Role of Francisella Lipid A Phosphate Modification in Virulence and Long-Term Protective Immune Responses

Abstract

Lipopolysaccharide (LPS) structural modifications have been shown to specifically affect the pathogenesis of many Gram-negative pathogens. In Francisella, modification of the lipid A component of LPS resulted in a molecule with no to low endotoxic activity. The role of the terminal lipid A phosphates in host recognition and pathogenesis was determined using a Francisella novicida mutant that lacked the 4′ phosphatase enzyme (LpxF). The lipid A of this strain retained the phosphate moiety at the 4′ position and the N-linked fatty acid at the 3′ position on the diglucosamine backbone. Studies were undertaken to determine the pathogenesis of this mutant strain via the pulmonary and subcutaneous routes of infection. Mice infected with the lpxF-null F. novicida mutant by either route survived primary infection and subsequently developed protective immunity against a lethal wild-type (WT) F. novicida challenge. To determine the mechanism(s) by which the host controlled primary infection by the lpxF-null mutant, the role of innate immune components, including Toll-like receptor 2 (TLR2), TLR4, caspase-1, MyD88, alpha interferon (IFN-α), and gamma interferon(IFN-γ), was examined using knockout mice. Interestingly, only the IFN-γ knockout mice succumbed to a primary lpxF-null F. novicida mutant infection, highlighting the importance of IFN-γ production. To determine the role of components of the host adaptive immune system that elicit the long-term protective immune response, T- and B-cell deficient RAG1^−/− mice were examined. All mice survived primary infection; however, RAG1^−/− mice did not survive WT challenge, highlighting a role for T and B cells in the protective immune response.

INTRODUCTION

The Gram-negative bacterium Francisella tularensis subsp. tularensis causes the human disease tularemia following infection through a variety of routes: skin contact, arthropod bite, or inhalation of aerosolized bacteria. F. tularensis subsp. tularensis is classified as a category A bioterrorism agent by the Centers for Disease Control (CDC) due to its low infectious dose and ability to be disseminated by an airborne route. Four Francisella species cause disease in mammals: F. tularensis subsp. tularensis (type A), F. tularensis subsp. holarctica (type B), F. tularensis subsp. mediasiatica, and Francisella novicida. Both type A and type B Francisella organisms can cause lethal human disease, whereas F. novicida causes a lethal disease in mice but is nonvirulent in immunocompetent humans. There also is an atypical type B live vaccine strain (LVS) that is attenuated in humans but virulent to mice by certain infection routes. All Francisella species share more than 95% DNA sequence homology; this close relationship allows F. novicida to serve as a representative biosafety level 2 (BSL-2) model organism for F. tularensis subsp. tularensis infection (18, 23, 31).

Recognition of lipid A, the biologically active component of lipopolysaccharide (LPS), is the cause of many pathological responses to Gram-negative bacterial infections. Lipid A from Escherichia coli is biphosphorylated, hexa-acylated with fatty acid chains of 12 to 14 carbons in length, and able to induce strong inflammatory responses through interaction with Toll-like receptor 4 (TLR4) (24, 29, 30, 34). In contrast, lipid A species that are tetra- or penta-acylated have significantly decreased or absent immunostimulatory activity (11, 14, 32). As previously shown, Francisella lipid A molecules, which are hypoacylated, lack immunostimulatory activity and are not recognized by TLR2 or TLR4 (4, 6, 15).

Unlike LVS, lipid A from both F. tularensis subsp. tularensis and F. novicida consists of a β-(1′-6)-linked glucosamine disaccharide backbone with amide-linked fatty acids at the 2 and 2′ positions and an ester-linked fatty acid at the 3 but not the 3′ positions (Fig. 1A). An additional secondary fatty acid is attached to the 2′ fatty acid, forming an acyloxyacyl group. Francisella lipid A has a single phosphate moiety at the 1 position; this phosphate can be further modified by the addition of a positively charged sugar, galactosamine. The 4′ phosphate normally present in most Gram-negative bacteria is missing in Francisella as it is removed by the 4′-phosphatase enzyme (LpxF). Deletion of lpxF (FTN_0295) results in a penta-acylated lipid A with phosphate groups at both the 1 and 4′ positions (Fig. 1B). An F. novicida mutant lacking the lpxF gene (lpxF-null F. novicida mutant) has been shown to be avirulent in the footpad injection model (35).

Fig 1.

Lipid A structural analysis of WT F. novicida and lpxF-null mutant. MALDI-TOF MS of the lipid A of F. novicida U112 (A), F. novicida lpxF-null mutant (B), and the F. novicida lpxF-null mutant complementation (C). Structures of dominant ion species are shown as insets.

However, given the importance of arthropod bites and inhalation in the transmission of Francisella, the virulence and immunogenicity of this mutant need to be further characterized in more representative routes of infection. In this work, we demonstrated that the lpxF-null F. novicida mutant is avirulent by both the pulmonary and subcutaneous routes of infection. Furthermore, we showed that inoculation with the lpxF-null F. novicida mutant protected mice from a lethal challenge with wild-type (WT) F. novicida.

MATERIALS AND METHODS

Mice.

