Attenuated Francisella novicida Transposon Mutants Protect Mice against Wild-Type Challenge

Abstract

Francisella tularensis is the bacterial pathogen that causes tularemia in humans and a number of animals. To date, there is no approved vaccine for this widespread and life-threatening disease. The goal of this study was to identify F. tularensis mutants that can be used in the development of a live attenuated vaccine. We screened F. novicida transposon mutants to identify mutants that exhibited reduced growth in mouse macrophages, as these cells are the preferred host cells of Francisella and an essential component of the innate immune system. This approach yielded 16 F. novicida mutants that were 100-fold more attenuated for virulence in a mouse model than the wild-type parental strain. These mutants were then tested to determine their abilities to protect mice against challenge with high doses of wild-type bacteria. Five of the 16 attenuated mutants (with mutations corresponding to dsbB, FTT0742, pdpB, fumA, and carB in the F. tularensis SCHU S4 strain) provided mice with protection against challenge with high doses (>8 × 10⁵ CFU) of wild-type F. novicida. We believe that these findings will be of use in the design of a vaccine against tularemia.

Francisella tularensis is a gram-negative, facultative intracellular pathogen that causes tularemia, a debilitating and potentially fatal disease that affects humans and a wide range of animals. Infections can be acquired through bites from an arthropod vector, skin lesions, ingestion of contaminated food or water, and, most dangerously, inhalation of as few as 10 bacteria (10). The low dose required to cause tularemia by the aerosol route resulted in the development of F. tularensis for use as a biological weapon by several national weapons programs. This has led the U.S. Centers for Disease Control and Prevention to classify F. tularensis as a category A bioterrorism agent; members of this category are considered the organisms that pose the most serious risk to national security (http://www.bt.cdc.gov/Agent/Agentlist.asp). There is currently no approved vaccine available in the United States or Europe. Thus, the development of a vaccine against F. tularensis has become an international research priority.

Although the molecular mechanisms of F. tularensis pathogenesis remain relatively obscure, it has been established that replication in human and animal macrophages is central to this organism's ability to cause tularemia (14). Several F. tularensis genes associated with intracellular growth have been identified, including iglB, iglC, mglA, pdpD, and a clpB homolog (2, 18, 20, 31, 33). Additionally, it is thought that many of the genes in the recently described F. tularensis pathogenicity island (FPI) contribute to the survival and growth of this organism in macrophages (32, 35). Of these, only iglC has been studied as the basis for a potential vaccine strain. Pammit et al. recently reported that intranasal vaccination with an F. novicida strain carrying an iglC deletion resulted in >50% protection against challenges with the wild-type organism (38). However, the capacity of mutant derivative strains with mutations in other FPI genes to confer protection against challenge with wild-type bacteria has not been studied.

Three main subspecies of F. tularensis are commonly recognized: F. tularensis subsp. tularensis (type A), F. tularensis subsp. holarctica (type B), and F. tularensis subsp. mediasiatica. All of these biotypes along with F. novicida exhibit more than 95% DNA sequence identity (3). Although type A and type B strains are highly infectious, only type A strains cause significant mortality in humans. The current live vaccine strain (LVS) is an attenuated type B strain that provides different levels of protection against challenge with type A F. tularensis strains depending on the route of immunization, the route of challenge, and the genetic background of the host (4-6, 21, 43, 49). Because the molecular basis for LVS attenuation is not known, this strain is not licensed as a tularemia vaccine.

F. novicida U112 provides an ideal model for studying Francisella pathogenesis for several reasons. While F. novicida is not considered a human pathogen, it exhibits a degree of virulence in mice similar to that of F. tularensis subspecies (27, 42). Moreover, F. novicida is easier and less dangerous to manipulate genetically than F. tularensis. In addition to the considerable genomic similarity (>95%), the close relationship between F. novicida and F. tularensis is further highlighted by their nearly identical 16S rRNA gene sequences (13). The degree of genetic identity suggests that the two organisms utilize similar virulence genes and that F. novicida is thus an apt platform for the development of a tularemia vaccine.

In this study, we used transposon mutagenesis to identify F. novicida genes required for intracellular growth. The resulting mutant strains were screened for attenuation in macrophages and mice and tested for the ability to provide protection against a wild-type challenge in mice. Five F. novicida mutant strains were found to protect mice against challenge with >8 × 10⁵ CFU of wild-type F. novicida. These results will be used in the future for construction of a Francisella vaccine.

MATERIALS AND METHODS

Bacterial strains and culture.

F. novicida strain U112 was a kind gift from Fran Nano (University of Victoria). All Francisella strains were cultured at 37°C in tryptic soy broth supplemented with 0.1% cysteine (TSBC) (Becton, Dickinson and Company, Sparks, MD) or on cysteine heart agar (CHA) (Difco/Becton, Dickinson and Company) plates. Kanamycin was added to a final concentration of 20 μg/ml to TSBC and to CHA (CHA/Kan20) for selection of U112 strains carrying the transposon. Escherichia coli Genehogs (Invitrogen, Carlsbad, CA) that were used in subcloning for sequencing were transformed according to the manufacturer's directions. Colonies containing the transposon were selected by growth at 37°C on Luria-Bertani (LB) agar plates containing 60 μg/ml kanamycin. The Salmonella strain used for Southern blotting was grown in LB broth and on LB agar plates at 37°C.

Generation of bacterial transposon mutant strains.

