|Home | About | Journals | Submit | Contact Us | Français|
Although Francisella tularensis subsp. tularensis is known to cause extensive tissue necrosis, the pathogenesis of tissue injury has not been elucidated. To characterize cell death in tularemia, C57BL/6 mice were challenged by the intranasal route with type A F. tularensis, and the pathological changes in infected tissues were characterized over the next 4 days. At 3 days postinfection, well-organized inflammatory infiltrates developed in the spleen and liver following the spread of infection from the lungs. By the next day, extensive cell death, characterized by the presence of pyknotic cells containing double-strand DNA breaks, was apparent throughout these inflammatory foci. Cell death was not mediated by activated caspase-1, as has been reported for cells infected with other Francisella subspecies. Mouse macrophages and dendritic cells that had been stimulated with type A F. tularensis did not release interleukin-18 in vitro, a response that requires the activation of procaspase-1. Dying cells within type A F. tularensis-infected tissues expressed activated caspase-3 but very little activated caspase-1. When caspase-1-deficient mice were challenged with type A F. tularensis, pathological changes, including extensive cell death, were similar to those seen in infected wild-type mice. In contrast, type A F. tularensis-infected caspase-3-deficient mice showed much less death among their F4/80+ spleen cells than did infected wild-type mice, and they retained the ability to express tumor necrosis factor alpha and inducible NO synthase. These findings suggest that type A F. tularensis induces caspase-3-dependent macrophage apoptosis, resulting in the loss of potentially important innate immune responses to the pathogen.
Tularemia is a zoonotic infectious disease that is caused by the facultative intracellular bacterium Francisella tularensis. Little is known about the virulence properties of this pathogen or the pathogenic mechanisms responsible for the diseases it causes. Two subspecies of F. tularensis account for infections in humans. F. tularensis subsp. tularensis (type A) is by far the more virulent of the two and causes most of the lethal cases of tularemia in North America. The minimum infectious dose for type A F. tularensis strains in humans challenged by aerosol inhalation has been estimated to be less than 15 CFU (24). In the mouse, the most lethal form of exposure to type A strains is through the inhalation of viable organisms (14, 29), and infection disseminates to the spleen and liver after only a few days (11). In contrast, F. tularensis subsp. holarctica (type B) is a frequent cause of nonfatal tularemia in Europe and Asia.
The pathological features of human tularemia have been reported by Lamps et al. (22), who examined autopsy samples from confirmed cases of naturally acquired disease. These authors noted the presence of granulomas and irregular microabscesses with coagulation necrosis in the liver, spleen, kidney, and lymph nodes. Most patients also showed diffuse necrotizing pneumonia ranging in appearance from abundant fibrin and cellular debris in the alveolar walls and airways to confluent necrosis and hemorrhage. Thus, the hallmark features of end-stage tularemia in humans include foci of necrosis in the lungs, liver, and spleen indicative of extensive cell death. Similar findings have been reported for mice infected with type A F. tularensis (11), indicating that extensive in situ cell death is a defining characteristic of disseminated disease. Despite this information, little is known about the microbial factors responsible for inducing cell injury or the intracellular signaling pathways of programmed cell death that are activated during type A F. tularensis infections.
Infection of mice with the live vaccine strain (LVS) of F. tularensis subsp. holarctica continues to be one of the most frequently used animal models for studying the pathogenesis of tularemia. The minimum lethal dose for LVS in C57BL/6 mice injected by the intranasal (i.n.) route is ~103 CFU (3, 23), and under these conditions viable organisms disseminate from the lungs to the liver and spleen by the second day postinfection (p.i.) (3, 12). LVS causes pneumonia and systemic disease in mice that is similar in some respects to what is seen in infected human beings (2, 3, 12, 27). The host response to primary infection is characterized by a high level of expression of proinflammatory genes, especially in the spleen and liver (10, 16). Cells that accumulate within infected tissues include CD11b+ macrophages (M), CD3+ T cells, and Ly-6G+ immature myeloid cells that are capable of further differentiation into granulocytes and dendritic cells (DC) (3, 12, 27). In the liver, microgranulomas containing M, T cells, and myeloid precursors form early in LVS infections and continue to grow in size until the death of the host. However, only sporadic cell death occurs within hepatic granulomas and splenic pyogranulomatous infiltrates in infected mice (3). Both granuloma formation and cell death in the liver are highly dependent on gamma interferon (IFN-γ) production (3, 9), which is primarily produced during primary infections by activated NK cells and T cells within the infected organs (3, 13, 23). Thus, in contrast to type A F. tularensis, which causes significant pathological changes indicative of extensive cell death in situ, the tissue response to LVS is more inflammatory and results in less overt cytotoxic tissue damage.