C57BL/6 female mice (age, 6 to 8 weeks) were purchased from Charles River Laboratories (Wilmington, MA). TLR2/TLR4/caspase-1 triple knockout, MyD88/interferon-alpha receptor (IFNAR) double knockout, and recombination-activating gene-1 knockout (RAG-1^−/−) mice were obtained from Christopher B. Wilson (University of Washington, Seattle, WA). MyD88^−/− mice were provided by S. Akira (Osaka, Japan). Gamma-interferon-deficient (IFN-γ^−/−) and B cell-deficient (μMT^−/−) mice were purchased from The Jackson Laboratory (Bar Harbor, ME). All genetic knockout mice were on a C57BL/6 background. All mice were housed in a specific-pathogen-free facility. All protocols were approved by the Institutional Animal Care and Use Committee of the University of Washington (Seattle, WA) and the Institutional Animal Care and Use Committee of the University of Maryland, Baltimore (Baltimore, MD).

Bacterial strains and preparation.

Wild-type F. novicida strain U112 was obtained from Francis Nano (University of Victoria, Canada). Type A F. tularensis subsp. tularensis strain FT1D (SchuS4) and type B F. tularensis subsp. holarctica strain FSC200 were obtained from Anders Sjöstedt (Umeå University, Sweden). The lpxF-null F. novicida and mglA-null mutants were generated previously (19, 35). Complementation was accomplished using the expression plasmid pMP831 (21) containing the gene FTN_0295 (lpxF) under the control of the promoter upstream of FTN_1480. Complementation restored both normal lipid A profiles and lethality in mice. Type A F. tularensis subsp. tularensis strain FT4D was a clinical isolate from the Washington State Department of Health. All bacterial strains were grown at 37°C in tryptic soy broth/tryptic soy agar supplemented with 0.1% cysteine. Log-phase bacterial cultures were used to inject mice, as described previously (17). Briefly, 1 ml of log-phase bacterial culture was resuspended in phosphate-buffered saline (PBS), and the optical density at 600 nm (OD₆₀₀) was measured and adjusted with PBS to achieve the desired bacterial concentration.

Confirmation of lipid A structures.

The lpxF-null F. novicida mutant's lipid A structure was confirmed by matrix-assisted laser desorption ionization–time of flight (MALDI-TOF) mass spectrometry, as described previously (10). Briefly, an overnight culture of Francisella was resuspended in 400 μl of isobutyric acid and 1 M ammonium hydroxide (5:3, vol/vol) and incubated at 100°C for 2 h. After a cooling step, the sample was centrifuged for 15 min at 2,000 × g, and the supernatant was collected and diluted 1:1 (vol/vol) with water. This dilution was frozen and lyophilized overnight. The dry material was then washed twice with 400 μl of methanol. Insoluble lipid A was extracted in 200 μl of a mixture of chloroform, methanol, and water (3:1:0.25, vol/vol/vol). One microliter of this extract was then spotted onto a MALDI plate followed by 1 μl of 5-chloro-2-mercaptobenzothiazole (CMBT) matrix and air dried. Samples were analyzed on a Bruker AutoFlex II (Bruker Daltonics, Billerica, MA) mass spectrometer, which was calibrated using Agilent Tuning Mix (Agilent Technologies, Foster City, CA).

Macrophage infection in vitro assays.

MHS or THP-1 cells were grown in RPMI medium supplemented with 10% fetal bovine serum (FBS) to ∼75% confluence and then plated at 8 × 10⁴ cells/well in 96-well flat-bottom plates. THP-1 cells were then differentiated with 50 ng/ml of phorbol myristate acetate (PMA) for 24 h. MHS cells were allowed to adhere overnight. Nonadherent cells were aspirated, and medium was replaced with serum-free RPMI medium. Bacteria were grown as described previously (17) and prepared for infection. Upon addition of bacteria, plates were centrifuged at 400 × g for 10 min at 4°C to favor bacteria-macrophage contact. Cultures were then placed at 37°C for 1 h. The medium was then removed and replaced with RPMI medium containing 10 μg/ml of gentamicin to kill any noninternalized bacteria. After 30 min the antibiotic medium was removed, cells were washed three times with serum-free RPMI medium, and fresh RPMI medium was replaced. Cells were then returned to 37°C. At the indicated time points (3 and 24 h), the culture medium was harvested, and macrophages were lysed with 0.5% saponin. This was combined with the medium and diluted for enumeration by serial dilution.

Virulence and protection studies.

Subcutaneous and pulmonary infections of mice with F. novicida U112 (WT) or lpxF-null F. novicida mutant bacteria were carried out as described previously (17). Briefly, for subcutaneous infection, mice were injected with ∼10⁶ CFU of lpxF-null F. novicida mutant or ∼200 CFU of WT F. novicida. For pulmonary infection, mice were exposed to aerosolized lpxF-null F. novicida mutant (∼10⁵ CFU) or WT F. novicida (∼400 CFU) using a nose-only chamber unit (In-Tox, Moriarty, NM). For protection studies, mice were initially primed with lpxF-null F. novicida mutant on either days −28 and −14 or on day −28 only and subsequently challenged with a lethal dose of WT F. novicida on day 0. Survival of the mice was recorded twice daily.

Bacterial burden.

Mice were injected subcutaneously with the lpxF-null F. novicida mutant and euthanized at the time points indicated in Fig. 4. Spleens and livers were harvested, homogenized, serially diluted, and plated on tryptic soy agar plates supplemented with 0.1% cysteine. Plates were incubated at 37°C for 24 to 72 h prior to determination of CFU counts.

Fig 4.

