Perforin- and Granzyme-Mediated Cytotoxic Effector Functions Are Essential for Protection against Francisella tularensis following Vaccination by the Defined F. tularensis subsp. novicida ΔfopC Vaccine Strain

Abstract

A licensed vaccine against Francisella tularensis is currently not available. Two Francisella tularensis subsp. novicida (herein referred to by its earlier name, Francisella novicida) attenuated strains, the ΔiglB and ΔfopC strains, have previously been evaluated as potential vaccine candidates against pneumonic tularemia in experimental animals. F. novicida ΔiglB, a Francisella pathogenicity island (FPI) mutant, is deficient in phagosomal escape and intracellular growth, whereas F. novicida ΔfopC, lacking the outer membrane lipoprotein FopC, which is required for evasion of gamma interferon (IFN-γ)-mediated signaling, is able to escape and replicate in the cytosol. To dissect the difference in protective immune mechanisms conferred by these two vaccine strains, we examined the efficacy of the F. novicida ΔiglB and ΔfopC mutants against pulmonary live-vaccine-strain (LVS) challenge and found that both strains provided comparable protection in wild-type, major histocompatibility complex class I (MHC I) knockout, and MHC II knockout mice. However, F. novicida ΔfopC-vaccinated but not F. novicida ΔiglB-vaccinated perforin-deficient mice were more susceptible and exhibited greater bacterial burdens than similarly vaccinated wild-type mice. Moreover, perforin produced by natural killer (NK) cells and release of granzyme contributed to inhibition of LVS replication within macrophages. This NK cell-mediated LVS inhibition was enhanced with anti-F. novicida ΔfopC immune serum, suggesting antibody-dependent cell-mediated cytotoxicity (ADCC) in F. novicida ΔfopC-mediated protection. Overall, this study provides additional immunological insight into the basis for protection conferred by live attenuated F. novicida strains with different phenotypes and supports further investigation of this organism as a vaccine platform for tularemia.

INTRODUCTION

Live attenuated vaccines have played a significant role in control of bacterial and viral infections (3, 4, 31, 32, 34, 38, 44). The immune response to a live attenuated vaccine closely resembles that produced by a natural infection, usually comprising potent cellular and humoral responses (14, 71). Owing to their relative complex genetic nature, attenuated bacterial vaccines are often difficult to develop despite the availability of molecular techniques (34, 49, 70). To this end, there is a continued interest in development of a live attenuated vaccine against pneumonic tularemia. Francisella tularensis, the etiological agent of tularemia, is a Gram-negative facultative intracellular bacterium (11, 69). Untreated cases of pneumonic tularemia caused by the virulent human type A strains (Francisella tularensis subsp. tularensis) may have mortality rates of 30 to 60% (66). A live vaccine strain (LVS), derived from a type B strain that is less virulent in humans (Francisella tularensis subsp. holarctica), has been used to immunize millions of individuals in the former Soviet Union (62). However, LVS has not been licensed for general use due to phenotypic inconsistency (12, 23) and a lack of understanding of the nature of attenuation. Recently, our laboratory evaluated vaccine strains derived from Francisella tularensis subsp. novicida (herein referred to by its earlier name, Francisella novicida) (U112 strain) in experimental rodent models and found that protection can be achieved against heterotypic pulmonary LVS and SCHU S4 (type A) bacterial challenge (9, 53). F. novicida is genetically related to type A F. tularensis [98.1% homology between sequences common to U112 and the type A strain SCHU S4 (55)] and is avirulent in humans.

Using F. tularensis as a model pathogen, this study aimed at gaining additional immunological insight into the basis for protection conferred by live attenuated bacterial vaccines. Specifically, two live attenuated F. novicida mutant strains, namely, F. novicida ΔfopC (46) and F. novicida ΔiglB (9), were directly compared in order to characterize the mechanistic details underlying the respective protective efficacy against pulmonary F. tularensis LVS challenge. F. novicida ΔiglB is a Francisella pathogenicity island (FPI) mutant, deficient in intramacrophage growth and phagosomal escape (7, 36, 47). In contrast, F. novicida ΔfopC has a deficiency in the outer membrane lipoprotein (FopC), which has been reported by us (46) to mediate evasion of gamma interferon (IFN-γ)-mediated signaling and by others (35, 56) to play a role in iron acquisition and to be an important virulence factor for type A F. tularensis. Moreover, F. novicida ΔfopC replicated similarly to wild-type F. novicida U112 in primary macrophages that had not been stimulated with IFN-γ (46), suggesting that the bacterium likely escaped from phagosomes and replicated within the cytosol. Given the difference in phagosomal escape between F. novicida ΔiglB and F. novicida ΔfopC, the host may likely utilize different antigen-processing pathways and T-cell subsets to generate protective immunity.

