Metabolic Reprogramming of Host Cells by Virulent Francisella tularensis for Optimal Replication and Modulation of Inflammation

Activation of macrophages and dendritic cells requires a complex network of cellular processes to generate proinflammatory cytokines and anti-microbial products for control of pathogen replication. One of these processes involves regulated changes in host metabolism. Specifically, activation and maturation of macrophages and dendritic cells following engagement of pattern recognition receptors (PRR) is dependent on a shift in metabolism from oxidative phosphorylation to aerobic glycolysis (1). Induction of aerobic glycolysis generates several products required for host cell activation and maturation (2). First, the glycolytic process evokes rapid production of ATP to fuel the increased energy needs of the activated cell. Glycolysis also initiates metabolic pathways that generate substrates required for activation of transcription factors responsible for promoting expression of genes involved in potentiating glycolysis and proinflammatory cytokines.

Francisella tularensis ssp. tularensis is a facultative intracellular bacterium that causes a lethal disease known as tularemia. There are ∼200 cases of tularemia each year in the United States and ∼500–1000 cases in Europe; thus, this pathogen represents a consistent public health problem (3, 4). Additionally, F. tularensis ssp. tularensis was developed as a biological weapon by the United States, the former Soviet Union, and Japan (5). These programs also developed at least one antibiotic-resistant strain. Given the lack of an approved vaccine for use against natural and nefarious exposures and the presence of antibiotic-resistant strains, there is a need for novel vaccines and therapeutics.

Although F. tularensis ssp. tularensis can infect a variety of cell types, the preferred cellular targets of this pathogen in vivo are macrophages and dendritic cells (6). An important mechanism of virulence for F. tularensis ssp. tularensis is its ability to evade and suppress induction of innate immune responses in these cells that would normally control replication and dissemination of F. tularensis ssp. tularensis (7–9). However, the mechanisms and bacterial products responsible for modulating the host response have not been comprehensively defined. In the last several years, the presence of an O-antigen capsule and the genes required for capsule generation in virulent F. tularensis ssp. tularensis have been confirmed (10–12). Since then, the F. tularensis ssp. tularensis capsule has been shown to contribute to resistance of F. tularensis ssp. tularensis to killing via the alternative pathway of complement activation following exposure to human serum (12). Also, under specific conditions, presence of capsule on Francisella has also been reported to influence bacterial replication in the intracellular compartment (11). Collectively, these features suggest that capsule contributes to the virulence of the bacterium. Indeed, defined mutations in capsule synthesis genes result in modest attenuation of Francisella in vivo (13). Despite these advances, the contribution of capsule in the evasion suppression of proinflammatory responses is largely unexplored.

Bacterial capsules are generally thought to contribute to evasion of host defenses by cloaking immunostimulatory structures present on the bacterial surface, providing resistance to complement-mediated killing and limiting phagocytosis by host cells (as reviewed in Ref. 14). However, there are some data suggesting that capsular material may take a more active role in modulating the immune response by suppressing the ability of host cells to mount protective inflammatory responses (15).

In this study, we demonstrate that F. tularensis ssp. tularensis capsule aids in direct suppression of inflammatory responses in vitro and in vivo. Furthermore, we provide evidence that F. tularensis ssp. tularensis capsule mediates this suppression by reprogramming host metabolism via inhibition of the shift to aerobic glycolysis required for macrophage activation. Finally, we show that inhibition of host cell glycolysis is essential for the early intracellular replication of F. tularensis ssp. tularensis. Taken together, our findings identify a unique intercept of bacterial pathogenesis and modulation of host metabolism.

C57BL/6J wild-type mice were purchased from The Jackson Laboratory (Bar Harbor, ME). Mice were housed in Animal Biosafety Level 2 and Animal Biosafety Level 3 animal facilities at Rocky Mountain Laboratories and were provided food and water ad libitum. All research involving animals was conducted in accordance with Animal Care and Use Committee guidelines, and animal protocols were approved by the Animal Care and Use Committee at Rocky Mountain Laboratories. Bone marrow–derived macrophages (BMDM) were generated as previously described (16). As indicated, BMDM were treated with 2-deoxyglucose (2-DG; Sigma-Aldrich, St. Louis, MO) 2 h prior to infection or exposure to capsule.