Bacterial burden in lpxF-null F. novicida mutant-infected mice. Bacterial burden in C57BL/6 liver (A and C) and spleen (B and D) after subcutaneous infection with the lpxF-null F. novicida mutant (5.7 ×10⁶ CFU). Each data point represents one mouse. Data are representative of three independent experiments.

In vivo immune cell depletion.

To deplete neutrophils, 250 μg of anti-Ly-6G monoclonal antibody (clone RB6-8C5) was administered intraperitoneally 24 h before infection with lpxF-null F. novicida mutant bacteria, and survival of the mice was recorded. For depletion of natural killer (NK) cells, 300 μg of anti-NK1.1 monoclonal antibody (clone PK-136) was given on days −2, 0, 5, 10, 15, 20, and 25 of the infection course. Depletion of T cells was accomplished by intraperitoneal injection of 250 μg of anti-CD4 (clone GK1.5) or anti-CD8 (clone 2.43) monoclonal antibody to deplete CD4⁺ or CD8⁺ T cells, respectively, 24 h before infection. All antibodies used for in vivo cell depletion were purchased from BioXCell (West Lebanon, NH).

Flow cytometry analysis.

Peripheral blood and spleen mononuclear cells were stained with a panel of murine markers and analyzed with a Becton-Dickson LSRII flow cytometer (BD Biosciences, San Jose, CA) and FlowJo software (Tree Star, Ashland, OR). Antibodies to murine CD3e (clone 145-2C11), CD4 (clone GK1.5), CD8a (clone 53-6.7), NK1.1/CD161c (clone PK-136), and Gr-1 (clone RB6-8C5) were purchased from eBioscience (San Diego, CA).

Sensitivity to cationic antimicrobial peptides (AMPs).

Susceptibility of the lpxF-null F. novicida mutant and WT F. novicida to polymyxin B and colistin was determined using a commercial gradient strip assay (Etest; AB bioMérieux, Solna, Sweden) according to the manufacturer's instructions.

Heterologous Francisella challenge.

C57BL/6 mice were immunized with the lpxF-null F. novicida mutant and later challenged with lethal doses of F. tularensis subsp. tularensis (type A strains SchuS4 and FT4D) or F. tularensis subsp. holarctica (type B strain FSC200) for the heterologous protection study or with F. novicida (U112) for homologous protection. Survival was recorded twice daily.

Statistical analysis.

All statistical analysis was performed using Prism, version 4 (GraphPad Software Inc., San Diego, CA). A log rank test was used for statistical analysis. Group differences were considered significant at a P value of <0.05.

RESULTS

An lpxF-null F. novicida mutant fails to replicate in macrophages in vitro.

Francisella has the ability to replicate in a variety of cell types, including epithelial cells and macrophages. Previously, F. novicida mutants (mglA null, acpA null, and iglC null) showed a decreased ability to replicate in macrophages compared to WT F. novicida (19, 25, 26). To determine the ability of the lpxF-null F. novicida mutant to replicate in macrophages, a murine alveolar cell line (MHS) and a human monocytic cell line (THP-1) were infected with WT U112 F. novicida, lpxF-null F. novicida, and mglA-null F. novicida, a known macrophage replication-deficient mutant, initially at a multiplicity of infection (MOI) of 50. While WT F. novicida was able to replicate in both cell types, the lpxF-null and mglA-null mutants failed to replicate (Fig. 2). However, by 3 h postinfection, recovery of the lpxF-null mutant was significantly lower than that of the WT (P = 0.0488 in MHS cells; P = 0.0298 in THP-1 cells) or the mglA-null mutant (P = 0.0268 in MHS cells; P = 0.0019 in THP-1 cells). This suggests that the lpxF-null mutant was cleared at a higher rate than the mglA mutant. To determine if decreased uptake or phagocytosis was responsible for the lack of recovery, the lpxF-null mutant was added at a range of MOIs (50 to 400) to normalize the initial uptake levels. While an MOI of 400 resulted in the recovery of equivalent levels of lpxF-null mutant bacteria equal to the level of WT F. novicida (MOI 50) at the 0-h time point, recovery of the lpxF-null mutant was decreased compared to both WT and mglA mutant levels at 3 h, and no lpxF-null mutant bacteria were recovered after 24 h (data not shown), suggesting that not only is the lpxF-null bacteria phagocytosed at a lower rate, it is also rapidly cleared from the macrophages. Lack of macrophage replication may be a result of the lpxF-null mutant growing more slowly in broth culture than the WT (see Fig. S1 in the supplemental material) though slow broth growth cannot be strictly equated to a lack of replication in macrophages (1).

Fig 2.

The lpxF-null F. novicida mutant fails to replicate in macrophage cell lines. Murine alveolar MHS cells (A) or human monocytic THP cells (B) were infected at an MOI of 50 for 1 h, and viable bacteria were recovered at the indicated time points and enumerated. Each bar is an average of triplicate wells. Data are representative of three independent experiments.

Avirulence of an lpxF-null F. novicida mutant in subcutaneous and pulmonary infection models.