Several mechanisms of programmed cell death have been associated with Francisella infections. Lai et al. (20, 21) first reported that mouse J774 M-like cells underwent apoptotic death when challenged in vitro with LVS at high multiplicities of infection. Because infected cells showed cytochrome c release from their mitochondria, the degradation of poly-ADP-ribose polymerase, and the cleavage of procaspase-9 and procaspase-3, the authors concluded that death signaling involved the intrinsic apoptosis pathway. The morphological features of apoptosis, which include cytoplasmic condensation, cell membrane blebbing, and chromatin condensation (pyknosis) (3, 9, 27), as well as the cleavage of procaspase-3 (3, 27), have also been observed in LVS-infected tissues.
A caspase-3-independent form of programmed cell death, termed pyroptosis, is initiated when procaspase-1 is activated after its recruitment to multiprotein cytosolic complexes known as inflammasomes. Activated caspase-1 catalyzes the processing of the interleukin-1β (IL-1β) and IL-18 precursor polypeptides and coordinates their release with the induction of cell death. Cells dying by pyroptosis show impaired membrane integrity, which leads to osmotic cell lysis and the release of additional proinflammatory cellular contents (15, 18, 28). Monack and coworkers (17, 18, 25) showed that procaspase-1 was activated in mouse bone marrow-derived M after their infection with either LVS or F. novicida strain U112. In their earlier studies, Lai et al. (21) had not observed the cleavage of procaspase-1 in LVS-infected J774 M-like cells, but this may reflect their use of a cell line that fails to activate the pyroptosis pathway (18).
The objective of the present study was to understand the basis for tissue injury caused by highly virulent type A F. tularensis strains in a mouse respiratory challenge model of tularemia. Specifically, we wanted to determine whether any of the mechanisms of cell death described for the less-virulent subspecies contribute to tissue injury caused by the tularensis subspecies. The results indicate that type A F. tularensis causes a caspase-3-dependent destruction of phagocytic cells early in infection, which results in the loss of potentially important innate immune responses to the pathogen.
The following primary antibodies were used in the present study: goat anti-activated caspase-1 (Santa Cruz Biotechnology, Santa Cruz, CA), rabbit anti-activated caspase-3 (Cell Signaling Technology, Danvers, MA), rabbit anti-activated capase-9 (Cell Signaling), rabbit anti-F. tularensis (BD Diagnostic Systems, Sparks, MD), rat anti-F4/80 (AbD Serotec, Raleigh, NC), rabbit anti-iNOS (Santa Cruz), and rat anti- tumor necrosis factor alpha (anti-TNF-α; BD Biosciences, San Jose, CA). Alexa Fluor 488-conjugated and Alex Fluor 568-conjugated secondary antibodies were obtained from Molecular Probes/Invitrogen (Carlsbad, CA). A universal Histostain-SP kit and biotin-conjugated secondary antibodies (Zymed Laboratories/Invitrogen, Carlsbad, CA) were used for immunoperoxidase staining. The concentrations of gamma interferon (IFN-γ), IL-18, IL-12p70, and IL-10 were determined by using enzyme-linked OptEIA immunoassays (BD Biosciences). Double-strand DNA breaks were detected by using the TACS terminal deoxynucleotidyltransferase apoptosis detection kit (R&D Systems). Granulocyte-monocyte colony stimulating factor was purchased from R&D Systems (Minneapolis, MN). Medium conditioned by the growth of L929 cells was used as a growth factor for bone marrow-derived M (19).