A library of F. novicida transposon insertion mutants was created by electroporating mini-Tn5 transposon-transposase complexes into appropriately prepared F. novicida. Although independently developed, our technique was similar to that of Kawula et al. (26, 45). The mini-Tn5 cycler transposon was constructed as previously described (17). The transposon-transposase complex was prepared as described by Goryshin et al. (19). F. novicida U112 was grown to confluence on CHA plates at 37°C and resuspended with 5 ml of ice-cold 10% glycerol-500 mM sucrose buffer. Aliquots (1 ml) were transferred to 1.5-ml microcentrifuge tubes, pelleted by centrifugation at 12,000 × g for 5 min at 4°C, and resuspended in 1 ml of buffer. This wash step was repeated until a total of four washes had been performed. After the final wash, each aliquot was resuspended in 100 μl buffer. One microliter of transposon-transposase complex was added to each tube, and the samples were electroporated in 1-mm-gap cuvettes at 1.5 to 1.7 kV, 200 Ω, and 25 μF. The bacteria were recovered in 1 ml TSBC in glass tubes for 4 h in a 37°C rotator and plated on CHA/Kan20 plates. The frequency of isolation of transposon insertion mutants was rather low (about 10 to 100 insertions per 10⁹ cells following electroporation).

Culture and infection of cell lines and primary macrophages.

The J774A.1 and RAW264.7 murine macrophage cell lines (American Type Culture Collection, Manassas, VA) were cultured in Dulbecco's modified Eagle's medium (DMEM) (Gibco-BRL, Rockville, MD) supplemented with 10% fetal bovine serum (FBS) (Gibco-BRL), 1 mM nonessential amino acids (Gibco-BRL), and 0.2 mM sodium pyruvate (Gibco-BRL) at 37°C in the presence of 5% CO₂. Bone marrow-derived macrophages (BMDM) were collected by flushing the femurs of BALB/c mice with serum-free DMEM and were cultured in DMEM supplemented with 20% L929 and an antibiotic cocktail of penicillin (10,000 U/ml) and streptomycin (10,000 μg/ml). For infection, bacteria were added to 50% confluent cells in 24- or 96-well culture dishes (Corning, Corning, NY) or four-chamber microscope plates (Nalge Nunc, Naperville, IL) at the multiplicities of infection (MOI) indicated below, and the cells were centrifuged at 1,000 × g for 5 min at room temperature and incubated at 37°C in the presence of 5% CO₂. One hour after infection, the cells were washed twice with phosphate-buffered saline (PBS), and DMEM containing 100 μg/ml of gentamicin was added to prevent the growth of any extracellular bacteria (29). Two hours after infection, the cells were washed twice with PBS and either lysed or incubated in the presence of 10 μg/ml gentamicin for an additional 22 h. Cells were lysed with TSBC containing 0.5% saponin (Sigma) for 30 min at 37°C in the presence of 5% CO₂.

Screening for reduced growth in macrophages.

RAW macrophages were seeded to obtain 50% confluence in 96-well tissue culture plates and infected (MOI of ∼1,000) with overnight cultures of F. novicida mutant strains that were grown in stationary 96-well tissue culture plates, as described above. At 24 h postinfection (p.i.), the macrophages were washed and lysed as described above. Three percent of each lysate was plated onto CHA/Kan20 plates and incubated overnight at 37°C. Mutants that exhibited growth defects were identified visually. To eliminate false positives, the potentially attenuated mutants were subjected to another round of selection by infecting RAW macrophages in 24-well plates, as described above, using an input MOI of 100, which corresponded to about one bacterium per macrophage. After lysis, 50 μl of each lysate was plated onto CHA plates and incubated overnight at 37°C. F. novicida mutants compromised for growth in macrophages were identified visually by comparison to wild-type U112 infection lysates; the attenuated mutants yielded individual colonies, while the wild-type bacteria grew to confluence.

Sequencing of mini-Tn5 insertion sites and sequence analysis.

The method described by Geddes et al. was used for sequencing mini-Tn5 insertion sites and sequence analysis (17). Briefly, chromosomal DNA from F. novicida mutants exhibiting reduced growth in macrophages was prepared (1), digested with EcoRI, and subcloned into pACYC184. Ligation reaction mixtures were electroporated into GeneHogs E. coli cells (Invitrogen) and selected for growth on LB medium containing 60 μg/ml kanamycin. Plasmids from kanamycin-resistant colonies were purified using a QIAprep Spin miniprep kit (QIAGEN, Valencia, CA) according to the manufacturer's instructions. The DNA sequence of the fusion junction was obtained using a primer complementary to bp 166 to 190 of the 5′ end of mini-Tn5 cycler (5′ GTTGACCAGGCGGAACATCAATGTG 3′). Sequence analysis was performed using the MacVector 7.2.3 software and the NCBI BLAST server at http://www.ncbi.nlm.nih.gov/BLAST/.

Mouse studies.

Six- to 8-week old female BALB/c mice were purchased from the Jackson Laboratory (Bar Harbor, ME). The animals were fed autoclaved food and water ad libitum. All experiments were performed in accordance with Animal Care and Use Committee guidelines. For vaccination and challenge studies, mice were inoculated intraperitoneally (i.p.) with bacteria in 150 μl (total volume) of PBS. Mice were vaccinated with the number of CFU indicated below. Surviving mice were challenged 28 days later with the doses indicated below. Dissemination and clearance of the bacteria were determined by harvesting the lungs, liver, and spleen on the days postinfection indicated below, homogenizing the organs with a stomacher, and plating serial dilutions. The 50% lethal doses (LD₅₀) were calculated by the method of Reed and Muench (39). Mice were checked for signs of illness or death twice each day following infection.

Bacterial growth in liquid media.

Overnight cultures of F. novicida were diluted into 10 ml of TSBC to obtain an optical density at 600 nm (OD₆₀₀) of 0.1. Optical densities were then recorded at the times specified below. Note that the cultures were diluted 1:10 to obtain OD₆₀₀ of >1 for accuracy. We previously determined by plating that an OD₆₀₀ of 1 was equivalent to approximately 4 × 10⁹ bacteria/ml.

