ABSTRACT
Brucellosis, caused by the intracellular bacterial pathogen Brucella, is a zoonotic disease for which arthritis is the most common focal complication in humans. Here we investigated the role of inflammasomes and their effectors, including interleukin-1 (IL-1), IL-18, and pyroptosis, on inflammation and control of infection during Brucella-induced arthritis. Early in infection, both caspase-1 and caspase-11 were found to initiate joint inflammation and proinflammatory cytokine production. However, by 1 week postinfection, caspase-1 and caspase-11 also contributed to control of Brucella joint infection. Inflammasome-dependent restriction of Brucella joint burdens did not require AIM2 (absent in melanoma 2) or NLRP3 (NLR family, pyrin domain containing 3). IL-1R had a modest effect on Brucella-induced joint swelling, but mice lacking IL-1R were not impaired in their ability to control infection of the joint by Brucella. In contrast, IL-18 contributed to the initiation of joint swelling and control of joint Brucella infection. Caspase1/11-dependent cell death was observed in vivo, and in vitro studies demonstrated that both caspase-1 and caspase-11 induce pyroptosis, which limited Brucella infection in macrophages. Brucella lipopolysaccharide alone was also able to induce caspase-11-dependent pyroptosis. Collectively, these data demonstrate that inflammasomes induce inflammation in an IL-18-dependent manner and that inflammasome-dependent IL-18 and pyroptosis restrict Brucella infection.
INTRODUCTION
Brucellosis, caused by the genus Brucella, is one of the most common zoonotic infections worldwide, infecting over 500,000 individuals each year (1, 2). Transmission typically occurs through consumption of unpasteurized dairy goods or by aerosol from contaminated animal products (3, 4). Several Brucella species are capable of infecting humans, including Brucella melitensis, Brucella abortus, and Brucella suis (5). Livestock vaccination efforts have reduced disease incidence; however, no licensed human vaccine exists and several wild animal reservoirs remain (6–8). Brucellosis has been named a neglected zoonotic disease by the World Health Organization, and many developing countries have identified brucellosis as a top priority infection (8, 9).
Osteoarticular and/or musculoskeletal inflammation are the most common focal complications of brucellosis, with an incidence of 40 to 80% in infected patients (10, 11). Arthritis can manifest as peripheral arthritis, sacroiliitis, and spondylitis and is thought to arise from Brucella infection within and around the joint (10–13). Active infection of the joints is presumed to be required for arthritis development, as viable brucellae along with polymorphonuclear and mononuclear leukocytes are often found within the synovial fluid of infected patients (11, 14–16). Brucella-induced arthritis is treated with prolonged antibiotic therapy; however, time to resolution is often extensive, and relapse of disease can occur (11, 17). Untreated articular brucellosis can result in synovial rupture, bone destruction, and/or bursitis, potentially causing lifelong complications (13, 18–20).
Identification of inflammatory responses contributing to articular brucellosis have been hindered by the lack of relevant mouse models. We have previously reported that intraperitoneal (i.p.) Brucella infection in gamma interferon (IFN-γ)-deficient mice results in arthritis and musculoskeletal inflammation of ankle joints as early as 15 days postinfection. Arthritis in this systemic model was partially dependent on interleukin-1R (IL-1R) and CXCR2, while adaptive immune cells were not required for inflammation (21, 22). Recently, we demonstrated that mice inoculated with Brucella locally in footpads resulted in synchronized infection and joint inflammation in wild-type (WT) mice, allowing us to investigate mechanisms of the host response that are specific to the joint (23).
Inflammasomes are multiprotein structures that recognize threats in the host cell cytosol. Canonical inflammasomes are equipped with sensors that detect intracellular antigens and damage-associated molecular patterns (DAMPs) (24). This initiates a cascade of protein oligomerization and recruitment, resulting in inflammasome assembly. Upon activation, canonical inflammasomes recruit pro-caspase-1, a cysteine protease, and cleave it into its functional form, caspase-1 (24, 25). Caspase-1 can then activate proinflammatory cytokines and trigger a programmed inflammatory cell death known as pyroptosis (24, 26). Due to the inflammatory nature of inflammasomes, they contribute to many models of inflammation, including pathogen-induced and sterile arthritis (27, 28).
Brucella infection in macrophages activates the canonical inflammasome cytosolic sensors: NLR family, pyrin domain containing 3 (NLRP3) and absent in melanoma 2 (AIM2). Studies have shown that the Brucella type IV secretion system and mitochondrial damage can stimulate NLRP3, while Brucella DNA activates the AIM2 inflammasome pathway (29–31). Following stimulation of the NLRP3 or AIM2 sensors and subsequent cleavage of caspase-1, active caspase-1 proteolytically cleaves IL-1β and IL-18 into their functional/secreted forms. IL-1R and IL-18 are known to restrict Brucella burdens during chronic Brucella infection but may also lead to immunopathology, as they play a proinflammatory role in rheumatoid arthritis (29, 31–33). While NLRP3, AIM2, caspase-1, and caspase-11 are protective against chronic systemic Brucella infection in vivo (29), there are multiple studies indicating that NLRP3 and/or AIM2 contributes to pathology in autoreactive arthritis models (28, 34, 35). Thus, while inflammasomes contribute to the restriction of systemic Brucella infection, they might also induce deleterious joint inflammation conferred by Brucella.
Unlike canonical inflammasomes, the noncanonical inflammasome, caspase-11, does not use additional cytosolic protein sensors to identify intracellular threats; rather, caspase-11 directly recognizes, and is activated by, cytosolic bacterial lipopolysaccharide (LPS). Like caspase-1, caspase-11 can also induce pyroptosis, which can aid in the clearance of Gram-negative intracellular pathogens (26). There is limited knowledge on Brucella's ability to activate caspase-11. In fact, one study showed that the Brucella effector protein TcpB can induce degradation of caspase-11 (36). However, to our knowledge, the role of caspase-11 in response to Brucella in vivo has not been directly examined. Here we describe a role for both caspase-1 and caspase-11 in Brucella-induced arthritis and define the inflammasome effectors responsible for inflammation and control of infection.