Six-week-old female C57BL/6J (B6) mice were purchased from Jackson Laboratories (Bar Harbor, ME) and used from ages 7 to 12 weeks. Caspase-1-deficient [NOD.129S2(B6)-Casp1tm1Sesh/LtJ] and wild-type control NOD/ShiLtJ mice were also purchased from Jackson Laboratories and used at 7 weeks of age. The NOD background carries a mutation in the Cdh23 gene, which results in age-related hearing loss and the onset of pancreatitis beginning at 12 weeks of age. IFN-γ-deficient mice (B6.129S7-Ifngtm1Ts/J) and caspase-3-deficient mice (B6.129S1-Casp3tm1Fv/J) were purchased from Jackson Laboratories and compared to B6 wild-type controls. All infected mice were maintained under biosafety level 3 containment on a 12-h-light/12-h-dark cycle with food and water provided ad libitum. Animal care and use protocols were approved by the University of Kansas Medical Center Animal Care and Use Committee.
F. tularensis subsp. holarctica LVS was obtained from Jeannine Petersen (Centers for Disease Control and Prevention, Ft. Collins, CO). The type A F. tularensis SCHU S4 strain was provided by Kevin King (Midwest Research Institute, Kansas City, MO) and was prepared under contract SHHSM266200400002C from the National Institute of Allergy and Infectious Diseases, National Institutes of Health. Type A F. tularensis strains KU49 and KU54 were from our collection, and their assignment to the tularensis subsp. was determined by PCR typing (32). All strains were grown in supplemented Mueller-Hinton broth as previously described (32). Frozen aliquots of log-phase cultures were stored at −80°C, thawed rapidly prior to use, washed once in Dulbecco phosphate-buffered saline (DPBS), and diluted in DPBS for injection into mice. For cell culture experiments, the bacteria were resuspended instead in antibiotic-free RPMI 1640 medium containing 2 mM l-glutamine and 10% heat-inactivated fetal bovine serum (FBS). Actual numbers of bacteria injected or added to cell culture were determined by serial dilution and culture on chocolate agar.
Following anesthesia with ketamine-HCl and xylazine (Phoenix Scientific, Inc., St. Joseph, MO), mice were challenged by the i.n. route by gradually administering 20 μl of a bacterial suspension containing either 20 to 50 CFU type A F. tularensis or 1 × 103 to 5 × 103 CFU LVS into one naris. Minimum lethal doses of the type A strains were defined as the smallest dose tested that resulted in lethality of all mice in a group of 5 animals within 10 days. The minimum lethal doses for the three type A F. tularensis stains studied here were <25 CFU, and mice died 5 to 7 days later with minor differences between strains. The minimum lethal dose for LVS was ~1,000 CFU, with a mean time until death of approximately 9 days (3). Total organ burdens were determined on a daily basis after i.n. challenge by mechanically disrupting the organs in an aerosol-free bag (Seward, West Sussex, United Kingdom) and culturing an appropriately diluted aliquot on chocolate agar. The minimum detection limit for this assay was 40 CFU per organ. In instances in which no colonies were detected, a value of 20 CFU was assigned for calculating group means. For calculating the total blood volume, it was assumed that blood constituted 6% of body weight. All procedures with type A F. tularensis strains were conducted under strict biosafety level 3 containment conditions. The University of Kansas Medical Center is a Centers for Disease Control and Prevention-approved select agent entity.
Mouse lungs, livers, and spleen tissues were fixed in 3% paraformaldehyde for 6 h at room temperature, followed by multiple washes with DPBS. Following sterility validation of each fixed tissue, samples were embedded in paraffin and sectioned by Histology Services at the University of Kansas Medical Center. For cryostat sectioning, replicate tissue samples were overlaid on graded concentrations of sucrose and embedded in tissue freezing medium (Electron Microscopy Sciences, Hatfield, PA) as previously described (3).