Stock cultures of F. tularensis ssp. tularensis SchuS4 (Dr. Jeannine Peterson, Centers for Disease Control and Prevention, Ft. Collins, CO), SchuS4Δ1238 (Δ1238), SchuS4Δ1236, and SchuS4Δ1464c (Dr. Bradley Jones, University of Iowa, Iowa City, IA) were generated and used as previously described (8). SchuS4Δ1236 and Δ1238 were previously characterized as capsule mutants due to deletions in the waa locus encoding enzymes responsible for adding O-antigen subunits to the LPS core or adding sugars to the LPS core, respectively (13). SchuS4Δ1464c was also previously described and is devoid of capsule due to the mutation in the dTDP-glucose 4,6-dehydratase that enables formation of 6-deoxy sugars that make up the capsular structure (17). All experiments were performed under approved Biosafety Level 3 protocols at Rocky Mountain Laboratories.

Capsule was purified from F. tularensis ssp. tularensis as previously described with minor modifications (10). Briefly, F. tularensis ssp. tularensis grown on modified Mueller–Hinton (MMH) agar was collected in buffer containing 6 mM Tris, 10 mM EDTA, and 2% (w/v) SDS at pH 6.8 and incubated for 24 h at 65°C, followed by addition of 50 μg/ml proteinase K and an additional incubation at 37–42°C for 24 h. SDS was removed by ethanol precipitation, samples were centrifuged at 12,000 × g at 4°C, and the resulting pellets were resuspended in 10 mM Tris-base/10 mM CaCl₂ (pH 7.4) containing 80 U micrococcal nuclease (Sigma-Aldrich). Samples were incubated at 37°C for 24 h followed by addition of an equal volume of phenol and an additional incubation at 65°C for 30 min. Samples were then cooled on ice and centrifuged at 2000 × g for 10 min at 4°C. The aqueous layer was collected. The phenol layer was back extracted with deionized water. Aqueous layers were combined and phenol removed via ethanol precipitations. Following precipitation, the pellet was resuspended in HPLC-grade water (Sigma-Aldrich) containing 5% (v/v) Triton X-114 and incubated at 4°C for 24 h followed by incubation at 37°C for 1 h. Samples were centrifuged at 2000 × g for 10 min and the upper aqueous phase containing capsule was collected and lyophilized. Capsule was resuspended in sterile tissue culture–grade water (Life Technologies, Grand Island, NY) before use.

BMDM were infected with various strains of F. tularensis ssp. tularensis as previously described (16). Briefly, bacteria were resuspended at an MOI of 50 in 250 μl complete DMEM (cDMEM; DMEM with glutamine, HEPES, and nonessential amino acids added [Life Technologies], plus 10% heat-inactivated FBS [Thermo Fisher, Waltham, MA]). Actual inoculum was confirmed by plating on MMH agar. Medium from BMDM was removed and reserved for replacement after infection. Bacteria were added to BMDM and incubated at 37°C for 90 min. Infected BMDM were subsequently incubated with gentamicin (50 μg/ml for 45 min at 37°C) to eliminate extracellular bacteria and washed three times with PBS. As indicated, in some experiments the gentamicin incubation step was eliminated. Where indicated, BMDM were infected with bacteria utilizing centrifugation as previously described (18). Cells were washed three times with PBS and cultured in reserved cDMEM. R848 (Enzo Life Sciences, Farmingdale, NY) was added at 5 ng/ml either 90 min or 8 h postinfection. Intracellular bacteria were enumerated at the designated time points as previously described (16). Briefly, BMDM were incubated with gentamicin as described above. Cells were washed three times followed by lysis with sterile water. Cell lysates were serially diluted and plated on MMH agar plates for enumeration of colonies. Eight or 16 h postinfection cell lysates and culture supernatants were collected for assessment of mRNA, cellular protein, and cytokines.