To this end, the major histocompatibility complex class I (MHC I) pathway samples intracellular antigens (such as proteins secreted or degraded from cytosolic F. novicida ΔfopC bacteria) to present to cytotoxic T lymphocytes (CD8⁺ T cells) (24). On the other hand, the MHC II pathway interacts with endocytic exogenous antigens (such as antigens generated from F. novicida ΔiglB in the phagosomes) for presentation to helper T cells (CD4⁺ T cells) (24). Given that the initial priming mechanisms for the two attenuated mutant vaccine strains may be different, we sought to investigate whether these strains utilized different host immune components to induce protection against pulmonary F. tularensis LVS challenge with a panel of knockout mice, including MHC I, MHC II, CD4⁺ T cells, CD8⁺ T cells, and perforin, a potent cytotoxic effector molecule produced primarily by CD8⁺ T cells and natural killer (NK) cells. In these studies, we found an important protective role for perforin following oral F. novicida ΔfopC but not F. novicida ΔiglB vaccination against pulmonary F. tularensis LVS infection. The protection conferred by F. novicida ΔfopC vaccination may be mediated by NK cells via the release of perforin and granzymes. This is the first report that definitively describes dissimilar host-protective mechanisms mediated by two live attenuated F. novicida vaccine strains with major differences in phagosomal escape phenotypes.

MATERIALS AND METHODS

Bacteria.

F. tularensis subsp. novicida strain U112 was provided by Francis Nano (University of Victoria, Victoria, Canada). F. tularensis subsp. holarctica LVS (lot 703-0303-016) was obtained from Rick Lyons (University of New Mexico). The iglB and fopC mutants of F. novicida U112 were generated as reported previously (36, 46). All strains were grown at 37°C in tryptic soy broth (TSB) or on tryptic soy agar (TSA) (BD Biosciences, San Jose, CA), each supplemented with 0.025% sodium pyruvate, 0.025% sodium metabisulfite, 0.025% ferrous sulfate, and 0.1% l-cysteine. Aliquots of bacteria were stored at −80°C in TSB containing all supplements and glycerol (24%).

Mice.

C57BL/6 mice (4 to 8 weeks) were purchased from the National Cancer Institute (Frederick, MD). C57BL/6 MHC I β₂-microglobulin^−/− mice (30), MHC II H2^−/− mice (39), μMT (B-cell-deficient) mice (29), CD4^−/−-T-cell mice (43), CD8^−/−-T-cell mice (18), FcγR^−/− mice (68), and perforin^−/− mice (26) were purchased from the Jackson Laboratory (Bar Harbor, ME). Mice were housed and bred at the University of Texas at San Antonio Animal Facility. Animal care and experimental procedures were performed in compliance with Institutional Animal Care and Use Committee (IACUC) guidelines. Experiments were performed with age-matched groups of animals.

Immunization and challenge.

Mice (5 to 7 animals/group) were anesthetized and orally vaccinated, using a 22-gauge, 25-mm-long, 1.25-mm-tip feeding needle (Fine Science Tools Inc., Foster City, CA) (22), with either 10³ CFU of live F. novicida iglB or fopC mutant cells in 100 μl of phosphate-buffered saline (PBS) or PBS alone (mock vaccination). The 50% lethal doses (LD₅₀) of ΔiglB and ΔfopC strains administered intranasally were greater than 10⁷ CFU and between 10⁴ and 10⁵ CFU, respectively (9, 46). Vaccinated mice were allowed to rest for 4 weeks and then challenged intranasally (i.n.) with 4 to 5 LD₅₀ of LVS (LD₅₀ = ∼8,000 CFU [data not shown]) in 25 μl PBS. Within individual experiments, the same dose of LVS was used for each respective test variable. The actual vaccination and challenge doses administered were determined by dilution plating on TSA with supplements. Animals were monitored daily for morbidity and mortality.

Determination of bacterial dissemination of LVS in F. novicida ΔfopC- and F. novicida ΔiglB-vaccinated mice.

C57BL/6 mice were orally vaccinated with 1,000 CFU of F. novicida ΔfopC or F. novicida ΔiglB in 100 μl of PBS on day 0 and challenged i.n. with 5 LD₅₀ of LVS on day 28. Groups of mice (n = 3 per group) were euthanized on days 2 and 5 postchallenge. Lungs, livers, and spleens were removed, and the bacterial counts in the homogenized tissues were determined by dilution plating.

Preparation of single-cell suspensions from spleens.

Immunized mice were sacrificed on day 14, and spleens were removed under aseptic conditions, disrupted with a 3 ml syringe plunger, and passed through a 0.76-μm cell strainer. Single-cell suspensions were prepared, and erythrocytes were lysed with ammonium chloride. Cells were washed, viability was assessed by trypan blue exclusion, and cells were resuspended in Dulbecco's modified Eagle medium (DMEM; Mediatech, Fairfax, VA) containing 10% (wt/vol) fetal bovine serum (FBS; HyClone, Logan, UT) supplemented with l-glutamine (D10). The T-lymphocytes and NK cells were enriched from single-cell suspensions of splenocytes using an EasySep mouse T cell enrichment kit and an EasySep mouse NK cell enrichment kit (Stemcell Technologies, Vancouver, Canada), respectively. For experiments involving antigen-presenting cells (APCs), splenocytes were treated with mitomycin C (25 μg/10⁷ cells) for 20 min at 37°C in 5% CO₂, followed by 2 h of incubation to separate the adherent APCs from nonadherent lymphocytes.