For immunoperoxidase staining of F4/80, TNF-α, and the activated caspases, deparaffinized sections were first treated for 5 min with 20 μg of proteinase K (Invitrogen)/ml, followed by treatment with 3% H2O2 in methanol for 10 min to inactivate endogenous peroxidases. Detection of iNOS did not require antigen unmasking techniques. The sections were then stained with primary antibodies against activated casapase-1, activated caspase-3, F4/80, iNOS, or TNF-α as previously described (3). After being washed, the sections were treated with biotinylated secondary antibodies and horseradish peroxidase-conjugated streptavidin. Reactions were visualized with the peroxidase substrate aminoethyl carbazole and then counterstained with hematoxylin.
Cells with double-strand DNA breaks were detected in fixed paraffin-embedded tissue sections by TUNEL (terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling) using the TACS terminal deoxynucleotidyltransferase kit as described previously (3). Sections were counterstained with methyl green.
Cryostat sections were prepared for immunofluorescence, treated with ice-cold acetone for 10 min, air dried, and rehydrated with DPBS. The sections were then blocked with 10% serum from the animal species of the secondary antibody diluted in DPBS containing 0.3% Triton X-100 and 1% bovine serum albumin. Sections were then treated with primary antibodies, washed, and incubated with Alexa Fluor 488- or Alexa Fluor 568-conjugated secondary antibodies as previously described (3). The sections were mounted in Prolong Gold containing DAPI (4′,6′-diamidino-2-phenylindole; Invitrogen) to localize dense inflammatory infiltrates. Sections were examined with a fluorescence microscope (Nikon Instruments, Inc., Melville, NY) equipped with a charge-coupled device camera. Captured images were processed by using Metamorph software (Molecular Devices Corp., Sunnyvale, CA).
Mouse bone marrow-derived M and DC were prepared as previously described (19). Briefly, M were grown from bone marrow cells for 12 days in 10% CO2 in Dulbecco modified Eagle medium (DMEM), 3.7 g of NaHCO3/liter, 10 mM HEPES, 7% L-cell-conditioned medium, 10% FBS, penicillin, and streptomycin. The medium was replaced after 7 days of culture. Adherent M were dislodged from the flasks by treatment with 0.25% trypsin-EDTA for 5 min, resuspended in antibiotic-free RPMI 1640 medium-10% FBS-10 mM HEPES, and transferred to 96-well plates (6 × 104 per well). For the preparation of DC, bone marrow cells were grown in 5% CO2 in RPMI 1640 medium, 10 mM HEPES, 10% FBS, 10 μM 2-mercaptoethanol, 20 ng of granulocyte-macrophage colony-stimulating factor/ml, penicillin, and streptomycin. The medium was replaced periodically (19) during the 12 days of culture. The nonadherent DC were then recovered, washed, resuspended in antibiotic-free complete tissue culture medium, and transferred to 96-well-plates (6 × 104 per well). The cultures of M and DC were challenged with bacteria as previously described (19), and culture supernatants were collected 24 h later.
To measure cytokine concentrations in serum samples and cell culture fluids, the OptEIA enzyme-linked immunoassays for mouse IFN-γ, IL-18, IL-12p70, and IL-10 (BD Biosciences) were performed according to the manufacturer's instructions. The limit of detection for IFN-γ in mouse serum was 60 pg/ml. The limits of detection for all cytokines in culture fluids were 5 pg/ml. The specificities of these assays have been described previously (19).