F. tularensis ssp. tularensis capsule was diluted in cDMEM, vortexed, and added to macrophages at the indicated concentrations. Water was diluted using the same scheme and was used as vehicle control (mock). Eighteen hours after addition of capsule, cells were stimulated with the following TLR agonists: 5 ng/ml Pam3CSK4 (TLR2) or 5 ng/ml R848 (TLR8) (both from Enzo Life Sciences). Eight or sixteen hours later, cell lysates and culture supernatants were collected for assessment of mRNA, cellular protein, and cytokines.

Mice were anesthetized by i.p. injection of 100 μl 12.5 mg/ml ketamine plus 3.8 mg/ml xylazine. Animals were intranasally inoculated with 50 CFU SchuS4, Δ1238, or 10 μg purified F. tularensis ssp. tularensis capsule in 25 μl PBS. Mice receiving PBS alone served as negative controls for infected or capsule-treated mice. Sixteen hours later mice were anesthetized as described above and given 50 μg/25 μl R848 in PBS or 25 μl PBS alone (mock-treated controls). Four hours later mice were euthanized, tracheas were exposed and cannulated with a disposable 18-gauge catheter, and airways were flushed with 0.5 ml PBS. Lavage fluid was collected for assessment of cytokines.

Cell death was measured as release of lactate dehydrogenase into culture supernatants using a CytoTox 96 nonradioactive cytotoxicity kit (Promega, Madison, WI) according to the manufacturer’s instructions. TNF-α, IL-6, IL-12p40 (all from BD Biosciences, San Jose, CA), and CCL5 and CXCL1 (both from R&D Systems, Minneapolis, MN) present in cell culture supernatants or bronchoalveolar lavage fluid were quantitated using commercially available ELISA kits following manufacturers’ instructions.

At the indicated time points, medium was removed from BMDM. Cells were lysed in 150 μl 1× cell lysis buffer (Cell Signal Technology, Danvers, MA) supplemented with PMSF (Sigma-Aldrich). Lysates were added to NuPAGE LDS sample buffer (Life Technologies), heated at 95°C for 10 min, homogenized by centrifugation using a QIAshredder (Qiagen, Valencia, CA), and immediately placed on ice prior to loading onto 4–12% SDS-NuPAGE gradient gels (Life Technologies). Western blots were generated from SDS-NuPAGE gels as previously described (19). Blots were probed with Abs to phospho-p44/42, p44/42, hypoxia inducible factor 1α (HIF-1α), or β-actin (13E5) (all from Cell Signaling Technology). RNA was purified using RNeasy kits (Qiagen), cDNA was generated using the SuperScript VILO kit (Life Technologies), and real-time quantitative PCR was run using primer/probe sets for pfkfb3 (ID no. Mm00504650_m1) (all from Life Technologies) and an ABI 7900HT (Life Technologies). Input RNA was normalized to hypoxanthine phosphoribosyltransferase. Fold change of the indicated genes as compared with untreated, uninfected controls was quantified as ΔΔCT.

Lactate present in culture medium was quantitated using an Amplex red glucose/glucose oxidase kit (Life Technologies) following the manufacturer’s instructions, substituting lactate and lactate oxidase (both from Sigma-Aldrich) for glucose and glucose oxidase.

Statistical differences between two groups were determined using a two-tailed Student t test. For comparison between three or more groups, analysis was done by one-way ANOVA followed by a Tukey multiple comparisons test. Significance using both methods was determined at p < 0.05.

Previous reports have shown that the presence of F. tularensis ssp. tularensis capsule affects host cell death and replication efficiency of the bacterium in a species-dependent manner. For example, in human monocyte-derived macrophages, F. tularensis ssp. tularensis capsule mutants replicated exponentially during the first 16 h of infection similar to wild-type F. tularensis ssp. tularensis, followed by induction of dramatic host cell death that was not observed in wild-type F. tularensis ssp. tularensis–infected cells (12). Following infection of mouse BMDM, the capsule mutant bacteria also appeared to induce increased cell death distinct from that observed in wild-type F. tularensis ssp. tularensis–infected cells. However, in BMDM capsule-deficient F. tularensis ssp. tularensis did not replicate during the first 24 h (11). The contribution of capsule to evasion of triggering inflammatory responses and how this might contribute to bacterial replication and/or host cell death was not evaluated in either study. Thus, we first determined whether we could replicate previous findings with regard to replication and cell death in addition to examining elicitation of cytokine production in BMDM infected with various F. tularensis ssp. tularensis capsule mutants.