The lpxF-null F. novicida mutant was previously shown to be avirulent using a mouse footpad injection model (35). In order to mimic more natural routes of Francisella infection, we conducted experiments in mice using the subcutaneous and pulmonary routes of infection, corresponding to arthropod bite and airborne spread, respectively. In mice, the lethal dose of F. novicida is <10 CFU by both the subcutaneous and pulmonary routes of infection (19). C57BL/6 mice infected with 10⁵ to 10⁶ CFU (10,000× to 100,000× the 100% lethal dose [LD₁₀₀]) of the lpxF-null F. novicida mutant by either infection route survived (Fig. 3A and B) and showed no signs of disease (weight loss, piloerection, or hunching). Additionally, mice challenged subcutaneously with 5 × 10⁷ CFU (5,000,000× LD₁₀₀) showed no signs of disease or virulence (data not shown), indicating the severe defect in pathogenesis of this mutant. As a control, mice infected subcutaneously with WT F. novicida (∼400 CFU) succumbed by day 2, while mice infected by the pulmonary route with WT F. novicida (∼500 CFU) died by day 4 postinfection. To confirm the role of the LpxF enzyme in virulence, mice were infected subcutaneously with the complemented strain lpxF-null (pMP831-lpxF); these mice died over the normal time course of 3 to 5 days (data not shown). Since survival rates after infection by both the subcutaneous and pulmonary routes were comparable, further experiments were conducted using the subcutaneous route, unless otherwise stated.

Fig 3.

The lpxF-null F. novicida mutant is avirulent in mice. (A) C57BL/6 (n = 4 each group) mice were inoculated twice with the lpxF-null F. novicida mutant (day 0, 1.2 × 10⁶ CFU; day 14, 1.1 × 10⁶ CFU) or once with WT F. novicida (day 0, 432 CFU) by the subcutaneous route. (B) C57BL/6 (n = 4 each group) mice were inoculated twice with the lpxF-null F. novicida mutant (day 0, 1.1 × 10⁵ CFU; day 14, 1.9 × 10⁴ CFU) or once with WT F. novicida (day 0, 547 CFU) by the pulmonary route. Data are representative of three independent experiments.

Bacterial dissemination after lpxF-null F. novicida mutant infection.

To determine the kinetics of bacterial replication and spread following infection, C57BL/6 mice were infected subcutaneously with ∼10⁶ CFU of the lpxF-null F. novicida mutant, and the bacterial burdens in the liver and spleen were determined at the time points indicated on Fig. 4A and B. As initial short-term experiments showed incomplete clearance by 6 days postinfection, a second group was challenged and followed for an extended time (Fig. 4C and D). As shown in both experiments, lpxF-null F. novicida mutant bacteria was observed to be steadily declining and became undetectable in all organs by 14 days postinfection, thus indicating that a chronic infection or carrier state was not established.

Increased susceptibility of the lpxF-null F. novicida mutant to antimicrobial peptides.

Components of the host innate immune system, specifically antimicrobial peptides, target and disrupt the membrane of bacteria through electrostatic interaction. As the overall phosphate content of the lpxF-null F. novicida mutant is increased (Fig. 1B), we used an antimicrobial peptide gradient assay to determine the sensitivity of the lpxF-null F. novicida mutant to killing by positively charged hydrophobic peptides (polymyxin B and colistin). The lpxF-null F. novicida mutant was shown to be significantly more sensitive to these peptides than WT F. novicida, with P values of 0.002 and 0.001 for polymyxin B and colistin, respectively (Table 1). This increased sensitivity indicates that altering the overall charge and lipid content of the outer membrane alters resistance to innate immune killing mechanisms.

Table 1.

Sensitivity of the lpxF-null F. novicida mutant and WT F. novicida to cationic antimicrobial peptides

F. novicida strain	MIC ± SEM (μg/ml)
	Polymyxin Ba	Colistinb
WT	54.4 ± 12.0	36.0 ± 7.7
lpxF-null mutant	0. 5 ± 0.1	0.6 ± 0.1

^aP = 0.0020 in a comparison of the WT and mutant.

^bP = 0.0012 in a comparison of the WT and mutant.

Clearance of lpxF-null F. novicida mutant is not mediated by TLR2, TLR4, or MyD88 signaling.

We have previously shown that Francisella LPS/lipid A is not recognized by the murine TLR2 or TLR4 systems (15). We investigated whether the altered lipid A structure of the lpxF-null F. novicida mutant allowed it to be sensed by the TLR2 or TLR4 system, facilitating clearance of these bacteria. LPS from both the F. novicida strain and the lpxF-null mutant failed to stimulate production of either interleukin-8 (IL-8) or tumor necrosis factor alpha (TNF-α) from THP-1 cells. As the literature indicates that human and murine TLR4 can respond differently to altered lipid A structures (14), we also tested the ability of the WT and lpxF-null lipid A in MHS cells. Once again, cells failed to produce cytokines in response to either the WT or the lpxF-null mutant lipid A (see Fig. S2 in the supplemental material). While the LPS failed to stimulate cytokines in vitro, we wanted determine if perhaps the mutant would signal through TLR2/TLR4 in vivo. In order to test this hypothesis, knockout mouse strains were exploited. Initially, C57BL/6 TLR2^−/− TLR4^−/− and WT C57BL/6 mice were infected subcutaneously with ∼10⁶ CFU of the lpxF-null F. novicida mutant. All mice survived the infection (data not shown). Subsequently, all triple knockout mice(C57BL/6 TLR2^−/− TLR4^−/− caspase-1^−/−) survived primary infection. Caspase-1 is a proteolytic enzyme that cleaves precursors of the interleukin-1β (IL-1β) and interleukin-18 (IL-18), cytokines which have been shown to be involved in clearance of LVS infections (see Fig. S3A in the supplemental material). This indicates that in addition to the TLR2-TLR4 systems, these two proinflammatory cytokines were not required during the course of lpxF-null F. novicida mutant infection. To confirm that other receptors in the TLR system were not involved in the detection of the lpxF-null F. novicida mutant, we used knockout mice lacking the signaling adaptor molecule, MyD88. MyD88 is used by all TLRs to activate host NF-κB-driven inflammatory responses. When C57BL/6 MyD88^−/− mice were infected subcutaneously with the lpxF-null F. novicida mutant, all but one survived the infection (see Fig. S3B). The fact that only one C57BL/6 MyD88^−/− mouse died and that death occurred at a very late time point (day 14 after the lpxF-null F. novicida mutant infection) suggest that MyD88 has a minimal effect on the control of infection with lpxF-null mutant. We further confirmed the dispensability of MyD88 using MyD88 and interferon-alpha receptor knockout (IFNAR^−/−) mice. All C57BL/6 MyD88^−/− IFNAR^−/− mice also survived infection with the lpxF-null F. novicida mutant (see Fig. S3C), emphasizing that MyD88 and thus the majority of the TLR signaling system were not essential for control of lpxF-null F. novicida mutant infection.