RESULTS
Inflammasomes have IL-1-independent effects on inflammation and control of infection during Brucella-induced arthritis.Following intraperitoneal Brucella infection, IL-1R contributes to the development of Brucella-induced arthritis in IFN-γ-deficient mice (22). Thus, to examine whether inflammasome-dependent IL-1 contributes to joint inflammation during local B. melitensis infection, we infected caspase-1/11−/− and IL-1R−/− mice in the rear footpads (Fig. 1). Beyond day 6 postinfection, caspase-1/11−/− mouse ankles displayed variable but slightly increased swelling compared to WT animals (Fig. 1A), a trend which was also evident when a higher dose of Brucella was used for infection (data not shown). This long-term study also revealed a minor role of caspase-1/11 in the clearance of Brucella from the spleen (Fig. 1B). Caspase-1/11−/− mice exhibited delayed swelling, with significantly reduced ankle diameters at days 2 and 3 compared to those of WT mice (Fig. 1C). Many inflammasome-dependent inflammatory cytokines peak at day 2 postinfection, and joint Brucella loads are also comparable between WT, caspase-1/11−/−, and caspase-11−/− mice at day 2 postinfection. Therefore, joint inflammation was assayed at day 2 postinfection, as it allows us to study the initiation of inflammation and immunity when Brucella loads are similar between mouse strains. Consistent with our swelling data, hematoxylin and eosin (H&E) staining revealed decreased inflammation in caspase-1/11−/− joints at day 2 postinfection (Fig. 1D and E). Additionally, caspase-1/11−/− mice exhibited lower levels of inflammatory cytokines/chemokines, such as IL-6, IFN-γ, CCL2, tumor necrosis factor alpha (TNF-α), CXCL1, and CXCL9, in their joints than WT mice (Fig. 2A to F) at day 2 postinfection. In contrast to the phenotype observed at days 2 and 3, caspase-1/11−/− mice had increased swelling by day 4 compared to WT animals (Fig. 1C). To explain this escalation of swelling in caspase-1/11−/− mice, we hypothesized that inflammasomes may be important for mediating bacterial clearance in the joints. When joint bacterial loads were examined, caspase-1/11−/− mice had numbers of CFU that were similar to those of WT mice at days 1 and 2 postinfection; however, by day 7, caspase-1/11−/− animals had significantly increased Brucella loads (Fig. 1F). To examine whether IL-1 was inducing this arthritis and musculoskeletal inflammation, swelling of WT and IL-1R−/− mouse ankles was recorded over 7 days post-Brucella infection. Interestingly, IL-1R−/− mice displayed only a modest reduction in swelling at day 2, and Brucella loads at day 7 were similar among groups (Fig. 1G and H). This disparity in swelling kinetics and joint Brucella burdens between IL-1R−/− and caspase-1/11−/− mice suggests that inflammasomes have IL-1R-independent effectors that contribute to Brucella-induced arthritis and control of infection.
Caspase-1/11 induces inflammation and controls joint Brucella burden. (A to H) All mice were infected in both rear footpads with 1 × 105 B. melitensis 16M bacteria. (A) WT and caspase-1/11−/− mouse joint swelling was recorded over time. (B) Mice were euthanized on day 34, and ankle joint and spleen bacterial loads were enumerated (n = 5 mice/group). Data are representative of one experiment. (C) Joint swelling was recorded over 7 days (4 to 5 mice/group). Data are representative of >4 experiments. (D and E) Mice were euthanized on day 7 postinfection, and H&E staining was conducted on the rear ankle joint sections and scored for inflammation severity (6 to 8 mice/group). (D) Images (×40) are depicted with amplified boxed regions (×200) displayed to the right of each image. (F) WT and caspase-1/11−/− mice were euthanized at days 1, 2, and 7 postinfection, and joint bacterial burdens were determined (3 to 5 mice/group); day 1 data are representative of one experiment, day 2 data are representative of >3 experiments, and day 7 data are representative of >3 experiments. (G) Joint swelling was measured in infected WT and IL-1R−/− mice over time. (H) Mice were euthanized and joint bacterial loads were determined at day 7. Data are representative of two independent experiments (3 to 5 mice/group). *, P < 0.05 (compared to results for WT mice). Error bars represent standard deviations (SD) of the mean.
Inflammasomes mediate proinflammatory cytokine production in Brucella-infected joints. (A to F) WT and caspase-1/11−/− mice were infected in each rear footpad with 1 × 105 B. melitensis 16M bacteria. Mice were euthanized at day 2 postinfection, and levels of joint cytokines were determined. Data are representative of two independent experiments (n = 3 mice/group). *, P < 0.05 (compared to results for WT mice). Error bars represent SD of the mean.