Quantification of bacterial entry and growth in macrophages.

J774 and RAW cells and BMDM were seeded in triplicate to obtain 50% confluence in 24-well tissue culture plates and were infected as described above with F. novicida mutant strains at an input MOI of 100. Cells were lysed at 2 or 24 h p.i. Serial dilutions of the lysates were plated onto CHA/Kan20 or CHA (wild-type and mock infection controls) plates. After overnight incubation at 37°C, the colonies on each plate were counted. Means and standard deviations were calculated using Microsoft Excel X for Mac. The 24-h data were statistically analyzed by paired two-tailed t tests using Microsoft Excel X for Mac.

Southern blot analysis.

F. novicida chromosomal DNA was prepared using the cetyltrimethylammonium bromide method (1), and 250 ng of each preparation was digested to completion with HindIII. Digested DNA was electrophoresed on a 0.8% agarose gel for 2 h at 90 kV and then transferred to a positively charged nylon membrane (Roche) using a standard capillary transfer method (1). DNA was cross-linked to the membrane at 120,000 μJ/cm² using a Stratalinker 1800 UV cross-linker (Stratagene, La Jolla, CA). The digested bacterial DNA was probed with a digoxigenin-labeled probe using a DIG High Prime II DNA labeling and detection starter kit (Roche, Indianapolis, IN), and the membrane was exposed to film (Kodak, Rochester, NY) for 2 or 8 min as described below. By using a DNA probe that spans a HindIII site in the transposon and hybridizes to two separate locations of the HindIII-digested chromosomal DNA, we were able to determine the number of transposon inserts in each strain.

Cytotoxicity assay.

A cytotoxicity assay was conducted as described by van der Velden et al. (48). Briefly, J774 cells seeded in 96-well culture plates were infected in triplicate with either the transposon mutants or wild-type F. novicida U112 at an input MOI of 100. After 48 h, the supernatants were removed and assayed for released lactate dehydrogenase (LDH) using the CytoTox 96 nonradioactive cytotoxicity assay (Promega, Madison, WI). Cytotoxicity was determined for each mutant strain by calculating the amount of LDH released as a percentage of the maximal amount released from macrophages infected with wild-type strain U112.

Microscopy.

J774 cells were infected at an input MOI of 100, as previously described, in four-well chamber plates (Nalge Nunc). After 24 h, the cells were washed twice with PBS, fixed for 1 h with 4% paraformaldehyde, and stored in PBS at 4°C. After three washes for 10 min in PBS, the cells were permeabilized with 0.5% Triton X-100 (Sigma Chemical) in PBS for 20 min at room temperature, blocked with 5% FBS in PBS for 30 min, and incubated for 1 h at 4°C with a polyclonal antibody against F. tularensis (Becton, Dickinson and Company). After three washes for 10 min in PBS, the cells were again blocked with 5% FBS. A goat anti-rabbit antibody conjugated to Alexa 488 (Molecular Probes, Eugene, OR) was applied to the cells overnight at 4°C. The cells were again washed three times for 10 min in PBS and incubated with a 1:1,000 dilution of FM 4-64 membrane stain (Molecular Probes) and 1:1,000 dilution of Draq5 DNA stain in PBS (Alexis Biochemicals, San Diego, CA) for 10 min at room temperature. The cells were washed twice with PBS and mounted in Fluormount-G antifade solution (Southern Biotechnology, Birmingham, AL), and images were obtained with an Applied Precision DeltaVision deconvolution microscope system (Advanced Precision Instruments, Issaquah, WA). All images were taken obtained a ×60 objective. Stacks of 10 z-plane images that were 1 μm apart were captured at 1024 × 1024 pixels and deconvolved for seven iterations. Selected images were saved in TIFF format and imported into Adobe Photoshop to be formatted.

Complementation of disrupted genes.

Plasmid pKK202 (30) was modified to include unique NotI, SfiI, and XhoI restriction sites by digestion with ClaI and XbaI, followed by ligation with a DNA fragment. Oligonucleotides CGGCGGCCGCTTGGCCTCGAGGGCC and CTAGGGCCCTCGAGGCCAAGCGGCCGC were annealed to obtain a double-stranded product encoding the new restriction sites. Using SCHU S4 DNA as a template, full-length genes were amplified by PCR. The dsbB gene was cloned using primers GCGGCCGCCTTCTTAACGTCCACAGTTTTGTCC and GGCCCTCGAGGCCCTTTCTGATGGTTTGTCATTTCTCC, FTT0742 was cloned using primers GCGGCCGCGCAGCATTACCTGGAATTACAAG and GGCCCTCGAGGCCCAAACAGCAAATAAATATACAACACC, and fumA was cloned using primers GCGGCCGCTAGTGATAAAATTAGCGAGG and GGCCCTCGAGGCCATTAACTATAATGCCGAG. The modified pKK202 vector and the PCR products were digested with NotI and XhoI and ligated. pKK202-dsbB was electroporated into the dsbB mutant F. novicida strain and used to infect J774 and RAW cells, primary macrophages, and mice as indicated below. pKK202-FTT0742 and pKK202-fumA were similarly tested with RAW cells.

RESULTS

Transposon mutagenesis and identification of disrupted loci.

An ongoing challenge to the establishment of a suitable set of genetic tools for F. tularensis is the difficulty of creating stable mutations in the genome. To address this issue, we developed a transposon mutagenesis technique independent of phages and shuttle vectors that yielded as many as 150 transposon mutants from a single electroporation procedure (see Materials and Methods). Our method, although independently developed, is quite similar to the technique used by the Kawula laboratory with LVS (26, 45). The bacteria were electroporated with a transposase-transposon complex that completed the transposition event once it was inside the bacteria. This approach yielded the library of 779 Francisella transposon mutants used in this study.