In agreement with previous work, F. tularensis ssp. tularensis capsule mutants were nearly eliminated from the intracellular compartment of BMDM within the first 24 h of infection. However, replication was partially restored during the last 48–72 h infection (Fig. 1A). During the first 24 h of infection, none of the capsule mutants tested induced significant cell death. Rather, statistically significant differences in cell death among BMDM infected with capsule mutants compared with mock-infected controls were not noted until 72 h postinfection (Fig. 1B). In contrast to induction of cell death, capsule was required for evasion of induction of inflammatory responses as indicated by the significant increase in secretion of IL-12p40 and CCL5 among BMDM infected with F. tularensis ssp. tularensis capsule mutants compared with cells infected with wild-type F. tularensis ssp. tularensis– and mock-infected controls (Fig. 1C). There were small, but notable, differences in the ability of each mutant to trigger cytokine production. Δ1238 induced the greatest amount of both IL-12 and CCL5 in the first 24 h postinfection. By 48 h postinfection all capsule mutants were eliciting cytokine secretion (Fig. 1C). Similar to our previous findings, wild-type F. tularensis ssp. tularensis grew exponentially during the 72-h culture period and did not induce detectable cytokine production until 72 h postinfection. However, this was paired with almost complete eradication of the cellular monolayer (unpublished observation). Thus, the direct contribution of microbial products versus massive cell death in wild-type–infected F. tularensis ssp. tularensis cultures was difficult to discern. In the last 72 h, statistically significant cell death among all infected cells compared with uninfected cells was noted (Fig. 1B).

The early control followed by recovery of F. tularensis ssp. tularensis capsule mutant replication in combination with triggering early cytokine response was similar to findings following in vivo infection with these mutants. However, our results did not replicate previously published in vitro work with mouse-derived cells (11). In the previous study, the method of infection was different and involved centrifugation of BMDM with bacteria. It was possible that this difference in methodology contributed to the disparity with our findings. To confirm this hypothesis we performed infections in BMDM duplicating the method of infection described in the previous study. Using this technique, we observed early control of F. tularensis ssp. tularensis capsule mutant replication accompanied by significant host cell death (Supplemental Fig. 1). We also observed that capsule mutants triggered cytokine secretion from BMDM similar to that observed using the infection method described in the present study. Therefore, the method of infection of BMDM altered the ability of F. tularensis ssp. tularensis capsule mutants to trigger cell death, but not cytokine production. Centrifugation of host cells can alter receptor expression and change the outcome of infection in other models of host–pathogen interaction (20). Our data suggest that a “passive” infection model, as opposed to centrifugation, may more closely mimic the interaction of F. tularensis ssp. tularensis with host cells in the in vivo environment, and thus this method was used for the remainder of our study.

The data presented above demonstrated that capsule is required for evasion of detection by macrophages early after infection. We hypothesized that capsule also contributes to the ability of F. tularensis ssp. tularensis to suppress induction of inflammatory responses. We first examined the ability of a capsule mutant to suppress cytokine secretion in macrophages early after infection that had also been treated with an unrelated secondary stimuli. Treating newly infected cells with gentamicin followed by washing with PBS resulted in elimination of nearly all of the capsule mutants within 24 h of infection. Failure of these mutants to inhibit host cell responses could be a factor of too few bacteria present in the cell. Therefore, we modified our protocol to ensure that similar numbers of wild-type and capsule mutant bacteria were present at the time of stimulation and that there was not significant cell death among infected macrophages as compared with uninfected controls (Fig. 2A). We also selected Δ1238 to test as a representative capsule mutant. As expected, wild-type F. tularensis ssp. tularensis–infected cells treated with R848 or Pam3CSK4 secreted significantly less IL-12p40 compared with treated, mock-infected cells (Fig. 2A). In contrast, despite similar numbers of intracellular bacteria, Δ1238 was unable to impair macrophage secretion of cytokine in response to R848 or Pam3CSK4 (Fig. 2A).