Caspase1/11-dependent IL-18 mediates joint swelling and Brucella clearance from the joint.In addition to IL-1β, inflammasomes also activate IL-18. To investigate if IL-18 is an inflammasome-dependent effector of Brucella-induced arthritis, IL-18 was neutralized in WT and caspase-1/11−/− mice and the impact on arthritis was assessed. Joint swelling 2 days postinfection was reduced in anti-IL-18-treated mice compared to WT animals, but IL-18 neutralization did not have as great an effect as caspase-1/11 deficiency. Additionally, the effects of IL-18 relied on caspase-1/11, as swelling in caspase-1/11−/− mice was similar to that in caspase-1/11−/−, anti-IL-18-treated animals (Fig. 3A). Bacterial loads at day 2 postinfection were similar between WT and caspase-1/11−/− mouse joints either with sufficient IL-18 or with IL-18 neutralized (Fig. 3B). To examine if IL-18 contributes to swelling or joint bacterial restriction beyond the initial initiation of inflammation, swelling in IL-18−/− mice was monitored over the course of 7 days. IL-18−/−/IL-1R−/− mice were also included in our study to further examine the role of IL-1R in our model. IL-18−/− mice had reduced joint swelling only at day 2 postinfection, while IL-18−/−/IL-1R−/− mice had significantly reduced ankle widths at days 2 and 3 compared to that of WT mice (Fig. 3C). By day 7 postinfection, IL-18 deficiency resulted in increased joint Brucella burdens; however, IL-1R deficiency in IL-18−/− mice had no additive effect on control of joint infection (Fig. 3D). These data indicate that IL-18 and IL-1R are involved in the early induction of swelling during Brucella-induced arthritis but that IL-18 alone is required for restriction of Brucella infection in the joint. Relative to WT mice, however, caspase-1/11 deficiency resulted in increased bacterial burdens and had a greater effect on joint swelling than IL-18/IL-1R deficiency. This suggests that IL-18 and IL-1 are not the sole effectors of caspase-1/11.
IL-18 contributes to inflammasome-dependent joint inflammation and Brucella clearance. (A and B) WT or caspase-1/11−/− mice were treated with an IgG isotype control or anti-IL-18 and then infected in each rear footpad with 1 × 105 B. melitensis 16M bacteria. At day 2 postinfection, ankle swelling was recorded (A) and joint CFU were enumerated (B). Data are combined from two separate experiments with 8 to 10 mice/group in total. (C) WT, IL-18−/−, and IL-1R/IL-18−/− mice were infected in the rear footpad with 1 × 105 B. melitensis 16M bacteria, and swelling was recorded over time. (D) Mice were euthanized on day 7 postinfection, and joint bacterial burdens were determined. Data are representative of two independent experiments (4 to 5 mice/group). Means with the same letter are not statistically different from each other, as determined by ANOVA, followed by Tukey's test. Error bars represent SD of the mean.
Prostaglandins, thromboxanes, and perforin-mediated cytotoxicity are dispensable for the control of Brucella infection in the joint.Although inflammasome activation of IL-1β and IL-18 are the best characterized modes of caspase-1/11-dependent antimicrobial activity, inflammasomes can protect against infection via other means, including induction of eicosanoid storms (37). Thus, we treated Brucella-infected mice with the cyclooxygenase inhibitor indomethacin to inhibit production of prostaglandins and thromboxanes. Indomethacin treatment reduced joint swelling at day 2 postinfection, but no significant difference in joint Brucella loads was observed at day 7 postinfection relative to WT mice (see Fig. S1A and B in the supplemental material). In fact, there was a trend of reduced Brucella burdens in indomethacin-treated mice, indicating that prostaglandins/thromboxanes are not bactericidal but actually may aid in Brucella colonization of the joint (Fig. S1B). Additionally, IL-18 has recently been shown to activate perforin-mediated cytotoxicity as a mode of intracellular bacterial clearance (38). Therefore, the role of cytotoxicity was evaluated using perforin−/− (perf−/−) mice, but no notable changes in joint Brucella loads was observed (Fig. S1C).
Inflammasomes do not restrain Brucella infection in macrophages due to their ability to restrict host l-lactate production.Previous studies have demonstrated that B. abortus thrives in glycolytic cells via its ability to utilize host lactate via the Brucella l-lactate dehydrogenase (LDH) gene (lldD) (39, 40). However, inflammasome activation resulting in pyroptosis releases host cell lactate dehydrogenase, which can degrade l-lactate (41). Additionally, other studies suggest that active caspase-1 can directly inhibit glycolysis and the subsequent production of l-lactate via cleavage of glycolytic enzymes (42). To determine if caspase-1/11 was protective against Brucella via its ability to restrict l-lactate, WT and caspase-1/11−/− mouse bone marrow-derived macrophages (BMDMs) were infected with B. melitensis, and l-lactate in cell supernatant was measured. l-Lactate production was induced during B. melitensis infection in both WT and caspase-1/11−/− macrophages; however, l-lactate levels in wells containing caspase-1/11−/− cells were elevated compared to those containing WT cells (Fig. 4A). While WT and caspase-1/11−/− cells had similar rates of Brucella phagocytosis (data not shown), caspase-1/11−/− BMDMs had an increased Brucella burden at 48 h postinfection (Fig. 4B). In order to determine if glycolysis and the subsequent production of l-lactate aided in Brucella survival, we made a B. melitensis l-lactate dehydrogenase mutant (ΔlldD) to inhibit Brucella's ability to utilize free l-lactate (39). Compared to WT B. melitensis, the ΔlldD mutant was attenuated in WT macrophages (Fig. 4C). Therefore, we investigated whether lldD conferred a greater competitive advantage in caspase-1/11−/− cells than in WT macrophages due to enhanced lactate production. Thus, we performed competition experiments between the ΔlldD mutant and a complemented ΔlldD strain in both WT and caspase1/11−/− cells. We found that our complemented strain outcompeted the mutant in both WT and caspase-1/11−/− macrophages; however, the competitive index of the lldD-complemented strain relative to the ΔlldD mutant was not significantly different between macrophage strains. These data indicate that caspase-1 and caspase-11 do not restrict Brucella survival via their ability to reduce l-lactate production (Fig. 4D).