Macrophages are the primary host cell type for Francisella in both humans and animals (14). It follows that one approach for developing a tularemia vaccine would be to discern which Francisella genes are necessary for growth in macrophages. Thus, we screened our F. novicida transposon mutant library for mutants that had a reduced ability to grow in macrophages, as described in Materials and Methods. Of the more than 700 F. novicida transposon mutants screened, 34 exhibited reduced growth in RAW macrophages. We obtained sequences for 28 of these mutant strains and identified the disrupted open reading frames (ORFs) by comparison to the SCHU S4 sequence (Table 1).

TABLE 1.

F. novicida transposon mutant strains generated in this study

Mutant	Corresponding SCHU S4 FTT no.	Gene disrupted	% Identity to SCHU S4a	Nucleotide location of transpson insertb	% Survivalc
1	FTT0107c	dsbB	100	114151	100
2	FTT0145	rpoC	99	163108	0
3	FTT0203c	purH	99	222340	100
4	FTT0334	rpsQ	100	342324	0
5	FTT0356	htpG	95	356504	100
6	FTT0504c	sucC	97	524250	0
7	FTT0583	fopA	98	599781	0
8	FTT0742	Hypotheticald	89	765155	100
9	FTT0893	purM	98	901556	100
10	FTT0893	purM	97	901647	100
11	FTT0893	purM	98	901848	100
12	FTT0894	purCD	99	904045	100
13	FTT0894	purCD	99	904160	100
14	FTT0917	maeA	98	926193	33
15	FTT1165c	aspC2	99	1179264	0
16	FTT1222	dedA2	98	1240288	33
17	FTT1241	glyA	97	11261475	67
18	FTT1269c	dnaK	100	1291446	100
19	FTT1345/1700	pdpBe	98	1384141/1777485	100
20	FTT1369c	tktA	97	1416905	100
21	FTT1535c	ocd	99	1597434	0
22	FTT1535c	ocd	97	1597841	0
23	FTT1600c	fumA	98	1667516	100
24	FTT1629c	Hypotheticalf	99	1692570	0
25	FTT1664	carB	99	1730805	100
26	FTT1720c	purL	98	1804171	100
27	FTT1720c	purL	97	1805882	100
28	FTT1769c	clpB	96	1858564	0

^aLevel of identity when our fragment sequence was used.

^bCorresponding to SCHU S4 location.

^cData for groups of three mice at 28 days after infection with 6 × 10³ CFU.

^dLipoprotein gene.

^eSCHU S4 contains two copies of pdpB.

^fMembrane protein gene.

Sixteen F. novicida mutants exhibited attenuation in mice.

To narrow our study to the mutant strains that had an attenuated phenotype in an animal model, we infected wild-type BALB/c mice with the 28 F. novicida mutants that were attenuated for growth in macrophages. Mice were inoculated intraperitoneally with 6 × 10³ bacteria in 150 μl of PBS, a dose which is about 100-fold greater than the wild-type F. novicida LD₅₀ for mice. At 28 days postinfection, the survival rate for 16 of the 28 groups of mice was 100%, indicating that 16 of the insertion mutants were highly attenuated in this animal infection model (Table 1).

Five F. novicida mutants protected mice against challenge.

The ideal living vaccine strain produces an asymptomatic infection that provides complete protection against subsequent exposure to the wild-type organism. To determine if any of our 16 attenuated F. novicida transposon mutants could confer protection against wild-type infection, we next challenged the surviving vaccinated mice with the wild-type parental strain. Four weeks after infection with mutant F. novicida strains, surviving mice were intraperitoneally challenged with 8 × 10⁵ CFU F. novicida U112. We considered this to be a very stringent challenge, as the dose was more than 10,000 times the LD₅₀ observed for wild-type infection. At 28 days after the challenge, 5 of the 16 mutants exhibited 100% protection after a single vaccination; these mutants had mutations in dsbB, the ORF corresponding to FTT0742 (referred to as FTT0742 below), pdpB, fumA, and carB (Table 2). In the same experiments, all wild-type control infections resulted in 0% survival.

TABLE 2.

F. novicida transposon mutants attenuated in mice

F. novicida mutant	% Survivala
dsbB	100
purH	0
htpG	0
FTT0742	100
purM	0
purM	0
purM	33
purCD	0
purCD	0
dnaK	0
pdpB	100
tktA	66
fumA	100
carB	100
purL	0
purL	33

^aLevels of survival after challenge with 8 × 10⁵ CFU of wild-type strain U112. Groups of three mice were inoculated intraperitoneally.

LD₅₀s for F. novicida mutants.

Additional infections of mice with the F. novicida transposon mutants were conducted to determine the LD₅₀s of the five protective strains. In our experiments, the F. novicida U112 parental strain was observed to have an LD₅₀ of 66.25 CFU (Table 3). Our carB mutant exhibited the least attenuation and had an LD₅₀ of 6.75 × 10³ CFU. The LD₅₀s for our dsbB and fumA mutants were 6.625 × 10⁵ CFU and 6.17 × 10⁵ CFU, respectively. The mutant strains with the highest levels of attenuation in an animal infection model were the FTT0742 and pdpB mutants, both of which were observed to have LD₅₀s of >6 × 10⁷ CFU. Taken together, these results showed that our five F. novicida transposon mutants were significantly attenuated in a mouse infection model compared to the wild-type parental strain.

TABLE 3.

LD₅₀s for protective F. novicida mutants

Strain	LD50 (CFU)a
Wild-type U112	66.25
dsbB	6.625 × 105
FTT0742	>6 × 107
pdpB	>6 × 107
fumA	6.17 × 105
carB	6.75 × 103

^aLD₅₀ when mice were inoculated intraperitoneally.