We also compared the ability of wild-type and capsule mutant F. tularensis ssp. tularensis to suppress inflammatory responses in vivo. Similar numbers of wild-type F. tularensis ssp. tularensis and Δ1238 were recovered from the lungs of infected mice (Fig. 2B). As previously observed, wild-type F. tularensis ssp. tularensis did not trigger inflammation within the first 24 h of infection, and the organism significantly impaired secretion of TNF-α and IL-6 in response to R848 (Fig. 2B) (7). Animals infected with wild-type F. tularensis ssp. tularensis also had less, but not significantly different, CXCL1 in their bronchoalveolar lavage following exposure to R848. Similar to wild-type F. tularensis ssp. tularensis, Δ1238 did not induce proinflammatory cytokines following a low-dose infection (Fig. 2B). However, consistent with our in vitro data, the capsule mutant was unable to inhibit secretion of TNF-α, IL-6, and CXCL1 in response to R848 (Fig. 2B). Taken together, our data show that the presence of capsule on viable bacteria was required for suppression of proinflammatory responses in vitro and in vivo.

We next confirmed that F. tularensis ssp. tularensis capsule, independent of intact bacteria, triggers a suppressive program in macrophages. We tested purified F. tularensis ssp. tularensis capsule for its ability to stimulate and/or suppress proinflammatory responses in macrophages. Capsule did not induce secretion of IL-12p40 from macrophages (Fig. 3A). However, capsule-treated cells secreted significantly less cytokine in response to R848 and Pam3CSK4 (Fig. 3A). We then determined whether purified capsule could inhibit production of cytokines in the airways in vivo. Similar to our in vitro findings, capsule did not trigger production of proinflammatory cytokines in the airways (Fig. 3B), nor did capsule induce an influx of inflammatory cells or depletion of alveolar macrophages (unpublished observations). Four hours after treatment with R848, no influx of inflammatory cells was observed in mock-treated or capsule-treated animals (unpublished observations), nor did capsule impair secretion of TNF-α or IL-6 into the airways following administration of R848. However, we did observe significantly less CXCL1 in bronchoalveolar lavage fluid among capsule-treated animals compared with mock controls that received R848 (Fig. 3B). Thus, F. tularensis ssp. tularensis capsule recapitulates the ability of intact F. tularensis ssp. tularensis to suppress host cell inflammatory cytokine production in vitro and partially mimics this response in vivo.

Activation of macrophages and secretion of many proinflammatory cytokines are dependent on a shift in metabolism from oxidative phosphorylation to aerobic glycolysis (as reviewed in Ref. 1). We hypothesized that F. tularensis ssp. tularensis capsule may inhibit this shift as a mechanism to impair cytokine production by macrophages. Therefore, we examined the ability of F. tularensis ssp. tularensis capsule to modulate induction of host cell glycolysis in three ways. Lactate is a secreted product of aerobic glycolysis and is increased in the supernatant following activation of cells with TLR ligands (21). In agreement with these previous findings, R848 induced an increase in lactate secretion among macrophages and treatment with F. tularensis ssp. tularensis capsule significantly impaired this response (Fig. 4A). A shift to aerobic glycolysis among host cells can also be monitored by the stabilization of the transcription factor HIF-1α and the gene it targets, that is, pfkfb3 (21). Similar to our findings with lactate secretion, R848 increased stabilized HIF-1α and increased expression of pfkfb3 whereas F. tularensis ssp. tularensis capsule significantly inhibited each of these responses (Fig. 4B, 4D). Translation of the HIF-1α gene and activation of HIF-1α are controlled, in part, by activation of the MAPK p44/42 (22). Thus, we also determined whether capsule inhibited activation of p44/42 triggered by R848. Stimulation of BMDM with R848 induced rapid phosphorylation of p44/42, and pretreatment of BMDM with capsule inhibited this activation (Fig. 4C).