Increased lactate production by caspase-1/11−/− macrophages does not promote Brucella infection. (A to D) Mouse strain BMDMs were infected for 6 h with B. melitensis 16M at an MOI of 100 as described in Materials and Methods. (A) At 48 h postinfection, l-lactate was measured in cell supernatant. Means with the same letter are not statistically different from each other, as determined by ANOVA, followed by Tukey's test. (B) At 48 h postinfection, cells were lysed and Brucella loads were determined. Data are representative of two independent experiments, with 6 to 8 wells/group. *, P < 0.05 (compared to results for infected WT BMDMs). (C) WT BMDMs were infected with either B. melitensis or B. melitensis ΔlldD. At 48 h postinfection, intracellular Brucella burdens were determined. Data are representative of one of two independent experiments (6 wells/group). *, P < 0.05 (compared to results for WT B. melitensis). (D) BMDMs were coinfected at a 1:1 ratio with the ΔlldD mutant and an ΔlldD complemented strain of B. melitensis. At 48 h postinfection, cells were lysed and the competitive index was determined for each BMDM strain. Data are representative of two independent experiments, with 5 wells/group. Error bars represent SD of the mean.
Caspase-11 contributes to Brucella-induced arthritis and restricts Brucella infection in the joint.Many studies utilize caspase-1/11−/− mice to study the role of caspase-1, despite the fact that these animals also lack functional caspase-11. To determine if the noncanonical inflammasome, caspase-11, contributes to the initiation of inflammation and/or the clearance of Brucella from the joint, WT and caspase-11−/− mice were infected in the rear footpads with B. melitensis. Indeed, caspase-11−/− mice displayed reduced swelling at day 2 postinfection, while day 2 joint bacterial loads remained similar to those of WT mice (Fig. 5A and data not shown). The reduction of joint swelling in caspase-11−/− mice relative to WT animals was not to the same magnitude as that seen in caspase-1/11−/− mice, suggesting a partial role of caspase-11 in inflammasome-dependent induction of swelling. Histopathology analysis at day 2 postinfection confirmed caspase-11's inflammatory role, as caspase-11−/− mouse joints had reduced inflammation compared to that of WT mice (Fig. 5B). Additionally, IL-6 and TNF-α levels were significantly decreased in caspase-11−/− joints at day 2 postinfection (Fig. 6A to F). Caspase-11−/− mice also displayed increased joint swelling by day 4 postinfection (Fig. 5A). To examine the effects of caspase-11 on bacterial clearance, we measured joint bacterial burdens on day 7. Joint CFU numbers were slightly, but significantly, elevated in caspase-11−/− mouse joints compared to WT mice (Fig. 5C). As with our swelling data, the difference in CFU numbers between caspase-11−/− and WT mice was not to the same magnitude as the difference between caspase-1/11−/− and WT mice. To assay what canonical inflammasome could be involved in caspase-1-dependent control of Brucella infection, NLRP3−/− and AIM2−/− mice were infected and evaluated for Brucella burdens at day 7 postinfection. Interestingly, neither NLRP3−/− nor AIM2−/− mice displayed impaired ability to control joint Brucella burdens (Fig. 5D and E).
Caspase-11 is inflammatory and limits Brucella infection in the joints. (A) All mice were infected in each rear footpad with 1 × 105 B. melitensis 16M bacteria. Joint swelling was measured over time in WT and caspase-11−/− mice. Data are representative of three independent experiments (4 to 5 mice/group). (B) Mice were euthanized at day 2 postinfection, H&E staining was conducted on mouse ankle sections, and the severity of pathology was scored. Histology scores are from one of two independent experiments (4 to 5 mice/group) from which sections were obtained. (C) Mice were euthanized on day 7 postinfection, and joint Brucella loads were enumerated. Data are representative of two independent experiments (4 to 5 mice/group). (D and E) NLRP3−/− (D) and AIM2−/− (E) mice were evaluated for joint Brucella burdens at day 7 postinfection. Data are representative of two independent experiments (4 to 5 mice group). *, P < 0.05 (compared to results for WT mice). Error bars represent SD of the mean.
Caspase-11 contributes to inflammasome-dependent proinflammatory cytokine production in Brucella-infected joints. (A to F) WT and caspase-11−/− mice were infected in each rear footpad with 1 × 105 B. melitensis 16M bacteria. Mice were euthanized at day 2 postinfection, and levels of joint cytokines were determined. Data are representative of one experiment (n = 3 mice/group). *, P < 0.05 (compared to results for WT mice). Error bars represent SD of the mean.
Caspase-1 and caspase-11 induce cell death in response to Brucella and restrict infection in macrophages.Pyroptosis and subsequent efferocytosis of dead host cells containing bacteria can be an important antimicrobial mechanism of inflammasomes (26). To demonstrate if pyroptosis occurs during Brucella infection, BMDMs were infected with B. melitensis and evaluated for cell death at 24 and 48 h postinfection. Brucella induced robust cell death in WT macrophages, while cell death was reduced in caspase-11−/− macrophages and abolished in caspase-1/11−/− cells at both 24 and 48 h postinfection, indicating that both caspase-1 and caspase-11 contribute to Brucella-induced macrophage death. Interestingly NLRP3−/− and AIM2−/− cells had similar levels of cell death as WT macrophages, suggesting that NLRP3 or AIM2 alone is not required to induce pyroptosis (Fig. 7A and B). Consistent with the differential levels of macrophage cell death, more viable bacteria were recovered in wells with caspase-1/11−/− and caspase-11−/− BMDMs than in wells with WT BMDMs at 24 and 48 h postinfection (Fig. 7C and D). Similarly, joint cells from footpad-infected caspase-1/11−/− mice displayed reduced terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL), which identifies DNA fragmentation during late-stage apoptosis or pyroptosis, compared to WT mice (Fig. 8A and B) (43). To determine if caspase-11 could be directly activated by Brucella LPS, we treated BMDMs with purified Brucella abortus S19 LPS, which has a structure similar to that of B. melitensis LPS (44). We found that Brucella LPS was able to induce caspase-11-dependent cell death, and this phenotype was enhanced upon the addition of the transfection reagent FuGENE, confirming that cytosolic Brucella LPS is capable of inducing caspase-11-dependent cell death (Fig. 8C). These data suggest that caspase-11- and caspase-1-induced pyroptosis is protective against Brucella infection.