For analysis of the frequencies of granulomas and TUNEL+, F4/80+, TNF+, iNOS+, or Francisella+ cells in tissues, Poisson distributions were assumed, and comparisons of Poisson regressions were made with the numbers of microscopic fields of view (in logarithm) as the offset variables. Because these data were overdispersed, scaling factors (square root of deviance divided by the degrees of freedom) were applied. Cytokine concentrations were assumed to follow log-normal distributions. IFN-γ logarithmic data (see Fig. Fig.2B)2B) were compared by using the two-sample t test, while the cytokine data in Fig. Fig.33 were compared using Tukey's procedure to control the type I error rate in multiple comparisons. Each of the three type A F. tularensis strains (KU49, KU54, and SCHU S4) induced foci of necrosis, extensive TUNEL, and caspase-3 activation on day 4 p.i. Strain KU49 was selected as representative of this group of pathogens for further study.
To characterize the pathological changes that occur in disseminated tularemia, we challenged C57BL/6 mice by the i.n. route with 21 CFU of the type A F. tularensis strain KU49. The mean survival time for mice infected with the KU49 strain was 5.7 days. Bacteremia was apparent by day 2 p.i. in infected mice, and the numbers of bacteria in the lungs, livers, and spleens increased rapidly during the first 4 days of infection to >108 CFU per organ (Fig. (Fig.1).1). Both the rate of the increase of organ burden and the number of bacteria in each organ on day 4 were greater in KU49-infected mice than in LVS-infected mice when each group was challenged with 5× the 50% lethal dose of the respective pathogen (Table (Table11).
At day 3 p.i., nascent microgranulomas appeared in the livers of infected mice (Fig. (Fig.2A),2A), and extensive pyogranulomatous infiltrates were seen in both the red pulp of the spleens (Fig. (Fig.2A)2A) and the peribronchial and perivascular areas of the lungs (data not shown). These inflammatory foci were similar in appearance to the inflammatory infiltrates seen in mice infected with LVS (3, 9, 27).
By day 4 p.i., the infected tissues had undergone significant changes that distinguished KU49-infected from LVS-infected mice. Large numbers of pyknotic cells were present in the livers of KU49-infected tissues, and distinct foci of necrosis replaced most of the inflammatory infiltrates (Fig. (Fig.2A).2A). As reported previously by Conlan et al. (11), numerous enlarged hepatocytes adjacent to necrotic granulomas contained basophilic cytoplasm, suggesting a high burden of intracellular bacteria. This was confirmed by immunoperoxidase staining of sections of infected livers with antibodies to Francisella antigens (3; data not shown). Cellular pyknosis and necrosis were also observed throughout the splenic red pulp of KU49-infected spleens on day 4 p.i. In the lungs, occasional peribronchial granulomas contained numerous pyknotic cells (data not shown), but these lesions were not present in all lung sections from infected animals. For this reason, further analysis of in situ cell death focused on the liver and spleen, organs in which pathological changes were more uniformly distributed.
The TUNEL reaction was then used to identify dying cells that contained double-strand DNA breaks. On day 3 p.i., only a few TUNEL+ cells were detected in infected livers and spleens (Fig. (Fig.2A)2A) and were predominantly restricted to the inflammatory infiltrates. By the following day, nearly all of the cells in these inflammatory foci contained cleaved DNA. The periarteriolar lymphoid sheaths were often the only areas of the spleen that contained significant numbers of normal appearing cells. This pattern of cell death was seen with three different type A strains (KU49, KU54, and SCHU S4) and contrasted with the comparatively infrequent TUNEL+ cells observed throughout the entire first week in mice infected with LVS (3). Neither sublethal (100 CFU) nor supralethal (105 CFU) doses of LVS produced the highly cytotoxic response that was observed in infections with type A F. tularensis, even if the tissues were collected as late as day 6 after LVS challenge. Thus, both the pace and extent of cell death induced by the virulent type A F. tularensis strains was greater than that induced by LVS.