To confirm that modulation of host cell metabolism by F. tularensis ssp. tularensis capsule was not a peculiarity of purified capsule, we also compared the ability of wild-type F. tularensis ssp. tularensis and Δ1238 to induce and/or inhibit host cell glycolysis. Because the bacteria are also metabolically active, we monitored the host cell–specific markers of aerobic glycolysis, including stabilization of HIF-1α, activation of p44/42, and transcription of pfkfb3. Consistent with its ability to evade triggering host cell processes, wild-type F. tularensis ssp. tularensis did not significantly affect stabilization of HIF-1α or increase expression of pfkfb3 (Fig. 5A, 5C). However, F. tularensis ssp. tularensis infection resulted in dephosphorylation of p44/42 in host cells (Fig. 5B). In contrast, Δ1238 stabilized HIF-1α, did not inhibit phosphorylation of p44/42, and induced expression of pfkfb3 (Fig. 5A–C). Following treatment with R848, wild-type F. tularensis ssp. tularensis–infected macrophages had significantly less HIF-1α, and F. tularensis ssp. tularensis impaired R848-mediated phosphorylation of p44/42 (Fig. 5D, 5E). Consistent with decreased HIF-1α, F. tularensis ssp. tularensis–infected cells also had significantly fewer transcripts encoding pfkfb3 following R848 treatment compared with uninfected controls (Fig. 5F). In contrast, Δ1238-infected cells had significantly increased expression of pfkfb3 upon stimulation with R848 compared with uninfected controls (Fig. 5F), suggesting that Δ1238 was promoting, or unable to inhibit, the development of an activated glycolytic program among infected cells. Therefore, wild-type F. tularensis ssp. tularensis modulates host cell metabolism by impairing a shift to aerobic glycolysis, and capsule contributes to this process.

Induction of aerobic glycolysis alters both the intracellular pool of metabolites and induces pathways that could contribute to control of bacterial replication. The killing of Δ1238 during the first 24 h of infection could be attributed to bacterial-driven induction of aerobic glycolysis. Thus, we tested whether inhibition of aerobic glycolysis would restore replication of Δ1238. Because F. tularensis ssp. tularensis capsule independently inhibited host cell glycolysis, we first determined whether addition of capsule to Δ1238-infected cells would impact bacterial survival and production of IL-12p40. Addition of exogenous capsule resulted in significantly higher numbers of Δ1238 bacteria 24 h postinfection compared with mock-treated controls (Fig. 6A). Similarly, in the presence of capsule Δ1238 triggered significantly less IL-12p40 compared with mock-treated, Δ1238-infected cells (Fig. 6A). Although the addition of capsule attenuated killing of Δ1238 during the first 24 h of infection, it did not enhance replication of the bacteria. It was possible that inhibition of aerobic glycolysis mediated by the concentration of capsule was not sufficient to reverse inhibition of Δ1238 replication. Therefore, we repeated the experiments in the presence of a specific and potent inhibitor of aerobic glycolysis, 2-DG (23). Addition of 2-DG had no effect on the ability of wild-type F. tularensis ssp. tularensis to replicate, nor was there any difference in cytokine production among wild-type F. tularensis ssp. tularensis–infected BMDM compared with vehicle controls (Fig. 6B). In marked contrast, BMDM treated with 2-DG had statistically significantly more Δ1238 compared with mock-treated controls (Fig. 6B). Treatment of BMDM with 2-DG also resulted in less secretion of IL-12p40 by Δ1238-infected cells (Fig. 6B).

Virulent F. tularensis causes a lethal disease following infection of host cells with as few as 1–2 bacteria per cell and 10–15 bacteria in the entire inoculum. Given this low infectious dose, it suggests that successful infection is greatly dependent on the ability of the bacteria to both evade and suppress antimicrobial responses in host cells from the moment of contact through the course of infection. The intersection of bacterial replication, manipulation of innate immunity, and host metabolism have not been fully explored with regard to tularemia. Our present study provides evidence that F. tularensis ssp. tularensis capsule plays a key role in the early evasion and suppression of proinflammatory responses in vitro and in vivo and that this process, in part, involves modulating host cell metabolism.