Brucella induces caspase-1- and caspase-11-dependent pyroptosis. (A to D) WT, caspase-1/11−/−, caspase-11−/−, NLRP3−/−, and AIM2−/− BMDMs were infected with B. melitensis 16M at an MOI of 100. (A and B) Supernatants were harvested at 24 and 48 h, and percent cell death was determined by LDH release. (C and D) The remaining cells were lysed, and Brucella loads were determined. Data are representative of two independent experiments (4 wells/group). Means with the same letter are not statistically different from each other, as determined by ANOVA, followed by Tukey's test. Error bars represent SD of the mean.
Brucella infection induces caspase1/11-dependent cell death in vivo, and transfection of Brucella LPS induces caspase-11-dependent cell death in vitro. WT and caspase-1/11−/− mice were infected in each rear footpad with 1 × 105 B. melitensis 16M bacteria. Mice were euthanized at day 2 postinfection, and joints were evaluated for TUNEL staining via flow cytometry. (A and B) Depicted are a representative histogram of the proportion of TUNEL-positive cells in WT versus caspase-1/11−/− mice (A) and the percentage of TUNEL-positive cells (n = 3 to 5 mice/group) in each mouse strain (B). *, P < 0.05 (compared to results for WT mice). Data are representative of one experiment. (C) WT or caspase-11−/− macrophages were treated with Brucella LPS alone or with Brucella LPS packaged in FuGENE. After 9 h of culture, supernatant was harvested and assessed for cell death. Data are representative of one experiment (4 wells/group). Means with the same letter are not statistically different from each other, as determined by ANOVA, followed by Tukey's test. Error bars represent SD of the mean.
DISCUSSION
Brucellosis is a globally prevalent zoonotic disease, and Brucella-induced arthritis is the most common focal complication of human brucellosis, occurring in 40 to 80% of patients (10, 11). Here we investigated the role of inflammasomes in a murine model of Brucella-induced arthritis. We show that inflammasomes initiate inflammation and arthritis in mouse ankle joints. Changes in articular inflammatory responses due to inflammasomes at day 2 postinfection were not dependent on joint Brucella burden, and thus day 2 was used to determine differences in initiation of joint inflammation.
Caspase-1 and caspase-11 both induced a joint inflammatory response at day 2 after Brucella footpad infection. Inflammasomes play a major role in inducing pathology in rheumatoid arthritis due to their ability to process IL-1β and IL-18 into their active forms (28, 33). Indeed, caspase-1/11-dependent inflammation was partially reliant on IL-18, while only modest inflammatory effects of IL-1R were observed. At day 7 postinfection, restriction of Brucella burdens in the joint was partially mediated by IL-18. IL-18 can induce the production of IFN-γ, which is a well-known mediator of protection against Brucella infection and chronic Brucella-induced arthritis (21, 22, 45, 46). Therefore, inflammasome-dependent IL-18 induction of IFN-γ could be important for control of joint infection by Brucella. Control of Brucella joint infection did not require IL-1R signaling in the presence or absence of IL-18. Additionally, IL-1R was not responsible for protection during chronic joint infection (data not shown). IL-1β and IL-1α can have opposing effects, and both signal through IL-1R (47); therefore, future studies will clarify the role of IL-1β and IL-1α in Brucella-induced arthritis and control of infection.
Although inflammasomes did not influence Brucella burdens early in infection, we did uncover an antimicrobial role for caspase-1 and caspase-11 beyond day 6 postinfection. Caspase-1/11, ASC, AIM2, and NLRP3 have all been shown to be protective against systemic Brucella abortus colonization 4 weeks following i.p. inoculation (29), but nothing is known about caspase-11's antimicrobial role. Here we demonstrated that caspase-1/11 reduced joint Brucella loads by roughly 1 log, and approximately one-third of this phenotype was attributed to caspase-11. Caspase-11 traditionally recognizes enterobacterial LPS with hexa- or penta-acylated lipid A (48, 49). Brucella's LPS structure is known to be highly immunoevasive in part due to its hepta-acylated lipid A, its core oligosaccharide, and its O-polysaccharide (44). Here we demonstrated that caspase-11 contributes to joint inflammation, and both live Brucella and Brucella LPS induce caspase-11-dependent cell death. Similarly, Acinetobacter baumannii, another Gram-negative bacterium with hepta-acylated lipid A, has also been shown to activate caspase-11, suggesting that caspase-11's repertoire of lipid A recognition may be more diverse than once thought (50, 51). It should be noted that the Brucella LPS concentrations employed here are much higher than those used by others who have tested caspase-11 activation in response to enterobacterial LPS (48), indicating that Brucella LPS may not be recognized by caspase-11 as efficiently as canonical LPS.