The inflammatory cytokine IFN-γ is required for both the formation of hepatic microgranulomas (3, 9) and the development of double-strand DNA breaks (3) in LVS-infected mouse tissues. Although KU49 induced high levels of circulating IFN-γ, the cytokine was not essential for cell death induction. IFN-γ-deficient mice showed foci of necrosis and nearly confluent areas of TUNEL in the liver, and there were no significant differences in the frequencies of granulomas or TUNEL+ cells between wild-type and IFN-γ-deficient mice by day 4 p.i. (Fig. (Fig.2B).2B). These findings suggested that the signals initiating cell death by KU49 were qualitatively different from those operating during infections with LVS and did not require the production of IFN-γ.
Monack and coworkers (18, 25) have reported that both F. novicida strain U112 and LVS activate inflammasomes in infected mouse M, resulting in the cleavage of procaspase-1, the release of IL-1β and IL-18, and the induction of pyroptotic cell death. To determine whether or not caspase-1 was activated in type A F. tularensis-infected cells, we challenged mouse bone marrow M and DC in vitro with either LVS or two different type A F. tularensis strains. Culture supernatant fluids were collected 24 h later, and the concentrations of IL-18 were measured by enzyme-linked immunosorbent assay. The release of IL-18 from cells requires caspase-1-catalyzed cleavage of the cytokine propeptide. The concentrations of two caspase-1-independent cytokines, IL-12 and IL-10, were also measured in these samples to verify Francisella-induced cell activation. Whereas LVS induced the release of IL-18 from both M and DC, neither KU49 nor KU54 stimulated IL-18 release (Fig. (Fig.3).3). The type A F. tularensis strains did induce significant quantities of IL-12 and IL-10 by DC and M, respectively.
We next examined the extent to which type A F. tularensis activated caspase-1 in infected mouse tissues. Mice were challenged i.n. with either KU49 or LVS, and the expression of activated caspase-1 or activated caspase-3 was determined by immunofluorescence microscopy. LVS induced both caspase-1 and caspase-3 activation in a portion of the cells within hepatic granulomas (Fig. (Fig.4).4). By comparison, larger numbers of cells in KU49-infected liver granulomas expressed activated caspase-3, whereas only a few cells expressed activated caspase-1 (Fig. (Fig.4A4A).
Immunoperoxidase staining of spleen sections indicated that activated caspase-3 was distributed throughout the splenic red pulp in KU49-infected mice (Fig. (Fig.4B),4B), an area in which numerous TUNEL+ cells had been detected (Fig. (Fig.2A).2A). In contrast, far fewer caspase-1-expressing cells were found in these spleen sections, were only seen on day 3 p.i., and were spatially restricted to the marginal zones between the red and white pulp. Activated caspase-9, an intermediate in the intrinsic apoptosis pathway, was observed by immunofluorescence in the areas of the splenic red pulp of KU49-infected mice that contained large numbers of cells expressing the M lineage marker F4/80. Thus, cell death in type A F. tularensis-infected tissues was observed throughout inflammatory infiltrates of the liver and spleen and was more often associated with the expression of activated caspase-3 and activated caspase-9 than with the expression of activated caspase-1.
To determine the potential role of caspase-1 in the induction of cell death, the tissues from infected wild-type and caspase-1-deficient mice were compared on day 4 p.i. For this purpose, NOD/ShiLtJ mice bearing a targeted mutation of the caspase-1 gene were compared to wild-type NOD/ShiLtJ mice. When challenged i.n. with KU49, wild-type NOD/ShiLtJ mice showed extensive cellular pyknosis, foci of necrosis, and TUNEL staining that was comparable to that seen in KU49-infected B6 mice (compare Fig. Fig.4C4C and Fig. Fig.2A).2A). Likewise, both the formation of granulomas and the frequency of TUNEL+ cells in caspase-1-deficient mice were equivalent to what was observed in the wild-type NOD/ShiLtJ mice (Fig. (Fig.4D),4D), indicating that caspase-1 was not required for granuloma formation, the induction of double-strand DNA breaks, or the development focal tissue necrosis in type A F. tularensis-infected tissues.