Initially, it was not surprising that strains of F. tularensis ssp. tularensis lacking capsule were unable to evade detection by host cells. It is appreciated that one function of bacterial capsules is to shield immunogenic structures present on the surface of the microorganisms (14). Indeed, attenuated strains of Francisella lacking capsule triggered even higher proinflammatory responses than did encapsulated wild-type strains (24, 25). It is likely in our model that F. tularensis ssp. tularensis capsule acts to shield immunogenic proteins present on the surface, and it is the recognition of these proteins by host cells infected with capsule mutants that triggers the early transition to aerobic glycolysis. In contrast to the activity of capsules shielding structures on the surface of bacteria, there is less evidence for bacterial capsules directly dampening the host response, as we observed in the present study with purified F. tularensis ssp. tularensis capsule. The best example of a bacterial capsule triggering anti-inflammatory responses is the polysaccharide A capsule derived from Bacteroides fragilis. However, the anti-inflammatory mechanism of this glycan lies not within direct impairment of macrophages or dendritic cells, but rather its ability to induce expansion of IL-10–producing CD4⁺ T cells following its presentation to T cells by dendritic cells (26). Thus, the ability of F. tularensis ssp. tularensis capsule to direct an anti-inflammatory program among macrophages in the absence of other cell types is a unique feature of virulence embodied by these bacteria.

After establishing that capsule played a role in inhibiting inflammatory responses in macrophages, we next determined the mechanism by which capsule executed this suppressive program. As a polysaccharide, F. tularensis ssp. tularensis capsule could have initiated anti-inflammatory responses in host cells following engagement of glycan-binding receptors present on the cell surface. We analyzed the potential role of a wide variety of receptors known to interact with microbial carbohydrates, including mannose receptor, scavenger receptor A, CD11b, Mincle, and Dectin-1. However, none of these receptors was required for capsule-mediated inhibition of proinflammatory responses in macrophages. We have previously established that F. tularensis ssp. tularensis lipids use TLR2 to antagonize general inflammatory responses in host cells (27). However, we confirmed that TLR2 was not required for capsule-mediated inhibition of proinflammatory responses in macrophages (unpublished observations). Alternatively, F. tularensis ssp. tularensis capsule may nonspecifically bind surface-exposed TLRs inhibiting the ability of TLR agonists to interact with their cognate receptor. However, we routinely observed F. tularensis ssp. tularensis–mediated inhibition of TLR7/8 signaling. These receptors are only found associated with acidic compartments within the host cytosol. The agonists for TLR7/8 are small nucleic acids. Thus, if capsule were covering these TLR receptors, it would have to gain access to the cytosol and acidic compartments and mimic the ability of nucleic acids to bind to these receptors. In our opinion, it does not seem likely that a carbohydrate such as F. tularensis ssp. tularensis capsule would be able to fulfill both of these requirements for inhibiting TLR7/8-mediated signaling.

As an intracellular bacterium that replicates in the cytosol of the host cell, by default the F. tularensis ssp. tularensis capsule could engage any number of other host pathways that guide macrophage activation and production of proinflammatory cytokines. Transition in host cell metabolism is a key feature for induction of inflammation. Specifically, induction of aerobic glycolysis is a requirement for macrophages and dendritic cells to efficiently activate production of proinflammatory cytokines and antimicrobial molecules (28–30). Additionally, shifts in host cell metabolism also alter the composition of intracellular metabolites and other pathways such as autophagy that may impact survival of intracellular bacteria (31). Thus, we hypothesized that capsule-mediated interference with induction of host cell aerobic glycolysis could explain both the inability of capsule-deficient mutants to evade and suppress inflammatory responses as well as their survival. We established that both capsule and wild-type F. tularensis ssp. tularensis impair induction of host aerobic glycolysis and provided evidence that a capsule mutant is incapable of this process. Importantly, supplementation of capsule to macrophages infected with capsule-deficient F. tularensis ssp. tularensis partially restored bacterial replication and dampened the production of cytokine triggered by infection. The mechanism by which F. tularensis ssp. tularensis capsule modulates the transition to aerobic glycolysis in host cells was not clear. The AKT/mammalian target of rapamycin (mTOR) pathway has been implicated in the induction of aerobic glycolysis in cancer cells. It has been suggested that F. tularensis ssp. tularensis impairs mTOR activity as measured by phosphorylation of the mTOR target, S6 ribosomal protein (32). Thus, we examined activation of mTOR among cells infected with wild-type versus capsule mutant F. tularensis ssp. tularensis. We did not observe significant changes in mTOR activity among wild-type or capsule mutant-infected BMDM (unpublished observations). This suggested that suppression of aerobic glycolysis by wild-type F. tularensis ssp. tularensis may be independent of the mTOR pathway. As an alternative to directly targeting host metabolic pathways, other bacterial capsules have been documented to act as mimics for host carbohydrates (33). It is possible that F. tularensis ssp. tularensis capsule is acting in that capacity in the present study. Specifically, F. tularensis ssp. tularensis capsule may engage specific host enzymes required to support aerobic glycolysis and as a consequence outcompete the natural ligands.