F. novicida mutants were highly attenuated for growth in mouse macrophage cell lines.

Wild-type strain F. novicida U112 and the five mutants that conferred protection were examined in order to quantify entry and attenuation in macrophage cell lines and primary mouse macrophages (BMDM). RAW and J774 macrophage-like cells and BMDM were infected in triplicate wells in duplicate plates with an input MOI of 100, as described in Materials and Methods. The contents of one plate containing each cell type were lysed at 2 h after infection to determine the abilities of the transposon mutants to enter host cells, and lysis of the contents of the second plates at 24 h allowed us to quantify the intracellular growth of the mutant strains.

In J774 mouse macrophage-like cells, the levels of entry were quite similar for the mutant and wild-type strains, except for the pdpB mutant (Fig. 1A). The dsbB mutant exhibited the highest level of attenuation in J774 cells, and for the FTT0742 mutant derivative there was also a decrease in the number of CFU after 24 h. Although fewer pdpB mutant bacteria than bacteria of the other strains entered J774 cells, this mutant still displayed an attenuation phenotype. For the fumA mutant there was very little difference between the level of entry and the level of replication at 24 h, indicating that there was no death, no replication, or a balance between the two. Interestingly, the carB mutant was able to replicate within J774 cells, albeit at a lower rate than the wild-type control. Each of the mutants exhibited statistically significant attenuation in J774 cells at 24 h p.i. (P < 0.005).

FIG. 1.

Five F. novicida transposon mutants are attenuated for growth in macrophages. J774 (A) and RAW (B) cell lines and mouse BMDM (C) were infected with the five F. novicida mutants and wild-type strain U112 at an MOI of 100 for 2 h and 24 h. Cells were lysed, and serial dilutions of the lysates were plated onto CHA/Kan20 (mutants) or CHA (U112 and mock infection controls). Colonies were counted, and the numbers of CFU/ml were calculated and converted to a log scale. Each column shows the average for three individual infections. No bacteria were detected in J774 or RAW cells infected with the pdpB mutant after 24 h. Asterisks indicate the statistically significant results for 24 h (J774 cells, P < 0.005; RAW cells, P < 0.05; BMDM, P < 0.01).

The levels of entry for the transposon mutants and the wild-type bacteria were similar when RAW mouse macrophage-like cells were used (Fig. 1B). Again, the dsbB and pdpB mutant derivatives exhibited the strongest attenuation phenotypes. Intriguingly, the FTT0742, fumA, and carB mutant strains all were able to replicate in RAW cells, whereas only the wild type and the carB mutant replicated in the J774 cell line. Compared to the wild type, each of the mutants strains was significantly impaired for replication inside RAW cells at 24 h p.i. (P < 0.05).

As observed for the J774 and RAW cells, the dsbB and pdpB mutants were the most attenuated strains in primary murine BMDM (Fig. 1C). Despite its high LD₅₀ for mice, the ability of the FTT0742 mutant derivative to enter or replicate within BMDM was not impaired, and both the fumA and carB strains displayed only slight attenuation. This apparent disparity underlies the differences in infection among various cell populations. It should also be noted that the BMDM were not stimulated prior to infection and were probably less microbicidal than tissue-resident macrophages in the mouse model. In BMDM, only the dsbB, pdpB, and carB strains were significantly attenuated for growth at 24 h p.i. (P < 0.01).

To visually assess the replication of F. novicida transposon mutants in macrophages, fluorescence microscopy was performed. J774 macrophages were infected and prepared for microscopy as described in Materials and Methods. As expected, macrophages infected with F. novicida U112 contained more bacteria than cells infected with our mutants contained (Fig. 2). Although several bacteria were observed inside host cells infected with the fumA mutant (Fig. 2F), infections with the dsbB, FTT0742, pdpB, and carB mutant strains resulted in only one or two intracellular bacteria at 24 h p.i. These findings corroborated our finding that the F. novicida transposon mutants were defective for replication or survival inside macrophages. Furthermore, although the macrophages were initially seeded at the same concentration, fewer cells remained in the wells after infection with wild-type strain U112 than after infection with the mutants and in uninfected controls. This observation indicated that host cell death occurred during the course of the wild-type infection but not during infection with attenuated mutant strains.

FIG. 2.

Fluorescence microscopy of macrophage cell line infections. J774 macrophages were infected in four-chamber microscope plates for 24 h at an MOI of 100. The cells were fixed in 4% paraformaldehyde, probed with a polyclonal antibody against Francisella (Alexa 488) (green), and stained with FM 4-64 (membrane not shown) and Draq 5 (DNA) (blue). Cells were imaged with an Applied Precision DeltaVision deconvolution microscope system using a ×60 objective. The arrows indicate individual bacteria. wt, wild type.

To address the possibility that the attenuation phenotypes of our F. novicida mutant strains could have been due to overall defects in replication, a simple growth curve was determined for each mutant. While the dsbB, FTT0742, and pdpB mutants replicated at levels similar to that of wild-type strain U112, both fumA and carB mutants exhibited defects in replication after 4 h of growth (data not shown). At 24-h after inoculation, the OD₆₀₀ of cultures of these two strains were approximately 1.5, and the OD₆₀₀ of cultures of the other strains ranged from 2.5 to 3.3. Interestingly, this phenotype was not rescued by supplementing the media with malate (fumA) or arginine (carB) (data not shown). It is possible that the addition of exogenous substrates could not complement the defects because the enzymes are part of multienzyme complexes from which additional intermediates are excluded. Nonetheless, the fumA and carB mutants remained potential vaccine candidates because they protected mice against wild-type challenge (Table 2).

Infection with F. novicida mutants did not reduce host cell integrity.