Generation of bone marrow-derived macrophages (BMDM) and in vitro infection.

Four- to five-week-old mice were euthanized, and femurs were removed and flushed with D10 supplemented with penicillin (100 U/ml) and streptomycin (100 μg/ml). Cells were collected by centrifugation and resuspended in D10 containing 10% culture medium from the L929 hybridoma cell line (6), seeded into 75-cm² tissue culture flasks, and incubated at 37°C with 5% CO₂ to allow differentiation into BMDM. Adherent cells were harvested after 6 to 7 days and stained with phycoerythrin (PE)-conjugated CD11b (BD Biosciences), and flow cytometry was used to determine the purity of the cell population. We routinely obtained >95% CD11b⁺ cells from the preparations.

For bacterial infection, BMDM were seeded at 2 × 10⁵/well in sterile 96-well plates in D10 and incubated at 37°C in 5% CO₂. After 2 h of incubation, Francisella bacteria were diluted from frozen stocks, added to BMDM at a multiplicity of infection (MOI) of 10:1 or 100:1, and incubated for 2 h to allow bacterial uptake. After 3 washes with DMEM, the BMDM were then incubated for 1 h in D10 supplemented with 50 μg/ml gentamicin to kill extracellular bacteria. Cells were washed three times and lysed with 0.2% deoxycholate solution (3-h time point) or incubated in D10 for additional 21 h prior to lysis (24-h time point). Intracellular bacteria were enumerated by plating serial dilutions of lysates on TSA containing supplements.

For coculture experiments, T cells or NK cells enriched from splenocytes were added to LVS-infected (10 MOI) BMDM at a 1:1 ratio (200 μl/well) following gentamicin treatment. Cultures were incubated at 37°C in 5% CO₂ for the remainder of the experiment. LVS replication within BMDM was determined as described above. Supernatants also were collected for measurement of cytokines and/or granzyme at 3, 6, 12, 24, and/or 72 h based on the specific experimental conditions. IFN-γ, interleukin 1β (IL-1β), and granzyme enzyme-linked immunosorbent assays (ELISAs) (BD Pharmingen, San Diego, CA) were used to determine cytokine levels in the supernatants according to manufacturer's recommendations.

Statistical analysis.

Data from all assays were evaluated using an ANOVA one-way test, except that the Kaplan-Meier test was used for survival analysis. The data are presented as means ± standard deviations. Each experiment was independently done at least twice.

RESULTS

F. novicida ΔfopC but not F. novicida ΔiglB escaped from phagosomes in primary macrophages.

We previously characterized two F. novicida (strain U112) mutants, F. novicida ΔiglB and F. novicida ΔfopC, which showed distinctive intramacrophage replication phenotypes (9, 46). As shown in Fig. 1A, the ΔiglB strain was attenuated in growth (less than a 1-log-unit increase from 3 h to 24 h after inoculation) in BMDM, whereas the ΔfopC strain, like its wild-type parent, U112, displayed marked replication (3- to 4-log-unit increase). Lack of intracellular replication for Francisella ΔiglB has been correlated with its deficiency in phagosomal escape (7). To assess whether F. novicida ΔfopC escaped from phagosomes and replicated in the cytosol, as seen with the U112 parental strain (57, 58), we analyzed IL-1β secretion by BMDM following bacterial infection. Release of IL-1β in infected macrophages as a result of inflammasome activation following escape of Francisella from the phagosome into the cytosol has been reported (8, 20, 21, 41). We observed a comparable high level (>3,000 pg/ml) of IL-1β released by BMDM upon live U112 and F. novicida ΔfopC infection but not with UV-inactivated bacteria (data not shown), suggesting escape of live F. novicida ΔfopC cells from phagosomes (Fig. 1B). In contrast, minimal amounts of IL-1β secreted by F. novicida ΔiglB-infected BMDM confirmed that F. novicida ΔiglB failed to disrupt phagosomes, similarly to other reported FPI-associated mutants (58, 59). Negligible amounts of IL-1β were secreted by uninfected macrophages (Fig. 1B) as well as those infected with UV-inactivated bacteria (data not shown). Collectively, such evidence indicates differences in the cytosolic presence and replication between the F. novicida ΔiglB and F. novicida ΔfopC strains, which might lead to some distinct immune responses due to differences in antigen processing and presentation by the host cells.

Fig 1.