As stated above, a shift in host cell aerobic glycolysis impacts the presence of antimicrobial effector molecules such as reactive oxygen species (ROS) (34). Thus, the early replication defect observed among capsule-deficient mutants may be a result of increased presence of ROS. We examined the role of NADPH-derived ROS utilizing gp91/nos2-deficient macrophages and found that absence of this molecule did not restore the ability of capsule mutants to replicate during the first 24 h (unpublished observations). Mitochondrial ROS (mROS) is an additional source of ROS in host cells that is increased during aerobic glycolysis (35). We have previously established that enhanced susceptibility of attenuated F. novicida to mROS contributes to the ability of these organisms to promote inflammatory responses in macrophages (36). However, inhibition of mROS also reverses aerobic glycolysis (35). Therefore, it is not currently possible to separate the role for mROS independent of other features of aerobic glycolysis that may affect F. tularensis ssp. tularensis replication.

In cancer cells and fibroblasts associated with tumors undergoing aerobic glycolysis, autophagy is increased. Autophagy is a critical part of maintenance of the general physiology of host cells, delivering spent or unnecessary cytosolic content to lysosomes for degradation. Autophagy also plays a role in control of intracellular pathogens (37). A recent study examining two capsule mutants of F. tularensis ssp. tularensis suggested that the impaired replication observed in mouse cells was due to failure of these mutants to evade being taken up into autophagic vacuoles (11). Although they did not observe an increase in autophagic vacuoles following infection with capsule mutants, it is possible that the method used to quantitate these compartments (microscopy) was not sensitive enough to detect differences triggered by capsule mutants versus wild-type F. tularensis ssp. tularensis. Therefore, it is possible that the early defect in survival of capsule mutants in mouse macrophages is due to increased autophagy triggered by aerobic glycolysis. Alternatively, as the metabolic content of host cells shifts during aerobic glycolysis, host cell metabolites critical for F. tularensis ssp. tularensis replication may be depleted, resulting in impaired replication of F. tularensis ssp. tularensis.

Regardless of the mechanism by which induction of aerobic glycolysis contributes to limiting F. tularensis ssp. tularensis replication, this process is transient and reflects the modest attenuation of F. tularensis ssp. tularensis capsule mutants in vivo. It is appreciated that sustained aerobic glycolysis can be lethal for macrophages and dendritic cells, and thus we propose that the recovery phase of F. tularensis ssp. tularensis mutants observed at 48 and 72 h postinfection reflects another shift in host cell metabolism toward one that is more favorable for replication of F. tularensis ssp. tularensis. Moreover, we suggest that a similar phenomenon occurs in vivo. Specifically, F. tularensis ssp. tularensis capsule mutants trigger aerobic glycolysis upon invading macrophages, resulting in a subset of the bacteria to be killed by the host cell. However, as the metabolism of the host cell shifts back to a resting oxidative phosphorylation, the surviving bacteria are free to replicate, which ultimately leads to a lethal outcome. Collectively, our data suggest that identification of the specific mechanism by which induction of aerobic glycolysis controls F. tularensis ssp. tularensis replication will yield new strategies and targets for novel therapeutics aimed to limit F. tularensis ssp. tularensis and resolve infection.

We thank Dr. Chris Bosio for helpful comments and review of this manuscript.

The authors have no financial conflicts of interest.