The observed attenuation phenotypes could have been a result of increased host cell killing, which would have yielded fewer live infected macrophages and thus fewer bacteria, as they would have been killed by the gentamicin in the extracellular media (29). One method for determining the degree of cytotoxicity that results from bacterial infection is to measure cell lysis by quantifying the release of the stable cytosolic enzyme LDH. J774 macrophages were infected with either wild-type strain U112 bacteria or one of the five mutant strains for 48 h at an input MOI of 100. The levels of LDH in the supernatants were then recorded. As shown in Fig. 3, the abilities of the five mutant strains to cause cell lysis were significantly impaired compared with the ability of wild-type F. novicida. With the levels of LDH released during wild-type infection normalized to 100%, the amounts of LDH released during infection with the five attenuated mutants ranged from 9.75% (FTT0742) to 24.52% (fumA). These results indicated that the attenuation phenotypes were not due to increased killing of host cells by the transposon mutants and that the intracellular replication of these strains was indeed compromised.

FIG. 3.

Infection with F. novicida mutants does not reduce host cell integrity. J774 macrophages were infected with the five F. novicida mutants and wild-type strain U112 (wt U112) at an MOI of 100 for 48 h. The levels of LDH in the extracellular medium were determined. The level of LDH release for the wild-type infection was defined as 100%, and the levels of LDH release for the five mutant strains were normalized to this level. Each column shows the average for three individual infections.

Each F. novicida mutant harbored a single transposon insertion.

Transposon mutagenesis has the potential to produce strains with more than one transposon insertion and thus multiple causes for an observed phenotype. Each of our five protective F. novicida transposon mutants was subjected to Southern blot analysis to ensure that the attenuation phenotypes of the mutant strains were the result of a single transposition event. To determine the number of inserts, a DNA probe that spanned a unique HindIII site in the transposon was designed so that digestion of chromosomal DNA harboring a single transposon insert would yield two targets for this probe. Chromosomal DNA from each of the five protective mutant strains, as well as wild-type strain U112 and a Salmonella strain known to contain a single copy of the transposon, was prepared and probed as described in Materials and Methods. The presence of two bands demonstrated that the F. novicida mutant strains each harbored a single copy of the transposon insert, as shown in Fig. 4.

FIG. 4.

Each F. novicida transposon mutant harbors a single transposon. Chromosomal DNA preparations from a Salmonella strain known to carry the mini-Tn5 cycler transposon (lane 1), wild-type strain U112 (lane 2), and the dsbB (lane3), FTT0742 (lane 4), pdpB (lane 5), fumA (lane 6), and carB (lane 7) mutant strains were digested with HindIII. The DNA was transferred to a membrane and probed with a digoxigenin-labeled DNA probe that spanned a HindIII site in the transposon. The membrane was exposed to film for 2 min (lanes 1, 2, 3, 5, and 7) or 8 min (lanes 4 and 6). The presence of two bands indicates a single transposon insertion event.

F. novicida mutants disseminated to the liver, spleen, and lungs and were subsequently cleared.

Acceptable vaccine candidates ideally infect mice transiently and are cleared before challenge with the parent strain. We inoculated groups of 15 BALB/c mice i.p. with 0.1 LD₅₀ of each mutant. Thus, for these infections, the vaccination dose varied from strain to strain. Three mice from each group were sacrificed at 1, 3, 5, 7, and 28 days after vaccination, and their spleen, liver, and lungs were harvested. As shown in Fig. 5, each mutant, with the possible exception of the carB mutant, disseminated to all three organs (spleen, liver, and lungs) from the original site of inoculation. Two of the five strains, the dsbB and fumA mutants, were completely cleared by day 28 following infection. Although relatively low numbers of bacteria remained in the spleen at day 28 after infection with the FTT0742 and pdpB mutants, it is possible that these organisms would have been cleared in the vaccination experiments because a lower dose (10- to 1,000-fold-fewer bacteria) was used for that procedure (Tables 2 and 4).

FIG. 5.

F. novicida mutants disseminate and are subsequently cleared. The dsbB (A), FTT0742 (B), pdpB (C), fumA (D), and carB (E) mutant derivatives were injected i.p. into BALB/c mice, using 0.1 LD₅₀. The spleen (⧫), liver (▪), and lungs (▴) were harvested at 1, 3, 5, 7, and 28 days p.i. Bacteria were liberated from the tissues and enumerated. Each point shows the results for three mice.

TABLE 4.

Results of challenge studies after vaccination with F. novicida transposon mutants

Mutant strain	Vaccine dose (CFU)	% Survival (n = 5)	Challenge dose (CFU)	% Survival
dsbB	6 × 105	100	6 × 107	100
	6 × 106	20	6 × 107	100
	6 × 107	0	NDa	ND
FTT0742	6 × 105	100	6 × 107	100
	6 × 106	100	6 × 107	100
	6 × 107	100	6 × 107	100
pdpB	6 × 105	100	6 × 107	100
	6 × 106	100	6 × 107	100
	6 × 107	100	6 × 107	100
fumA	6 × 105	100	6 × 107	100
	6 × 106	0	ND	ND
	6 × 107	0	ND	ND

^aND, not determined.

Expressing full-length genes in trans complemented the attenuation phenotype.

While the Southern hybridization experiments strongly indicated that each mutant derivative contained only a single transposon insertion, we wished to determine if cloned copies of the genes could complement the observed virulence defects. This would provide as additional evidence that the attenuation phenotype of each strain was a result of a single mutation, marked by the transposon insertion.

The dsbB gene was amplified from SCHU S4 DNA by PCR and cloned into plasmid pKK202 (30). Following transformation into the dsbB mutant, the abilities to replicate within macrophages and cause disease in mice were determined. As shown in Fig. 6, in trans expression of the cloned dsbB gene provided nearly complete complementation of the virulence defect in three different cell types. Further analysis showed that the LD₅₀ was 60.25 CFU, which is comparable to the wild-type LD₅₀ (66.25 CFU).

