give rapid identification and guide further testing. While obtaining a culture is an irrefutable confirmation of the presence of live F. tularensis, PCR is an invaluable diagnostic when a sample is unculturable or contaminated or when results are needed immediately.

In conclusion, while other PCR assays exist for the detection of F. tularensis, our multitarget TaqMan PCR assay has many advantages because it consists of three genetic targets, is a real-time rapid assay, can differentially diagnose F. tularensis and F. philomiragia, and is sensitive to 1 CFU or 1 GE. In a bioterrorism event, when the testing of clinical and environmental samples would need to be done quickly and accurately for immediate assessment and response, the multitarget assay would provide an indispensable tool. It should also prove to be of great value for testing a wide variety of environmental and field samples. In the future, the ISFtu2, 23kDa, and tul4 TaqMan assays can be combined into a single multiplex reaction by labeling the probes with emitters of different wavelengths, with the caveat that the multiplex assay will need to be carefully evaluated to ensure that there is no compromise to the sensitivity. While this TaqMan PCR assay was developed on a LightCycler instrument, it can be translated upon evaluation to any real-time platform, such as the I-Cycler, SmartCylcer, or ABI 7000 instrument.