The bactericidal activity of caspase-1 and caspase-11 may be due in part to their ability to induce pyroptosis. This inflammatory form of cell death can damage intracellular bacteria and render them confined to pore-induced intracellular traps, thereby allowing bacteria to be easily killed via efferocytosis (26). In Brucella-infected macrophages, we observed caspase-1- and caspase-11-dependent cell death, suggesting that infected macrophages die via pyroptosis. Wells with reduced cell death contained increased Brucella burdens per well, indicating that pyroptosis may protect the host by eliminating macrophages as a reservoir of Brucella infection. This was surprising, since canonically, Brucella is thought to inhibit the death of host cells (52). It is known that caspase-11 strictly requires priming prior to activation (53). Therefore, the high multiplicity of infection (MOI of 100) and extended infection time (6 h) employed in our study may have efficiently primed caspase-11 and led to pyroptosis. In contrast, the lower MOIs and shorter infection times often employed by others (54, 55) may not lead to priming of caspase-11, which could explain why robust cell death was not observed. Additionally, we demonstrated that transfection of macrophages with B. abortus S19 LPS results in induction of caspase-11-dependent cell death. To our knowledge, this is the first report of caspase-11 recognizing Brucella LPS, although one study did demonstrate that chemical inhibition of caspase-4 and caspase-5 (human orthologues of murine caspase-11) reduces Brucella LPS-induced upregulation of annexin V on human neutrophils (56). Although considerable studies identify murine caspase-11 to be activated similarly to caspase-4 and caspase-5, a recent study has demonstrated that human caspase-4 has a broader spectrum of activation, as it is activated by tetra-acylated LPS (57). Additionally, caspase-4 is constitutively expressed in human phagocytic cells, while like caspase-11, caspase-5 must be primed by other signals (53, 58). Future studies in our laboratory will investigate whether Brucella and Brucella LPS induce caspase-4/5-dependent pyroptosis of human macrophages.
DAMPs generated by caspase-11-dependent pyroptosis can activate NLRP3 and, in turn, caspase-1 (59). NLRP3 did not influence joint Brucella burdens, but caspase-11 was modestly protective against Brucella colonization. Similarly, in our in vitro studies, caspase-11 was required for inducing the death of macrophages, while NLRP3 was dispensable, identifying that caspase-11 has effects independent of the NLRP3–caspase-1 axis. Taken together, these results suggest that caspase-11 induces cell death, leading to the restriction of Brucella infection.
Elevated lactate levels are observed in multiple forms of arthritis, including septic arthritis and rheumatoid arthritis (60). Patients with Brucella-induced arthritis display synovial levels of lactate similar to those of patients with rheumatoid or crystal-induced arthritis but have higher lactate levels than osteoarthritis patients (61). Active caspase-1 can directly inhibit glycolysis and, in turn, l-lactate by cleaving important glycolytic enzymes, such as α-enolase, aldolase, and GAPDH (glyceraldehyde-3-phosphate dehydrogenase) (42). To address whether inflammasomes restricted Brucella survival by reducing l-lactate, we made a Brucella mutant defective in l-lactate dehydrogenase. Similar to another study using B. abortus (39), we found that Brucella l-lactate dehydrogenase was important for intracellular B. melitensis survival. However, inflammasomes do not appear to aid in B. melitensis restriction through their ability to limit l-lactate, as the ΔlldD complemented strain did not outcompete the ΔlldD mutant to a greater degree in caspase-1/11-deficient cells than in WT BMDMs.
The ASC-dependent inflammasomes, AIM2 and NLRP3, are known to recognize Brucella and restrict B. abortus splenic burdens during chronic infection (29, 31). Surprisingly, AIM2 and NLRP3 were dispensable for control of joint infection. NLRP3 or AIM2 can compensate for one another, as has been described in pulmonary aspergillosis, which could explain why we did not observe altered Brucella joint burdens in mice lacking only NLRP3 or AIM2 (62). Alternatively, caspase-1 may have an AIM2/NLRP3-independent activator specific to control of joint tissue.
Collectively, we have demonstrated that caspase-1 and caspase-11 contribute to pyroptosis, inflammation, and control of infection during articular brucellosis. Therefore, a proper level of inflammasome activation which controls infection while minimizing deleterious inflammation is likely critical to individuals with brucellosis.
MATERIALS AND METHODS
Bacterial strains and growth conditions.All experiments with live Brucella melitensis were performed in biosafety level 3 (BSL-3) facilities. B. melitensis 16M, obtained from Montana State University (Bozeman, Montana), was grown on Brucella agar (Ba) at 37°C (Becton Dickinson). Colonies were picked from Brucella agar plates, and strains were cultured in Brucella broth (Becton Dickinson) overnight at 37°C with shaking. The overnight Brucella concentration was estimated by measuring the optical density at 600 nm, and the inoculum was diluted to the appropriate concentration in sterile phosphate-buffered saline (sPBS). The actual viable titer was confirmed by serial dilution of the B. melitensis inoculum onto Brucella agar plates.
Deletion and complementation of the B. melitensis l-lactate dehydrogenase gene (lldD).The lldD gene in B. melitensis 16M (BMEII0377) was mutated using a nonpolar, unmarked gene excision strategy. Briefly, a 1,250-bp gBlock fragment containing, from 5′ to 3′, ∼600 bp upstream of lldD and then the first and last three codons of lldD, followed by ∼600 bp downstream of lldD (see Table S1 in the supplemental material), was synthesized by Integrated DNA Technologies. In addition, the 5′ end of the gBlock fragment contained 30 bp homologous to 30 bp upstream of the BamHI site in pNTPS139, and the 3′ end of the gBlock contained 30 bp homologous to the 30 bp downstream of the SalI site of pNTPS139. The suicide plasmid pNTPS139 (63) was digested with BamHI and SalI. The gBlock fragment and digested pNPTS139 were then ligated using the NEB Hi-Fi DNA assembly kit according to the manufacturer's instructions (New England BioLabs). This plasmid was then introduced into B. melitensis 16M, and merodiploid transformants were obtained by selection on Brucella agar plus 25 μg/ml kanamycin. A single kanamycin-resistant clone was grown overnight in Brucella broth and then plated onto Brucella agar containing 6% sucrose. Genomic DNA from sucrose-resistant, kanamycin-sensitive colonies was isolated and screened by PCR using primers (lldD screen forward, lldD screen reverse) shown in Table S1 in the supplemental material for loss of the lldD gene. Genetic complementation of the lldD mutation was performed by expressing wild-type lldD from its native promoter in pBBR1MCS-2 (64). The lldD gene, along with the native promoter, was amplified from B. melitensis 16M genomic DNA by PCR using primers (lldD clone forward and lldD clone reverse) shown in Table S1 in the supplemental material. These primers also contained overlapping sequences to the SacI and XhoI sites of pBBR1MCS-2. The resulting PCR fragment was then ligated with SacI/XhoI-digested pBBR1MCS-2 via the NEB Hi-Fi assembly kit. This construct was sequenced and then introduced into B. melitensis ΔlldD by conjugation.