The adaptive immune system is not required for control of lpxF-null F. novicida mutant infection.

The resistance of TLR/MyD88-deficient mice to lpxF-null F. novicida mutant infection may be a result of priming of adaptive immune components that confer protection to a primary infection. To test this possibility, the susceptibility of C57BL/6 RAG-1^−/− mice, which lack mature T and B cells, to the lpxF-null F. novicida mutant was investigated. C57BL/6 and C57BL/6 RAG-1^−/− mice were infected subcutaneously with ∼10⁶ CFU of lpxF-null F. novicida mutant bacteria. C57BL/6 RAG-1^−/− mice did not show any signs of disease and survived the infection, similar to WT mice (see Fig. S3D in the supplemental material). This finding indicated that T and B cells were not essential in the control of primary infection with the lpxF-null F. novicida mutant.

IFN-γ is required for clearance of the lpxF-null F. novicida mutant.

The importance of IFN-γ in the clearance of intracellular bacterial infections has been previously demonstrated (9, 16). To explore the role of IFN-γ in the clearance of an lpxF-null F. novicida mutant infection, C57BL/6 IFN-γ^−/− mice were infected subcutaneously with ∼2 × 10⁶ CFU of the lpxF-null F. novicida mutant or ∼200 CFU of WT F. novicida. C57BL/6 IFN-γ^−/− mice died at day 3 after WT F. novicida infection, similar to WT mice infected with WT F. novicida (Fig. 5A). In contrast to WT mice infected with the lpxF-null F. novicida mutant, all of the C57BL/6 IFN-γ^−/− mice succumbed to the lpxF-null F. novicida mutant although the time of death was delayed to day 7 postinfection (Fig. 5A). This finding was confirmed in C57BL/6 IFN-γ^−/− mice using a pulmonary infection model. C57BL/6 IFN-γ^−/− mice exposed to ∼200 CFU of aerosolized WT F. novicida died at day 2 postinfection. When C57BL/6 IFN-γ^−/− mice were exposed to ∼6 × 10⁵ CFU of aerosolized lpxF-null F. novicida mutant bacteria, all mice succumbed to the infection by day 11 postinfection, a significant delay compared to survival with WT F. novicida infection of these mice (Fig. 5B).

Fig 5.

IFN-γ^−/− mice are sensitive to lpxF-null F. novicida mutant infection by both the subcutaneous and pulmonary routes. C57BL/6 or IFN-γ ^−/− mice (n = 4 each group) were exposed to either WT F. novicida (U112) (A, 244 CFU; B, 225 CFU) or the lpxF-null F. novicida mutant (A, 2.2 × 10⁶ CFU; B, 6 × 10⁵ CFU) by either the subcutaneous (A) or pulmonary (B) route. Data are representative of two independent experiments.

Natural killer cells and neutrophils are not required for clearance of the lpxF-null F. novicida mutant.

Subsets of innate immune cells are capable of producing IFN-γ during the course of bacterial infection, including NK cells and neutrophils. To evaluate the role of NK cells, C57BL/6 mice depleted of NK cells by antibody injection or mock depleted with PBS were infected subcutaneously with the lpxF-null F. novicida mutant, and survival was monitored. Mice depleted of NK cells were able to survive lpxF-null F. novicida mutant infection (see Fig. S4A in the supplemental material). Depletion of NK cells was confirmed by fluorescence-activated cell sorting (FACS) analysis of spleens from both depleted and PBS-treated control mice (see Fig. S4B).

Neutrophils have also been shown to play an important role in bacterial clearance, particularly during the primary phase of infection (13, 28). Depletion of neutrophils using an anti-Ly-6G antibody had no effect on lpxF-null F. novicida mutant infection (see Fig. S5A in the supplemental material), suggesting that neutrophils, as well as NK cells, are not essential for clearance of the lpxF-null F. novicida mutant. Complete depletion of neutrophils was confirmed by FACS analysis of both anti-Ly-6G and PBS-treated mice (see Fig. S5B).

LpxF-null F. novicida mutant infection can protect against lethal challenge with WT F. novicida.