In contrast to the findings with caspase-1-deficient mice, the targeted deletion of the caspase-3 gene had significant effects on cell death in KU49-infected mice. Three days after infection with KU49 and prior to extensive cell death in the spleen, large numbers of cells expressing the M marker F4/80 were observed throughout the splenic red pulp (Fig. (Fig.5).5). By 4 days p.i., very few F4/80+ cells were detected in the necrotic red pulp of wild-type-infected spleens. However, the spleens of infected caspase-3-deficient mice contained significantly greater numbers of F4/80+ cells throughout the red pulp at this time point. Importantly, the expression of both TNF-α and iNOS showed a similar pattern, being sparse on day 4 in wild-type-infected tissues but much more abundant in the spleens of infected caspase-3-deficient mice. Caspase-1-deficient mice did not show an increase in the frequency of F4/80+ cells compared to wild-type controls (data not shown). These findings indicate that type A F. tularensis induces caspase-3-dependent apoptosis of F4/80+ cells and the loss of potentially important innate immune responses to the pathogen.
To determine whether or not caspase-3-dependent cell death affected the dissemination of the infection within the liver, tissue sections from KU49-infected wild-type or caspase-3-deficient mice were stained with a polyclonal antibody to Francisella antigens (Fig. (Fig.6).6). LVS-infected mice served as controls, and the frequencies of Francisella-containing cells located outside the granulomas were determined for each experimental group. The results indicate that many more Francisella-containing cells were found outside granulomas in wild-type livers than in the livers of caspase-3-deficient mice. This suggests that caspase-3-dependent cell death facilitates dissemination of the infection from cells within the granulomas to cells located throughout the liver parenchyma.
Our knowledge of the pathological changes that occur in disseminated tularemia in humans has been primarily based on the examination of biopsy or autopsy specimens (22, 30). This approach has not provided much insight into the likely pathophysiological progression of disease or the underlying mechanisms of tissue injury. Fortunately, the mouse respiratory challenge model of tularemia reproduces many of the characteristic features of disseminated disease in humans, including the recruitment of inflammatory cells and the development of foci of necrosis in the lungs, spleen, and liver (11). In the present study this model was used to determine to what extent apoptosis signaling pathways contributed to cell death, particularly among infiltrating phagocytic cells, which are thought to mediate significant innate immune responses to this pathogen.
Evidence presented here indicates that cell death induction in vivo by type A F. tularensis does not require the activation of procaspase-1. Cells expressing activated caspase-1 were found infrequently in the tissues of mice infected with the type A strain KU49, despite high levels of intracellular Francisella antigens and double-strand DNA breaks. Consistent with this finding, the caspase-1-dependent release of IL-18 from bone marrow-derived M and DC was not observed after challenge in vitro with type A F. tularensis but was seen after challenge with LVS. This difference could not be attributed to cell death induction by KU49, because DC and M did produce IL-12 and IL-10, respectively, in response to the pathogen. Likewise, similar to LVS (7) and the type A SCHU S4 strain (31), the KU49 strain was taken up by and replicated within bone marrow M (data not shown), a property that is required for inflammasome activation. Conversely, caspase-1-deficient mice showed pathological responses to type A F. tularensis challenge that were indistinguishable from those of wild-type infected animals. Collectively, these findings distinguish type A F. tularensis from several less-pathogenic Francisella subspecies for which inflammasome and caspase-1 activation appear to be characteristic features (18, 25).
The spleens of mice infected with type A F. tularensis contained cells expressing activated caspase-3 throughout the red pulp, an area that also showed large numbers of TUNEL+ cells. This pattern of cell death in caspase-3-deficient mice was substantially altered, especially among F4/80+ cells. Cells expressing F4/80 were abundant in the splenic red pulp of infected mice on day 3 p.i. but were almost completely depleted the following day. In mutant mice lacking caspase-3, significantly greater numbers of red pulp F4/80+ cells were observed, and both the spleen and the liver appeared to be less necrotic. Death among spleen cells expressing the myeloid marker Ly-6G was not decreased in caspase-3-deficient animals (data not shown), suggesting that distinct signaling pathways initiate apoptosis in Ly-6G+ cells versus F4/80+ cells. Bosio et al. (4) reported that neutrophils and DC infiltrating the lungs of mice 3 days after infection with the type A SCHU S4 strain only rarely expressed activated caspase-3 or bound annexin V. In studies not reported here, we have also examined lung tissues of KU49-infected mice for the expression of cleaved caspase-3. Inflammatory cells in perivascular and peribronchial pyogranulomas did not show significant caspase-3 activation and primarily expressed Ly-6G rather than F4/80 (26a).