Phenotypic differences between F. novicida ΔiglB and F. novicida ΔfopC in bacterial replication and IL-1β production in BMDM. BMDM were infected (MOI, 10 or 100) with WT U112, F. novicida ΔfopC, or F. novicida ΔiglB. (A) Cell lysates were dilution plated at 3 and 24 h postinfection for bacterial enumeration. Differences in bacterial numbers were significant (*, P < 0.05; **, P < 0.01) for F. novicida ΔiglB compared to U112 or F. novicida ΔfopC at 24 h at MOIs of 10 and 100. (B) Supernatants were collected at 6, 12, and 24 h postinfection for determination of IL-1β levels by ELISA. Differences in IL-1β production at 12 and 24 h between F. novicida ΔiglB and either U112 or F. novicida ΔfopC were significant (***, P < 0.001). Results are representative of three independent experiments with triplicates in each group.

Antigen presentation by APC following F. novicida ΔfopC and F. novicida ΔiglB infection.

Gene knockout (KO) mice with deficient MHC I or MHC II were used to study whether different pathways were used to present F. novicida ΔfopC and F. novicida ΔiglB antigens. APCs were generated from naïve wild-type (WT), MHC I KO, and MHC II KO mice by treating the total splenocyte population with mitomycin C, followed by removal of nonadherent cells. APCs (2 × 10⁵/well) were then infected (MOI of 10) with live or UV-inactivated F. novicida ΔfopC or F. novicida ΔiglB for 2 h, treated with gentamicin for an additional 1 h, and then washed to remove all extracellular bacteria. These antigen-loaded APCs were incubated for 72 h with T-lymphocytes enriched from spleens of WT mice which had been inoculated orally with 1,000 CFU of F. novicida ΔfopC, F. novicida ΔiglB, or PBS. Following incubation, IFN-γ concentrations in the culture supernatants were assayed, and these served as an indicator of T-cell activation by the APCs.

As shown in Fig. 2, T cells from PBS-treated (mock-vaccinated) mice, regardless of whether they were cocultured with APCs isolated from WT, MHC I KO, or MHC II KO mice and previously infected with F. novicida ΔfopC, F. novicida ΔiglB, or hen egg lysozyme (HEL, an unrelated antigen used as a control), had minimal IFN-γ induction, as expected. In contrast, T cells obtained from WT mice primed by oral vaccination with F. novicida ΔfopC or F. novicida ΔiglB, upon incubation with WT APCs infected with respective bacterial mutant strains, exhibited similar elevated levels of IFN-γ production, indicating that WT APCs effectively processed and presented both F. novicida ΔfopC and F. novicida ΔiglB antigens. Similarly, comparable IFN-γ levels were observed in T cells primed with F. novicida ΔfopC and F. novicida ΔiglB, upon activation by respective mutant-infected APCs from either MHC I or MHC II KO mice, although levels for both mutants were lower in MHC II KO mice. Additionally, there was no significant difference in IFN-γ production by primed T cells upon activation by either live or UV-inactivated mutant-infected APCs from WT, MHC I^−/−, or MHC II^−/− mice. Collectively, these data suggest that there may be cross antigen presentation of the two mutant strains by MHC I and II pathways.

Fig 2.

Presentation of F. novicida ΔiglB and F. novicida ΔfopC antigens by MHC I and II. T cells obtained from WT mice orally immunized with F. novicida ΔfopC, F. novicida ΔiglB, or PBS were incubated for 72 h with respective WT, MHC I^−/−, or MHC II^−/− APCs that had been infected with bacteria (live or UV inactivated) and treated with mitomycin C. IFN-γ production by the T cells was measured by ELISA. APCs incubated with an unrelated antigen, hen egg lysozyme (HEL), also were included for comparison. Results are representative of two independent experiments with triplicates in each group.

Roles of CD4⁺ and CD8⁺ T cells in F. novicida ΔfopC and F. novicida ΔiglB oral vaccination-mediated protection against intranasal LVS challenge.

In the ex vivo system described above, we observed no difference in antigen processing and presentation by major MHC pathways following F. novicida ΔfopC and F. novicida ΔiglB uptake by APCs. Thus, we then determined whether CD4⁺ and CD8⁺ T cells played comparable roles in F. novicida ΔfopC and F. novicida ΔiglB vaccination-mediated protection. Mice (WT, CD4^−/−, and CD8^−/−) were vaccinated orally with F. novicida ΔfopC, F. novicida ΔiglB, or PBS and challenged i.n. with a lethal dose (5 LD₅₀) of LVS on day 28. Organs (lungs, livers, and spleens) were collected on days 2 and 5 postchallenge and homogenized to determine bacterial burdens. In mock-vaccinated mice, although there was 1 log unit less LVS in the spleens of CD4^−/− and CD8^−/− mice than in those of WT mice, we observed no significant difference in mortality among these three groups of mice following a lethal LVS challenge (all mice succumbed to infection by day 10 [data not shown]), which may have been due to the relatively high LVS burdens in the target organs that led to uncontrolled inflammation and death (40, 73). In contrast, all mice (WT, CD4^−/−, and CD8^−/−) vaccinated with either F. novicida ΔfopC or F. novicida ΔiglB had significantly reduced LVS burdens on days 2 and 5 in the primary challenge site (lungs) and in secondary targeted organs (livers and spleens) compared to mock-vaccinated mice (Fig. 3). However, in CD8^−/− mice, greater bacterial burdens were evident in the lungs and livers of F. novicida ΔfopC-vaccinated mice than F. novicida ΔiglB-immunized mice, and the increased organ burdens correlated with reduced survival rate (66.7% for F. novicida ΔfopC immunization versus 100% for F. novicida ΔiglB immunization [data not shown]), suggesting that the CD8 component may be required for F. novicida ΔfopC- but not F. novicida ΔiglB-mediated protection against LVS challenge.