To investigate if a primary infection with the lpxF-null F. novicida mutant could protect mice from a subsequent lethal WT F. novicida challenge, mice were inoculated either subcutaneously or by the pulmonary route with the lpxF-null F. novicida mutant on days −28 and −14. Mice were then challenged with ∼200 to 400 CFU (20× to 40× LD₁₀₀) of WT F. novicida on day 0 and followed for 30 days postchallenge. Mice inoculated with the lpxF-null F. novicida mutant by either the subcutaneous (Fig. 6A) or pulmonary (Fig. 6B) route showed complete protection from a WT F. novicida challenge. To determine if inoculation with the lpxF-null F. novicida mutant could protect against a higher dose of WT F. novicida, mice were inoculated subcutaneously, as in the experiment shown in Fig. 6A, but were challenged with a WT F. novicida dose of 7.4 × 10⁴ CFU (7,400× LD₁₀₀). All lpxF-null F. novicida mutant-primed mice survived this lethal F. novicida challenge, whereas all naïve controls died by day 3 postchallenge (Fig. 6C). Finally, the protective efficacy of administration of only a single dose of the lpxF-null F. novicida mutant was examined by inoculating mice subcutaneously with ∼1.3 × 10⁶ CFU of the lpxF-null F. novicida mutant and challenging them 28 days later with 220 CFU of WT F. novicida. Again, all mice that received a single dose of the lpxF-null F. novicida mutant survived WT challenge, whereas all naïve controls died by 3 to 4 days postinfection (data not shown).

Fig 6.

lpxF-null F. novicida mutant infection can protect mice against a subsequent WT F. novicida challenge. C57BL/6 (n = 5 each group) mice were inoculated by subcutaneous injection (A and C) or pulmonary exposure (B) on day −28 (A and C, ∼1.0 × 10⁶ CFU; B, ∼1.0 × 10⁵ CFU) and day −14 (A and C, ∼1.1 × 10⁶ CFU; B, ∼1.0 × 10⁵ CFU). The mice were then challenged with various doses of WT F. novicida (A, 400 CFU; B, 400 CFU; C, 173 CFU [naïve], 7 × 10³ CFU [low], or 7.4 × 10⁴ CFU [high]), and survival was followed. Each graph is representative of at least three independent experiments.

To determine if long-term immunity to a lethal WT challenge was achieved in mice immunized with lpxF-null mutant, C57BL/6 mice were inoculated subcutaneously with two doses of the lpxF-null mutant (1.5 × 10⁵ CFU) 14 days apart, as in previous experiments. These groups were then allowed to rest for either 90 or 120 days before being challenged with ∼200 to 500 CFU of WT F. novicida. All mice that received the lpxF-null mutant 90 or 120 days before challenge survived, while all naïve controls died over a normal time course (data not shown).

Innate and adaptive immune components are essential for development of protective immunity.

To determine if components of the host innate immune system were required in the generation of the protective immune response, C57BL/6 TLR2^−/− TLR4^−/− caspase-1^−/− and WT C57BL/6 mice were immunized by the subcutaneous route with the lpxF-null F. novicida mutant on days −28 (5 × 10⁵ CFU) and −14 (3 × 10⁶ CFU) and challenged subcutaneously with a lethal dose (200 CFU) of WT F. novicida on day 0. All C57BL/6 TLR2^−/− TLR4^−/− caspase-1^−/− mice succumbed while WT mice survived (see Fig. S6A in the supplemental material). To examine the role of MyD88 in eliciting protective immunity, C57BL/6 MyD88^−/−, C57BL/6 MyD88^−/− IFNAR^−/−, and WT C57BL/6 mice were immunized with the lpxF-null F. novicida mutant on days −28 and −14, as in the experiment shown in Fig. S6A, and challenged with WT F. novicida on day 0. All WT mice survived, while all C57BL/6 MyD88^−/− and all of the C57BL/6 MyD88^−/− IFNAR^−/− mice succumbed to the challenge (see Fig. S6B and C, respectively).

The role of adaptive immunity was subsequently examined using C57BL/6 RAG-1^−/− mice. C57BL/6 RAG-1^−/− and WT C57BL/6 mice were immunized with the lpxF-null mutant on days −28 and −14 and challenged with WT F. novicida on day 0. As expected, all C57BL/6 RAG-1^−/− mice succumbed to the infection, indicating a role for T and B cells in the protective immune response (see Fig. S6D in the supplemental material). To dissect the essential role of individual cell types of the adaptive immune system, B cell-deficient mice (C57BL/6 μMT^−/−) and in vivo depletion of T cell subpopulations were utilized. C57BL/6 μMT^−/− or WT C57BL/6 mice were immunized with lpxF-null F. novicida mutant bacteria on day −28 and later challenged with WT F. novicida on day 0. All WT mice were protected against WT F. novicida challenge, whereas all C57BL/6 μMT^−/− mice died over a normal time course of 3 to 5 days (data not shown). This finding indicated that the protection from WT F. novicida challenge exhibited by lpxF-null mutant-immunized mice was dependent on B cells.