Mice.Both male and female mice were used. Experiments were conducted using 6- to 12-week-old age- and sex-matched mice on a C57BL/6 background. Breeding pairs of caspase-1/11−/−, caspase-11−/−, IL-1R−/−, IL-18−/−, perf−/−, NLRP3−/−, and AIM2−/− mice were obtained from the Jackson Laboratory (Bar Harbor, ME), and breeding pairs of IL-1R−/−/IL-18−/− mice were gifted by Edward Miao (University of North Carolina, Chapel Hill, NC). Mice were infected in both rear footpads with a volume of 50 μl containing 1 × 105 Brucella bacteria (23). After infection, all mice were maintained in individually ventilated cages under high-efficiency particulate air-filtered barrier conditions of 12 h of light and 12 h of darkness in animal BSL3 facilities and were provided with sterile food and water. All studies were conducted in accordance with University of Missouri Animal Care and Use Committee guidelines.
Antibody neutralizations/treatments.In order to neutralize IL-18, mice were administered 2.5 mg of anti-IL-18 antibody (clone YIGIF74-IG7; BioXCell) into the peritoneum 1 day prior to infection. Control mice received rat IgG (Southern Biotech) as an isotype control. For eicosanoid experiments, mice received intraperitoneal delivery of 2 mg/kg indomethacin (Sigma) on day −1 and daily thereafter.
Bacterial burdens.At various times after infection, mice were euthanized, and spleens and mouse paws (following skin removal) were harvested. Spleens and paws were homogenized mechanically in sPBS (23). A series of 10-fold serial dilutions was performed in triplicate in sPBS, and the dilutions were plated onto Brucella agar. Plates were incubated 3 to 4 days at 37°C, colonies were enumerated, and the number of CFU/tissue was calculated.
Macrophage generation and infections.Bone marrow-derived macrophages (BMDMs) were derived by flushing the bone marrow from the femurs and tibias of C57BL/6 mice with sPBS. Isolated bone marrow was then cultured in complete medium (CM; RPMI 1640, 10% fetal bovine serum [FBS], 10 mM HEPES buffer, 10 mM nonessential amino acids, 10 mM sodium pyruvate) containing 30 ng/ml recombinant murine macrophage colony-stimulating factor (M-CSF) (Shenandoah). After 3 to 4 days of culture, fresh CM containing 30 ng/ml M-CSF was added to culture flasks. Adherent cells were then collected at 6 to 7 days post-bone marrow harvest by adding 0.05% trypsin (Life Technologies). Cells were plated at 1 × 106 cells/ml in fresh CM and left to adhere overnight. Cells were infected at a multiplicity of infection (MOI) of 100 B. melitensis 16M bacteria. Cells were infected for 6 h and washed with sPBS, CM containing 50 μg/ml gentamicin was added, and the cells were incubated for 30 min. Cells were then washed with sPBS and left to incubate in CM containing 2.5 μg/ml of gentamicin for the remainder of the experiment. Ten percent FBS was used for all studies unless cell death was measured (Fig. 7), in which case 2% FBS was used. Supernatants were harvested, and intracellular Brucella bacteria in lysates were enumerated at 24 or 48 h postinfection. Supernatants were used for l-lactate quantification as described below. l-Lactate dehydrogenase (LDH) release was measured utilizing the Cytotoxicity Detection Kit Plus (LDH) by adhering to the vendor's protocols (Roche). To determine bacterial burdens, BMDMs were washed with sPBS and lysed in double-distilled water (ddH2O), and the CFU were enumerated by serial dilutions onto Brucella agar (Ba). For competitive index experiments, both the ΔlldD mutant and the lldD complemented strain (containing kanamycin resistance [KanR]) were combined at a 1:1 ratio and titers were determined at the time of infection by plating inoculum on regular Ba and on Ba containing 25 μg/ml kanamycin. Phagocytosis was determined by plating the cell lysate collected at 0 h postinfection on regular Ba and on Ba with kanamycin. Cell lysates were also collected at 48 h postinfection and plated onto regular Ba and on Ba with kanamycin and then the CFU were enumerated. The final competitive index was calculated by determining the ratio of the two strains at 48 h postinfection with respect to their phagocytosis.
Lactic acid quantification.In order to remove cell debris, BMDM supernatants were centrifuged at 13,000 × g for 10 min. Supernatants were then deproteinated using a Vivaspin 500 10,000-molecular-weight-cutoff (MWCO) filter (Fisher) according to the manufacturer's instructions. l-Lactate was measured using an l-lactate kit (Eton Bio) according to the manufacturer's instructions.