Fig 3.

Role of CD4⁺ and CD8⁺ T cells in F. novicida ΔfopC- and F. novicida ΔiglB-mediated protection (bacterial burden) against intranasal LVS challenge. Mice (3 per time point) were vaccinated orally with 1,000 CFU of F. novicida ΔfopC or F. novicida ΔiglB or mock treated with PBS on day 0 and i.n. challenged with a lethal dose of LVS (44,000 CFU, ∼5 LD₅₀) on day 28. Lungs, livers, and spleens were collected on days 2 and 5 after LVS challenge to determine the bacterial burdens. Significant differences in LVS burdens on day 2 and day 5 in three tested organs from either F. novicida ΔfopC- or F. novicida ΔiglB-vaccinated mice, compared to those in mock-vaccinated mice, are indicated (*, P < 0.05; **, P < 0.01; ***P < 0.001). The difference in LVS burdens in the lungs and livers between F. novicida ΔfopC- and F. novicida ΔiglB-vaccinated CD8^−/− mice was also significant (*, P < 0.05; ***, P < 0.001). Results are representative of two independent experiments.

Perforin is required for F. novicida ΔfopC-mediated protection against LVS challenge.

Given that CD8⁺ T cells may control LVS infection in F. novicida ΔfopC-vaccinated mice and perforin is a key effector of CD8⁺ cell granule-mediated cytotoxicity (26, 33), we sought to determine the role of perforin in inducing protective immunity following oral F. novicida ΔfopC vaccination. WT and perforin KO (PRF^−/−) mice were vaccinated with F. novicida ΔfopC, F. novicida ΔiglB, or PBS and challenged i.n. with 5 LD₅₀ of LVS on day 30. Bacterial burdens in the lungs, livers, and spleens were measured at days 2 and 5 after LVS challenge, and survival was monitored for 30 days. We observed consistently high LVS burdens in all organs of PBS control WT mice at day 2 postchallenge (Fig. 3 and 4), which might be due to the high challenge inoculum (38,000 to 44,000 CFU) used as described in Materials and Methods. Vaccination of WT mice with either F. novicida ΔfopC or F. novicida ΔiglB resulted in significant reductions of LVS in all target organs compared to mock-vaccinated mice (Fig. 4, top). In contrast, while PRF^−/− mice retained their ability to control LVS growth effectively when vaccinated with F. novicida ΔiglB, the deficiency in perforin abrogated the reduction of LVS burdens mediated by F. novicida ΔfopC immunization, resulting in organ burdens as high as those in the mock-vaccinated animals (Fig. 4, bottom). Although no significant increase of LVS burdens from day 2 to day 5 was observed in target organs of F. novicida ΔfopC-vaccinated PRF^−/− mice, high levels of bacterial burden correlated with high mortality (67%) of these perforin-deficient mice following lethal LVS challenge, compared to a 17% mortality rate in the F. novicida ΔfopC-vaccinated WT mice (Fig. 5). These results suggest that perforin plays an important role in F. novicida ΔfopC-mediated protective immunity against LVS infection. There was no difference in survival between F. novicida ΔiglB-immunized, LVS-challenged WT and PRF^−/− mice, suggesting that while perforin is essential for F. novicida ΔfopC-mediated protection, it has a minimal role in F. novicida ΔiglB-generated protective immunity.

Fig 4.

Perforin is required for control of LVS replication and dissemination in F. novicida ΔfopC-vaccinated mice. WT and PRF^−/− mice were vaccinated orally with 1,000 CFU of F. novicida ΔfopC or F. novicida ΔiglB or mock treated with PBS on day 0 and i.n. challenged with a lethal dose of LVS (38,000 CFU, ∼5 LD₅₀) on day 28. Lungs, livers and spleens (3 mice per time point) were collected on days 2 and 5 after LVS challenge to determine the bacterial burdens. In WT mice (top), the difference in LVS burdens in all organs between PBS and mutant (F. novicida ΔfopC or F. novicida ΔiglB)-vaccinated mice were significant (P < 0.001). In PRF^−/− mice (bottom), LVS burdens were significantly (P < 0.001) reduced only when mice were vaccinated with F. novicida ΔiglB, not F. novicida ΔfopC, compared to PBS treatment. Results are representative of two independent experiments.