To investigate whether this protection from WT F. novicida challenge was also reliant on T cells, C57BL/6 mice were depleted of CD4⁺ or CD8⁺ T cells by antibody depletion. These mice were inoculated with the lpxF-null F. novicida mutant on day −28, depleted of T cell subpopulations on day −1, and challenged with WT F. novicida on day 0, and survival was followed. CD8⁺ T cell-depleted mice that received a single lpxF-null F. novicida mutant inoculation were able to clear WT F. novicida challenge, whereas half of the CD4⁺ T cell-depleted mice succumbed, indicating that protection is at least partially dependent on CD4⁺ T cells (Fig. 7). Complete depletion of T cells was confirmed by FACS analysis of anti-CD4 and anti-CD8, as well as PBS-treated mice (see Fig. S7 in the supplemental material).

Fig 7.

Role of T cells in protection from lethal WT challenge. C57BL/6 mice (n = 5 each group) were inoculated subcutaneously with the lpxF-null mutant on days −28 and −14 (1 × 10⁵ CFU). Mice were then depleted of CD4⁺ or CD8⁺ T cells at day −1 and challenged with WT F. novicida (U112) (200 CFU) on day 0, and survival was followed. Data are representative of two independent experiments. α, anti.

LpxF-null F. novicida mutant infection fails to protect mice from lethal challenge with type A or type B F. tularensis subsp. tularensis.

Following the finding that infection with the lpxF-null F. novicida mutant could protect mice from WT F. novicida challenge, this protection was tested for the ability to protect mice from heterologous type A or type B F. tularensis subsp. tularensis challenge. C57BL/6 mice were inoculated with the lpxF-null F. novicida mutant on days −28 and −14 and then challenged with F. tularensis subsp. tularensis type A strain SchuS4 (FT1D) or FT4D (a clinical isolate), type B strain FSC200, or WT F. novicida as a control. All mice that were challenged with the heterologous type A strains (see Fig. S8A and C in the supplemental material) or type B strain (see Fig. S8B) failed to survive the challenge, whereas complete protection was observed when mice were challenged with the homologous WT F. novicida strain. Mice challenged with a 10-fold lower dose of F. tularensis subsp. tularensis also succumbed (data not shown).

DISCUSSION

Francisella is a Gram-negative intracellular pathogen that has potential for use as a bioweapon. This bacterium, once taken up by phagocytes, can inhibit phagolysosomal fusion and escape from the phagosome into the cytoplasm. The lipid A structure of Francisella differs from the classical lipid A structure present in enteric bacteria and does not exhibit proinflammatory activity via TLR2 or TLR4, either in vivo or in vitro (15). This lack of recognition may contribute to the high lethality of Francisella infection. One contributing factor in the lack of proinflammatory responses for F. novicida lipid A is the lack of the 4′-position phosphate, removed by the inner membrane/periplasmic enzyme, LpxF. The absence of this enzyme in the lpxF-null F. novicida mutant causes retention of both the 4′ phosphate moiety on the lipid A backbone and surprisingly one additional fatty acid chain at the 3′ position, suggesting steric hindrance of a currently unidentified 3′-position deacetylase if the phosphate moiety is retained. This mutant was reported to be highly avirulent in the mouse intradermal footpad infection model. In addition, in this infection model system, a strong in vivo proinflammatory response was observed (35).

We showed here that the lpxF-null F. novicida mutant is rapidly cleared in macrophages and is also avirulent when mice are infected via the subcutaneous and pulmonary routes, even at a very high dose. We next investigated which arm of the host immune system was essential for clearance of primary infection with the lpxF-null F. novicida mutant. Because the lpxF-null F. novicida mutant has an altered lipid A structure and induces proinflammatory responses in vivo (35), we hypothesized that the lipid A modification in the lpxF-null F. novicida mutant might render it recognizable by the TLR system and thus stimulate a host immune response able to clear the bacteria. However, all TLR2^−/− TLR4^−/− caspase-1^−/− mice and WT mice infected subcutaneously with the lpxF-null F. novicida mutant survived the infection, thus indicating that TLR2 and TLR4 do not play a crucial role in the initial immune response to an lpxF-null F. novicida mutant infection. This result was contrary to the immune response to infection with the type B F. tularensis subsp. tularensis live vaccine strain (LVS) that was reported to require the host TLR2 system for survival (22).

In addition to the TLR system, caspase-1 activation was previously reported to mediate the innate immune response to LVS infection (20). Caspase-1 activation leads to the cleavage and release of mature forms of IL-1β and IL-18. In contrast, in our studies mice lacking caspase-1 along with TLR2 and TLR4 survived infection with the lpxF-null F. novicida mutant, indicating that the innate immune response to this mutant does not require caspase-1 activation.

Because the lpxF-null F. novicida mutant activates a proinflammatory response in mice, the mutant may be detected by components of the TLR system other than TLR2 and TLR4. The major functional pathway downstream of TLR stimulation requires MyD88 signaling leading to the activation of target genes involved in the host proinflammatory response (2). We tested whether TLR stimulation and, thus, signaling through the MyD88 adaptor protein were important for control of lpxF-null F. novicida mutant infection. MyD88^−/− mice infected with the lpxF-null F. novicida mutant showed no signs of disease throughout the study, indicating that the MyD88 pathway was not required for the response to infection with this strain. Furthermore, the survival of MyD88^−/− IFNAR^−/− mice not only confirmed the dispensability of MyD88 but also implied that TLR activation through an MyD88-independent pathway involving IFN regulatory factors (IRFs) and type I interferon was not involved in clearance of the lpxF-null F. novicida mutant. Taken together, the data suggested that although an innate immune response played a role in lpxF-null F. novicida mutant infection, this response did not require the TLR system and its MyD88 adaptors, caspase-1 or type I interferon.