Between days 3 and 4 after challenge with KU49, the expression of activated caspase-3 increased significantly in the liver and spleen and was accompanied by the sudden appearance of cellular pyknosis and double-strand DNA breaks in cells infiltrating these infection sites. During this 24-h period, the number of viable bacteria in infected tissues increased 100-fold, and the spatial distribution of the bacterial antigens changed. On day 3 p.i., bacterial antigens were restricted to the nascent granulomas in the liver but were disseminated throughout the liver parenchyma by day 4 of infection. In contrast, the livers of mice infected with LVS showed only minimum caspase-3 activation, TUNEL, necrosis or bacterial dissemination as their infections progressed. Although the basis for these differences is not currently known, it is possible that the relative rates of growth of these two pathogens (KU49 versus LVS) or the total intracellular bacterial burdens determine different cell death outcomes in infected tissues. However, either challenging mice with 200 50% lethal doses of LVS or delaying tissue collections until day 6 p.i. did not produce the pathological changes in LVS-infected mice that were seen in KU49 infections on day 4.
The identity of the various cells that die during type A F. tularensis infections has not been fully determined, since it has proven difficult to detect many cell surface markers on dying cells that express high intracellular levels of activated caspase-3. However, many of the cells that expressed activated caspase-9, an early component of the intrinsic apoptosis pathway, appeared to also express F4/80. Two important M-associated innate immune responses—TNF-α and iNOS expression—were also significantly elevated in caspase-3-deficient mice, supporting the conclusion that M constitute a significant proportion of the cells undergoing caspase-3-dependent apoptosis during type A F. tularensis infections. These findings are also consistent with the conclusion that M mediate the retention of the bacteria within hepatic granulomas in caspase-3-deficient mice.
Chen et al. (8) reported that thymic atrophy associated with the depletion of CD4+CD8+ thymocytes occurs in mice challenged with type A F. tularensis but not in mice challenged with type B strains. Circulating cortisone levels were significantly elevated at 4 days p.i. with type A F. tularensis, and thymocyte depletion was decreased in adrenalectomized mice or mice lacking receptors for tumor necrosis factor. To what extent corticosteroids or TNF-α contribute to the induction of apoptosis in type A F. tularensis-infected peripheral organs is not currently known, but splenic T cells within the periarteriolar lymphoid sheath did not show extensive apoptosis in the present study until just before death of the animals.
Several immune evasion mechanisms have been attributed to F. tularensis, including the induction of immunosuppressive cytokines (4, 5) and prostaglandin E2 (33) in the lungs of infected mice, the inhibition of respiratory bursts in human neutrophils (1, 26), and the impairment of phosphatidylinositol 3-kinase-mediated signaling in human monocytes (6). The present study indicates that the rapid induction of apoptosis in tissue M that have been recruited to sites of infection may constitute another important mechanism of immune evasion and explains many of the histopathological features of tularemia. The sudden onset of M death would be expected to facilitate the dissemination of infection to adjacent cells (e.g., hepatocytes) and dampen key innate immune responses to the pathogen.
We thank Kevin King for providing the SCHU S4 strain.
This study was supported by NIH grant R21 AI062939 and a bridging grant from the University of Kansas Medical Center Research Institute. S.M.B. was a recipient of a Biomedical Research Scholars Program fellowship.
Editor: W. A. Petri, Jr.
Published ahead of print on 24 August 2009.