Fig 5.

Perforin is required for F. novicida ΔfopC-mediated protection against LVS lethal challenge. WT and PRF^−/− mice were vaccinated orally with 1,000 CFU of F. novicida ΔfopC or F. novicida ΔiglB or mock treated with PBS and challenged i.n. with a lethal dose of LVS (35,000 CFU, ∼4 LD₅₀) on day 28. Mice were monitored for 30 days for survival. There were significant (P < 0.05) differences in survival between F. novicida ΔfopC- and F. novicida ΔiglB-vaccinated PRF^−/− mice. Results are representative of two independent experiments.

Perforin plays an important role in T-cell- and NK cell-mediated inhibition of LVS intramacrophage replication.

Cytotoxic T lymphocytes and NK cells are the major cell types that produce perforin upon activation (42, 61). To dissect the potential mechanism of perforin in control of LVS infection in F. novicida ΔfopC-vaccinated mice, we applied an in vitro killing coculture system to compare the inhibition of LVS intramacrophage replication in perforin competent and deficient T cells and NK cells. Total T cells were enriched from spleens of either F. novicida ΔfopC- or mock-immunized WT or PRF^−/− mice and cocultured for 21 h with WT BMDM which had previously been infected with LVS (2-h infection and 1-h gentamicin treatment). The levels of viable LVS within the BMDM were determined, and the production of granzymes, proteases delivered by perforin which trigger apoptosis of target cells (10), as well as IFN-γ, a potent cytokine produced by T cells that controls Francisella replication (2, 13, 46, 48), were measured in the culture supernatant at the end of coculture. LVS increased by 3 log units from 3 to 24 h in BMDM that were cultured alone (Fig. 6A). In contrast, there was a 25-fold (1.4-log-unit) reduction of LVS replication in BMDM cocultured with F. novicida ΔfopC-primed WT T cells (a 1.6-log-unit increase from 3 h to 24 h) compared to WT PBS-treated T cells (3-log-unit increase from 3 h to 24 h). BMDM cocultured with LVS WT F. novicida ΔfopC-primed PRF^−/− T cells exhibited only a 3-fold (0.5-log-unit) inhibition of LVS replication, suggesting that perforin played an important role in control of LVS intracellular replication. Granzyme levels were markedly higher in WT, F. novicida ΔfopC-vaccinated T cells cocultured with LVS-infected BMDM than the minimal granzymes produced by PRF^−/− F. novicida ΔfopC-primed T cells (Fig. 6B). On the other hand, comparable high levels of IFN-γ were produced by both WT and PRF^−/− F. novicida ΔfopC-primed T cells (Fig. 6C), suggesting that perforin-mediated inhibition of LVS replication in BMDM cells may be IFN-γ independent but granzyme dependent. Since CD8⁺ cells are the most predominate perforin-secreting T-cell type and PRF-mediated bacterial clearance was evident in our total T-cell population, a role of CD8⁺ cells in PRF-mediated bacterial clearance in F. novicida ΔfopC-vaccinated mice seems likely.

Fig 6.

Inhibition of LVS replication in BMDM by F. novicida ΔfopC-primed T cells is perforin and granzyme dependent. LVS infected BMDM were cocultured with T cells enriched from spleens of PBS treated or F. novicida ΔfopC immunized WT and PRF^−/− mice. (A) Viable LVS within BMDM was enumerated at 3 and 24 h after infection. Statistical significance between indicated groups (*, P < 0.05; **, P < 0.01). Granzyme (B) and IFN-γ (C) levels in the supernatants of cocultures were determined by ELISA. The difference in granzyme production between WT and PRF^−/− T cells was significant (P < 0.001). Results are representative of two independent experiments.

Similar results were observed in LVS-infected BMDM cocultured with NK cells enriched from spleens of naïve WT and PRF^−/− mice. Replication of LVS within BMDM over a period of 24 h was markedly decreased in the presence of WT NK cells but not PRF^−/− NK cells, suggesting that NK cell-mediated LVS killing required perforin (Fig. 7A, left). This NK cell-mediated LVS killing could be further enhanced by preincubation of NK cells with F. novicida ΔfopC immune sera (Fig. 7A, right). In these experiments, NK cells were incubated with serum from mock-immunized or F. novicida ΔfopC-immunized mice (1:20 dilution) for 20 min before coculture with BMDM. As shown in Fig. 7A, WT NK cells incubated with sera from PBS-treated mice significantly reduced LVS replication in BMDM compared to non-serum-treated NK cells (5.7 ± 1.5 ×10⁴ versus 1.8 ± 0.2 ×10⁶ CFU/well at 24 h). Incubation of WT NK cells with serum from F. novicida ΔfopC-immunized mice further reduced LVS replication within BMDM (8.5 ± 1.9 ×10³ CFU/well at 24 h). Deletion of perforin totally abrogated NK cell-mediated LVS killing even in the presence of F. novicida ΔfopC immune sera (Fig. 7A, right). Similar to T cells, NK cell perforin-mediated LVS killing was associated with granzyme secretion and was independent of IFN-γ production (Fig. 7B and C).