LPS purification and transfection.LPS was purified from the B. abortus S19 vaccine strain (obtained from the University of Wyoming) by using hot-phenol extraction (65) Briefly, S19 was suspended in deionized water at 66°C and then combined with 90% phenol (Sigma), preheated to 66°C. The suspension was stirred for 20 min, followed by cooling in an ice bath before centrifugation. The phenol layer was aspirated using a 16-gauge cannula, and cellular debris was removed via filtration using a Whatman no. 1 filter. The LPS was then precipitated via the addition of methanol (Sigma) containing 1% methanol saturated with sodium acetate (Sigma) and allowed to incubate for 2 h at 4°C before centrifugation. The precipitate was stirred with deionized water for 18 h at 4°C. Following incubation, the LPS was centrifuged and the supernatant was reserved and maintained at 4°C, while the precipitate was resuspended in deionized water and stirred for 2 h at 4°C. The supernatant solution was recovered as described above and pooled with the reserved supernatant sample. Trichloroacetic acid (Sigma) was added to the supernatant containing the crude LPS and stirred for 10 min, and the precipitate was removed via centrifugation. The supernatant solutions were dialyzed against deionized water and then freeze-dried.
BMDMs were cultured at 1 × 106 cells/ml and primed overnight with 1 μg/ml poly(I·C) (InvivoGen). Cells were washed and then cultured in Opti-MEM reduced serum medium (Thermo-Fisher). Purified Brucella LPS was packaged in FuGENE 6 (Promega) at room temperature for 15 min according to the manufacturer's instructions. Wells receiving 10 μg/ml LPS received a final concentration of 0.25% (vol/vol) FuGENE 6, while wells receiving 100 μg/ml LPS received a final concentration of 2.5% FuGENE. Cells were incubated for 9 h, at which time LDH release was measured as described above.
Assessment of pathology.Ankle swelling was evaluated in relation to basal joint measurements made prior to infection, as described previously (23). Ankle swelling was measured at various time points by collective measurements of both tibiotarsal joints. The difference of the recorded measurement from the basal measurement was reported as mean joint swelling. For histology of mouse ankles, skin was removed, and ankles were fixed in 10% buffered zinc formalin, decalcified in 15% formic acid, rinsed in tap water, dehydrated with an alcohol gradient and clearing agent, and then embedded in paraffin. Tissue sections (5 μm) were mounted on glass slides, and serial sections were stained with hematoxylin and eosin (H&E) and covered with a coverslip with aqueous mounting medium. Tissues were evaluated for inflammation (neutrophils, macrophages, lymphocytes), which involved multiple tissues (synovial membrane, bone, tendons, skeletal muscle). Two sections at different levels of each ankle joint and associated structures were evaluated for each mouse by histopathology and scored for lesion severity (inflammation). The following scale was used: 0, none (no inflammation); 1, minimal, with inflammation involving <5% of tissue; 2, moderate, with focally extensive areas of inflammation (5% to 25% of tissue and involving one or more tissues); 3, moderate to severe, with focally extensive areas of inflammation (>25% to 50% of tissue and involving multiple tissues and mild distention of tissues associated with joint); 4, severe, with large confluent areas of inflammation (>50% of tissue and involving all tissues and severe distention of tissues associated with joint) (23).
Flow cytometry.Rear legs underwent degloving, hip dislocation, and muscle trimming and were put into a sterile 4-ml PBS solution containing 125 U/ml collagenase (Sigma), 2 μl/ml of DNase (l U/ml; Thermo Scientific), and 4% fetal bovine serum (FBS; Sigma) and then were incubated for 1 h at 37°C with shaking. Joint slurry was put into sterile dishes, stripped of surrounding tissue, strained through an 80-μm mesh, and washed with complete medium. Cells were resuspended in RPMI 1640 medium with 0.1 mM HEPES, and red blood cells were lysed with lysis buffer (final concentration, 0.15 M NH4Cl, 10 mM KHCO3, and 0.1 mM EDTA) (21). Cells were then washed and stained in fluorescence-activated cell-sorting (FACS) buffer (PBS and 2% FBS). Cells were then washed with FACS buffer and fixed overnight in 4% paraformaldehyde at 4°C. After fixation, TUNEL staining was conducted on some samples using the Click-iT Plus TUNEL assay kit (Life Technologies) according to the manufacturer's instructions and then reconstituted in FACS buffer. Fluorescence was acquired on a CyAn ADP analyzer (Beckman Coulter), and FlowJo (Tree Star) software was used for analysis.
Ankle tissue processing for cytokine measurements.Following skin removal, the ankles from each mouse were excised, combined, and then homogenized in toto in 3 to 4 ml of sPBS treated with Halt protease inhibitor cocktail (Thermo Scientific) (23). Homogenized tissues were then centrifuged at 2,000 × g for 5 min, and supernatants were filter sterilized (0.2 μm) and stored at −70°C until analysis using a Luminex MagPix instrument with Milliplex magnetic reagents according to the manufacturer's instructions (Millipore). Luminex data were analyzed with Milliplex Analyst software (Millipore) and normalized to total protein levels determined by a bicinchoninic acid (BCA) protein assay (Thermo Scientific).
Statistical analysis.Statistical analysis of the difference between two mean values was conducted using a two-tailed Student t test with the significance set at a P of ≤0.05, while comparisons of three or more mean values was done using analysis of variance (ANOVA), followed by Tukey's test, with the significance set at a P of ≤0.05. The statistical significance of differences in histology scores was determined by a Mann-Whitney U test.
ACKNOWLEDGMENTS
This work as supported in part by NIH/NIAID R21AI119634 and by funding from the University of Missouri College of Veterinary Medicine and Research Board.
FOOTNOTES
- Received 11 May 2018.
- Returned for modification 7 June 2018.
- Accepted 20 June 2018.
- Accepted manuscript posted online 25 June 2018.
Supplemental material for this article may be found at https://doi.org/10.1128/IAI.00361-18.
REFERENCES
- Copyright © 2018 American Society for Microbiology.