It is possible that the lpxF-null F. novicida mutant may replicate very poorly in vivo, similar to its slow growth in broth culture (see Fig. S1 in the supplemental material), and is thus eliminated by other components of the innate immune system, such as neutrophils, natural killer cells, macrophages, complement, or antimicrobial peptides. To evaluate this possibility, in vivo depletion of neutrophils or NK cells was performed by administration of cell-specific antibodies before infection with the lpxF-null F. novicida mutant. Both the neutrophil-depleted and NK cell-depleted mice survived the infection, suggesting that neither of these cell types was required for clearance of lpxF-null F. novicida mutant bacteria.

IFN-γ has been shown to be important in several intracellular bacterial infections (9, 16). Engagement of the IFN-γ receptor initiates a signaling pathway involving activation of Janus kinases (JAK) and signal transducer and activator of transcription (STAT) proteins, followed downstream by the transcriptional induction of target IFN-stimulated genes (33). IFN-γ is important in the activation of macrophages to effectively kill intracellular bacteria and in the induction of CD4⁺ Th1 and CD8⁺ cytotoxic T cell responses. IFN-γ has been demonstrated to suppress Francisella replication inside macrophages, inhibit phagosome escape, and also induce the expression of IL-23 (3, 5). Conversely, Francisella infection may suppress activation of STAT1 expression and phosphorylation, possibly through upregulated expression of suppressors of cytokine signaling 3 (SOCS3) (27). We observed here, consistent with other reports (19), that absence of IFN-γ shortened the time to death in mice with pulmonary tularemia, emphasizing the importance of IFN-γ in WT F. novicida infection (26). Our data also showed that IFN-γ was crucial for clearance of lpxF-null F. novicida mutant bacteria in both subcutaneous and pulmonary infections. Although IFN-γ^−/− mice infected with lpxF-null F. novicida mutant through the pulmonary route died significantly more slowly than WT mice infected with WT F. novicida (day 10 or 11 versus day 4, respectively), it is likely that this was due to the very low replication rate of the mutant and, thus, slow onset of disease.

Multiple sources of IFN-γ in the early course of LVS infection have been reported, including NK cells, T cells, dendritic cells, NK-T cells, NK DCs, and neutrophils; production of IFN-γ during LVS infection was not dependent on T and B cells (8). In the lpxF-null F. novicida mutant infection model, the production of IFN-γ is very likely independent of TLR2, TLR4, MyD88 adaptor, type I IFN, and T and B cells. IFN-γ production is also independent of NK cells since mice depleted of NK cells were able to survive primary lpxF-null F. novicida mutant infection. The sources of IFN-γ in lpxF-null F. novicida mutant infection remain to be determined.

As there are no Food and Drug Administration (FDA)-approved vaccines to Francisella available, there is a need to identify well-defined vaccine targets conferring protection against a potential outbreak. We determined that the highly avirulent lpxF-null F. novicida mutant could induce protection against WT F. novicida infection and defined the mechanisms underlying the development of protection. The lpxF-null F. novicida mutant provided complete protection against a lethal WT F. novicida challenge by both the subcutaneous and pulmonary routes. This protection is dependent on both innate and adaptive immunity as knockout mice remained susceptible to lethal WT challenge after vaccination with the lpxF-null F. novicida mutant. Specifically, we determined that B cells from the adaptive immune system were essential since all μMT^−/− mice succumbed to infection. RAG-1^−/− mice died later than any of the mice with knockout of the innate immune system, suggesting that the innate immune system controls early infection, which can then be cleared by antibody.

Protection was achieved even when immunized mice were challenged with a very high, lethal dose of WT F. novicida, suggesting that this mutant may be a potential candidate for vaccine development. Protection was also long-lived and partially dependent on CD4⁺ T cells, indicating that protection is most likely mediated by class-switched high-affinity antibodies and not natural IgM, as is seen in some LVS models of protection (7). The class of antibodies induced by lpxF-null mutant infection will be determined in a future study. The lpxF-null F. novicida mutant did not cross-protect against heterologous type A and type B F. tularensis subsp. tularensis infection, possibly due to antigenic differences among different Francisella subspecies, specifically differences in the O-antigen composition. Therefore, whether mutations in the lpxF gene in type A and type B F. tularensis subsp. tularensis genomic backgrounds would provide comparable protection to that observed in the F. novicida model is being determined.

Finally, the lpxF-null F. novicida mutant may be recognized by intracellular receptors, such as nucleotide-binding domain, leucine-rich repeat-containing proteins (NLRs), although whether the lpxF-null F. novicida mutant can escape phagolysosomal fusion and activate intracellular receptors remains to be determined. Recent studies have implicated the pattern recognition receptor AIM2 (absent in melanoma 2) as being critical for survival of primary F. novicida infection (12). AIM2 is unlikely to play a role in lpxF-null F. novicida mutant infection since it functions upstream of caspase-1, which our data suggest was dispensable for survival. In conclusion, this study described the lpxF-null F. novicida mutant, which is rapidly cleared and sensitive to AMPs, as a potential vaccine candidate for the subcutaneous and airborne routes of transmission; in addition this study provided an understanding of the fundamental basis of the immune response to this mutant. This knowledge should help guide development of vaccines against human tularemia to achieve maximum efficacy.