Fig 7.

Inhibition of LVS replication in BMDM by NK cells preincubated with anti-F. novicida ΔfopC serum is perforin and granzyme dependent. LVS-infected BMDM were cocultured with NK cells enriched from spleens of naïve WT or PRF^−/− mice and either treated with serum from mice that had been mock or F. novicida ΔfopC immunized or left untreated. (A) Viable LVS within BMDM was enumerated at 3 and 24 h after infection. Replication of LVS within BMDM over a period of 24 h was markedly decreased (P < 0.05) in coculture with WT NK cells compared to coculture with PRF^−/− NK cells and BMDM alone (left). LVS replication within BMDM was significantly reduced when cells were cocultured with immune serum-treated WT NK cells, compared to coculture with WT NK cells treated with serum from mock-immunized mice (**, P < 0.01) or immune serum-treated PRF^−/− NK cells (***, P < 0.001) (right). Granzyme (B) and IFN-γ (C) levels in the supernatants of BMDM alone or in BMDM-NK cell coculture were determined by ELISA. The difference in granzyme production between WT and PRF^−/− NK cells was significant (P < 0.01). Results are representative of two independent experiments.

Collectively, our data suggest that perforin- and granzyme-mediated control of LVS replication in macrophages by T and NK cells is one of the mediators of the immune response against F. tularensis infection induced by F. novicida ΔfopC vaccination.

Inhibition of LVS replication in BMDM by NK cells is mediated via an antibody-dependent cellular cytotoxicity pathway.

F. novicida ΔfopC-immune sera appeared to play an important role in enhancing the cytotoxic activity of NK cells in control of LVS replication. NK cells express an activating Fc receptor (FcγRIIIa) that mediates antibody-dependent cellular cytotoxicity (ADCC) and production of immune-modulatory cytokines in response to antibody-coated targets (67, 68). To confirm that ADCC was indeed employed by NK cells in the killing of F. tularensis, we compared LVS replication in BMDM cocultured with NK cells isolated from WT mice and from FcγR^−/− mice (68), which are deficient in FcγRI and FcγRIII. We observed that LVS replication increased the number of organisms by about 3 log units in BMDM from 3 to 24 h, in contrast to almost no viable LVS increase in BMDM cocultured with WT NK cells pretreated with F. novicida ΔfopC-immune serum. However, this marked LVS inhibition was completely abrogated by coculture with FcγR^−/− NK cells (Fig. 8A). High levels of both granzyme (Fig. 8B) and IFN-γ (Fig. 8C) secretion in coculture of BMDM with NK cells coincided with marked LVS inhibition by the WT NK cells. Additionally, a significant reduction of granzyme (Fig. 8B) and IFN-γ (Fig. 8C) production was found in infected BMDM cocultured with FcγR^−/− NK cells compared to those cultured with WT NK cells, suggesting that both IFN-γ and granzymes may be involved in NK cell FcγR receptor-mediated LVS killing. Collectively, these results further demonstrate that NK cells employ ADCC in control of LVS infection.

Fig 8.

NK cell-mediated ADCC is FcγR dependent. LVS-infected BMDM were cocultured with WT and FcγR^−/− NK cells treated with serum from mock-immunized or F. novicida ΔfopC-immunized mice. (A) Viable LVS within BMDM was enumerated at 3 and 24 h after infection. LVS replication within BMDM was significantly reduced when BMDM were cocultured with immune serum-treated WT NK cells, compared to coculture with mock serum-treated WT NK cells (**, P < 0.01) or immune serum-treated FcγR ^−/− NK cells (***, P < 0.001). Granzyme (B) and IFN-γ (C) levels in the supernatants of BMDM alone or BMDM-NK cell cocultures were determined by ELISA. The difference in granzyme and IFN-γ production between WT and PRF^−/− NK cells was significant (***, P < 0.001; *, P < 0.05). Results are representative of two independent experiments.

DISCUSSION

In this study, we showed that the two live attenuated F. novicida mutant strains, F. novicida ΔfopC and F. novicida ΔiglB, utilize different host immune components to induce protection against pulmonary Francisella LVS challenge. Specifically, we demonstrated that perforin may play an essential role in F. novicida ΔfopC-mediated immunity but is dispensable for F. novicida ΔiglB-mediated protection against LVS infection. These analyses provide additional insight into effector components that induce protective immune responses against pulmonary tularemia.