FIG. 6.

Expressing full-length genes in trans complements the attenuation defects in cells. The levels of entry (2 h) and replication (24 h) were determined for wild-type strain U112 (wt U112), the dsbB mutant, and the dsbB mutant complemented with pKK202-dsbB in the J774 (A) and RAW (B) cell lines and in primary BMDM (C). Entry and replication rates in RAW cells were determined for complementation of the FTT0742 (D) and fumA (E) mutants. Each column shows the average for three separate infections.

In a parallel experiment, in trans expression of the full-length FTT0742 gene in the corresponding mutant derivative resulted in incomplete complementation. In RAW cells, intracellular replication of the complemented FTT0742 strain was 10-fold greater than intracellular replication of the mutant, but the value was still nearly 2 orders of magnitude less than the wild-type value (Fig. 6D).

Like complementation of the dsbB mutation, complementation of the fumA mutation with the full-length gene restored the level of intracellular growth to the level of wild-type F. novicida (Fig. 6E). Taken together, these findings show that the observed attenuation phenotypes were due to mutations in dsbB, FTT0742, and fumA.

Complementation of pdpB will be attempted in the future, as the transposon insertion is located in the second gene of a 12-gene operon and is undoubtedly polar on expression of downstream genes. Based on its in vitro growth defect, comparatively low LD₅₀, relative lack of intracellular attenuation, and questionable dissemination patterns, we felt that the carB mutant was not a strong enough candidate to include in further development of a vaccine against tularemia, and therefore we did not attempt to complement the carB gene.

Mutant strains protected mice against very high doses of wild-type bacteria.

To further assess the level of protection provided by the F. novicida transposon insertion mutants, we challenged vaccinated mice with higher doses of the wild-type U112 parental strain. The dsbB, FTT0742, pdpB, and fumA mutant strains were used to infect groups of five mice, and the doses used were 6 × 10⁵, 6 × 10⁶, and 6 × 10⁷ CFU (Table 4). Mice infected with each of the three doses of our FTT0742 and pdpB mutants had a survival rate of 100%, as did the animals infected with the lowest doses of the dsbB and fumA mutants. Four weeks after vaccination, surviving animals were challenged with 6 × 10⁷ CFU of wild-type strain F. novicida U112, which is approximately 10⁶ times the observed LD₅₀ for wild-type infection. All of the mice challenged survived without any symptoms of disease. These results demonstrated that four of our F. novicida transposon mutants were capable of protecting mice against infection with very high levels of the wild-type organism. Overall, our findings indicate that Francisella strains carrying mutations in these genes are candidates for a vaccine against tularemia.

DISCUSSION

The classification of F. tularensis as a category A bioterrorism agent by the Centers for Disease Control and Prevention demonstrates that this organism is widely acknowledged to be a potential threat to national security. Thus, there is an immediate need for an approved tularemia vaccine. The lack of genetic tools with which to manipulate F. tularensis remains a great barrier to the development of such a vaccine. Hence, we developed a transposon mutagenesis technique to create random insertions in the F. novicida genome and analyzed the resulting mutant strains for intracellular growth defects in macrophages, attenuation in mice, and the ability to provide protection against wild-type infection. We identified 28 F. novicida transposon mutants that have a defect in intracellular growth in macrophage cell lines, 16 of which exhibited 100% attenuation in mice at a dose that was more than 100 times the wild-type LD₅₀. When mice were challenged with the wild-type organism, five transposon mutant strains were found to protect the mice against infection with high doses of parental F. novicida strain U112. The disrupted genes in our five protective F. novicida mutants correspond to dsbB, FTT0742, pdpB, fumA, and carB in F. tularensis strain SCHU S4.

Disulfide bond formation protein B is encoded by dsbB. This integral membrane protein is part of a pathway that leads to disulfide bond formation between cysteine residues in periplasmic proteins in E. coli and other bacteria (23). The functional folded conformation of a protein often relies upon correct disulfide pairing of the cysteine residues. Thus, one explanation for why our dsbB mutant strain is attenuated is that a protein(s) required for replication inside host cells does not achieve its active conformation. It is also exciting to speculate that, with its potential influence on periplasmic proteins, the dsbB gene product may be involved in the secretion of virulence factors, possibly to ensure correct folding of components of a secretion apparatus.

The FTT0742 ORF codes for a hypothetical lipoprotein that is predicted to have transmembrane regions. Therefore, it is possible that the gene product is a component of the F. novicida cell wall and/or may be involved in molecule transport. Because in vitro growth and entry into the host cell were not compromised, we speculate that FTT0742 affects a function necessary for virulence and growth inside macrophages. Characterization of the FTT0742 protein should further clarify its role in virulence.

The product of the pdpB gene is an uncharacterized protein encoded on the FPI that exhibits some similarity to the conserved bacterial protein IcmF (http://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi). It has been shown that icmF is required for Legionella pneumophila intracellular growth, so we hypothesized that pdpB plays a similar role in F. novicida intracellular growth (50). Also of note, this gene exhibits some similarity to the genes encoding rhoptry proteins of the parasite Plasmodium, which mediate attachment to host red blood cells (40). Although the pdpB mutant did not exhibit defects during in vitro growth, its ability to enter host cells was significantly compromised and it displayed prominent intracellular attenuation since no CFU were detected in the lysates of RAW and J774 cells at 24 h p.i. Because pdpB is the second ORF in a 12-gene operon, it is likely that the transposon insert in this gene has polar effects on downstream genes. These findings indicate that pdpB or other genes in the pdp operon are needed for both entry into and replication within host cells. So far, the functions of the genes in the pdp operon have not been elucidated.

Fumarate hydratase A, the component of the Kreb's cycle (citric acid cycle) that converts fumarate to malate, is encoded by fumA (46). The citric acid cycle is one of the three metabolic pathways of cellular respiration and is necessary for fuel catabolism and ATP production. Precursory molecules for compounds such as amino acids are also generated by the citric acid cycle. Thus, the observation that our fumA mutant exhibited lower levels of in vitro replication than wild-type strain U112 exhibited may have been a result of energy deficiency or a lack of molecules needed for replication.

The carB gene encodes the large subunit of the heterodimeric enzyme carbamoyl phosphate synthase, which is required for pyrimidine biosynthesis (28). As pyrimidines are required for replication, it is clear that a mutation in this biosynthesis pathway would lead to both in vitro and intracellular growth defects. Indeed, we observed an in vitro growth defect with this mutant strain. Nevertheless, our carB mutant was able to infect macrophages and protect mice against challenge with the wild-type organism, which suggests that strains with mutations in the pyrimidine pathways of F. novicida may be useful as potential vaccine strains. However, compared to the other F. novicida mutants used in this study, the carB mutant had a lower LD₅₀ and less intracellular attenuation and did not appear to disseminate from the initial site of infection. For these reasons, we decided not to pursue use of this mutant as a possible vaccine candidate.

Although we observed only partial complementation of the FTT0742 mutant phenotype, we still consider this strain to be a vaccine candidate. Complementation experiments similar to those described in this paper carried out with Salmonella have also resulted in mixed success because the copy number of the plasmid, as well as the regulation of the gene itself, can influence complementation. In fact, the Forsberg group recently demonstrated the importance of correct gene regulation during complementation in Francisella with their studies of pilA; there was functional complementation in cis, but expression of PilA was barely detectable in the strain complemented in trans (12). The problem was further compounded in our studies because the F. novicida genome has not been published yet; therefore, the genes used for complementation were amplified from SCHU S4 DNA and may be incompatible with F. novicida due to variations between the subspecies.

Intriguingly, another finding of this study was the lack of protection conferred by pur mutants in a murine model. It has been postulated previously that mutations affecting the F. tularensis purine synthesis pathway could be used to generate a live attenuated tularemia vaccine (25). In fact, defined allelic replacement mutants with mutations that disrupt this pathway have been used to produce vaccine strains attenuated for replication in host cells in a variety of other bacterial species (36). Our F. novicida transposon library contained eight unique pur mutants, including mutants with mutations in purA, a purCD fusion (two strains), purL (two strains), and purM (three strains). Each of these strains exhibited 100% attenuation in mice when 6 × 10³ CFU was used, yet all of them failed to protect mice against a wild-type parental challenge with 8 × 10⁵ CFU. Although these results do not necessarily eliminate purine biosynthesis mutants as potential live vaccine strains, it was noteworthy that these mutants did not protect against challenge with wild-type F. novicida strain U112.

As an alternative to live vaccines, the possibility of developing a subunit vaccine must also be examined. Studies in which the efficacy of whole killed cells as a crude tularemia vaccination was evaluated demonstrated the feasibility of a subunit vaccine and prompted research into identification of antigens that induce protective immunity (8, 24). Although several polysaccharide and protein components of F. tularensis have been shown to react with convalescent-phase sera or T cells (22), the only antigen that has exhibited the ability to induce a protective immune response against tularemia is lipopolysaccharide (36). However, the protection was effective only against F. tularensis subsp. holarctica strains and was incomplete (7, 15, 16). Protection against a highly virulent F. tularensis type A strain likely depends on a Th1-mediated cellular immune response (44). So far, there is no way to administer antigens or killed bacteria that are as effective as a living attenuated vaccine, a point that has been highlighted by work in the Pamer laboratory showing that the immune system distinguishes between living and dead bacteria (37).

In contrast, a live attenuated vaccine would be effective in inducing the appropriate cellular responses (36). In fact, the type B LVS strain provides the only current means of tularemia vaccination. However, several limitations prevent the licensing of this vaccine. Foremost among these constraints is the fact that the genetic basis of LVS attenuation and protection remains unknown. Second, culturing LVS under certain conditions can lead to poorly immunogenic colony variants, which demonstrates this organism's inherent genetic instability (9, 11). Also, this vaccine does not provide protection to every individual vaccinated (34, 41). Finally, LVS protection against aerosol challenge is variable and depends on the route of immunization, as well as the host (4-6, 43). The last point is especially critical when F. tularensis is considered as a biological weapon, as aerosol dispersal is the most likely route of delivery. Taken together, these limitations clearly show that the development of an approved tularemia vaccine requires the development of a rationally attenuated, nonreverting live vaccine strain.

While the manuscript was in preparation, the Conlan Sjöstedt groups published the first description of a defined gene deletion mutant of a type A strain that protected mice against challenge with the wild-type organism (47). Indeed, the ideal live vaccine strain would be derived from a virulent F. tularensis strain; however, we used F. novicida as a model for preliminary analysis of potential tularemia vaccines because this taxon is more amenable to genetic manipulation without the danger of infection. Furthermore, because all of the Francisella taxa are closely related (3), genes necessary for intracellular growth in F. novicida are likely to have the same function in F. tularensis. Consequently, F. novicida strains provide relatively safe, genetically significant organisms with which to conduct exploratory investigations prior to studies with the more virulent F. tularensis subspecies.

Here, we describe discovery of five F. novicida transposon mutants that exhibit attenuation in macrophages and are capable of protecting mice against infection with the wild-type parental strain at doses that are up to 10⁶ times the observed wild-type LD₅₀. An approved tularemia vaccine must be a highly attenuated nonreverting derivative of a type A strain, as such strains are most likely to be used in a bioterrorism attack and there is no certainty that one taxon will protect against another. Accordingly, we will now focus on extending this work to Francisella type A strains by creating deletions of each of the genes identified here and assaying for virulence in a mouse model.