Despite the significant differences in phagosomal escape and cytosolic bacterial replication, antigenic epitopes of F. novicida ΔfopC and F. novicida ΔiglB seem to be processed and presented by APCs using both MHC I and MHC II pathways. In the case of F. novicida ΔfopC, the bacterium escapes from phagosomes, replicates in the cytosol, and stimulates IL-1β, but it also is susceptible to phagolysis upon activation of IFN-γ (46). This may explain the presentation of F. novicida ΔfopC epitopes by both MHC I and II pathways. In contrast, small numbers of F. novicida ΔiglB may escape and replicate in the cytosol, accounting for the minimal increase in bacterial replication after 24 h within primary macrophages. However, cross-presentation of exogenous (F. novicida ΔiglB) antigens by MHC I is also possible using cellular mechanisms, as reviewed by Ackerman and Cresswell (1). One such mechanism is based on endoplasmic reticulum-mediated phagocytosis. The MHC I peptide-loading complex is recruited to phagosomes, along with a Sec61 pore complex by which phagocytosed exogenous proteins are shuttled into the cytosol, and proteolysed by the proteasome, and the generated antigenic peptides are then loaded onto the MHC I complex and presented to CD8⁺ T cells (19, 72). In support of cross-presentation, oral vaccination of WT mice with F. novicida ΔfopC and F. novicida ΔiglB resulted in comparable serum anti-LVS IgG titers (S. Sanapala and B. P. Arulanandam, unpublished observations) and similar survival rates following pulmonary LVS challenge. Additionally, comparable antigen-specific cell-mediated immune responses were induced following oral immunization with F. novicida ΔfopC and F. novicida ΔiglB, as seen by elevated IFN-γ produced by F. novicida ΔfopC- or F. novicida ΔiglB-primed T cells upon cellular recall with UV-inactivated LVS (Sanapala and Arulanandam, unpublished).

Cytotoxic T lymphocytes and NK cells may employ several mechanisms to clear intracellular bacteria, including cytokine release (15), induction of apoptosis of the host cell (27), and antimicrobial activity (64). In this regard, IFN-γ is essential for control of F. tularensis infection, and NK cells are important producers of IFN-γ during primary LVS infection in the lungs and liver (5, 37); however, the role of cytolytic activity of NK cells in control of Francisella infection is largely unknown. We show here that perforin-deficient NK cells are impaired in effective inhibition of LVS replication in macrophages. Perforin was originally thought to cause cell lysis by forming pores on the target cell membrane (51), but later studies favor the theory that perforin functions by enabling granzymes to escape from the endosomes into the cytosol of the target cell (45, 50), where the granzymes trigger caspase activation and lead to cell apoptosis. In addition to mediating the cell death pathway, the internal peptide of granzyme B (HPAYNPK) displays bactericidal action in vitro, suggesting the possibility of antibacterial activities (60). In this study, we observed minimal production of granzymes by PRF^−/− T cells and NK cells upon coincubation with LVS-infected BMDM along with the loss of the ability to control LVS replication, further suggesting a granzyme-dependent mechanism in perforin-mediated protection. This was supported by the observation of greater bacterial dissemination and reduced survival rates in F. novicida ΔfopC-vaccinated PRF^−/− mice than was observed following F. novicida ΔiglB immunization or in F. novicida ΔfopC-vaccinated WT mice. Thus, CD8⁺ T cells and NK cells may play a greater role in F. novicida ΔfopC-mediated protection against LVS challenge than in protection mediated by F. novicida ΔiglB, and this likely occurs through cytolysis-induced mechanisms. CD8⁺ T cell perforin-mediated cytotoxicity also has been reported to play a role in control of other intracellular bacteria, such as Listeria monocytogenes and Mycobacterium tuberculosis (25, 63).

Of note, optimal control of LVS replication in BMDM by NK cells via perforin release was achieved in the presence of anti-F. novicida ΔfopC sera and was dependent on FcγR receptor-mediated signaling. The importance of antibody in F. novicida ΔfopC-mediated protection is further supported by the increased susceptibility to pulmonary LVS challenge in F. novicida ΔfopC-vaccinated B-cell-deficient mice (50% survival rate [Sanapala and Arulanandam, unpublished]) compared to wild-type mice (100% survival). Antibody-mediated protection against F. tularensis also has been reported by us (48, 52, 54) and others (16, 17, 28, 65), showing that adoptive transfer of anti-F. novicida mutant, anti-LVS, or anti-Francisella LPS antibodies conferred partial protection against LVS and type B Francisella challenge. Using genetically deficient mice, Kirimanjeswara et al. (28) further demonstrated that FcγR-mediated opsonophagocytosis plays an important role in antibody-mediated protection against i.n. LVS lethal challenge. Thus, specific antibodies generated by F. novicida ΔfopC immunization may contribute to the control of LVS infection both by facilitating LVS uptake via FcγR-mediated opsonophagocytosis into APCs and by LVS killing via FcγR-dependent ADCC by NK cells.

Taken together, results of our study suggest that perforin- and granzyme-mediated cytotoxic effector functions are essential for protective immunity against F. tularensis following F. novicida ΔfopC vaccination.