Caspase-11 Protects Against Bacteria That Escape the Vacuole

See allHide authors and affiliations

Science  22 Feb 2013:
Vol. 339, Issue 6122, pp. 975-978
DOI: 10.1126/science.1230751


Caspases are either apoptotic or inflammatory. Among inflammatory caspases, caspase-1 and -11 trigger pyroptosis, a form of programmed cell death. Whereas both can be detrimental in inflammatory disease, only caspase-1 has an established protective role during infection. Here, we report that caspase-11 is required for innate immunity to cytosolic, but not vacuolar, bacteria. Although Salmonella typhimurium and Legionella pneumophila normally reside in the vacuole, specific mutants (sifA and sdhA, respectively) aberrantly enter the cytosol. These mutants triggered caspase-11, which enhanced clearance of S. typhimurium sifA in vivo. This response did not require NLRP3, NLRC4, or ASC inflammasome pathways. Burkholderia species that naturally invade the cytosol also triggered caspase-11, which protected mice from lethal challenge with B. thailandensis and B. pseudomallei. Thus, caspase-11 is critical for surviving exposure to ubiquitous environmental pathogens.

Canonical inflammasomes, such as NLRP3, NLRC4, and AIM2, are cytosolic sensors that detect pathogens or danger signals and activate caspase-1, which leads to secretion of the proinflammatory cytokines interleukin-1β (IL-1β) and IL-18 and pyroptosis, a form of programmed cell death (1). Pyrin domain–containing inflammasomes, including NLRP3, signal through the ASC adaptor protein to recruit caspase-1 (fig. S1). Many diverse agonists cause cytosolic perturbations that are detected through NLRP3; however, the underlying mechanisms remain obscure (2). In contrast, the CARD domain–containing inflammasome NLRC4 can signal directly to caspase-1, which results in pyroptosis, as well as indirectly through ASC to promote IL-1β and IL-18 secretion (fig. S1) (1, 3). NLRC4 detects bacterial flagellin and type III secretion system (T3SS) rod or needle components within the macrophage cytosol (46). Together, NLRC4- and the ASC-dependent inflammasomes account for all known canonical caspase-1 activation pathways.

Burkholderia pseudomallei is a Gram-negative bacterium endemic to Southeast Asia that causes melioidosis and is a potential biologic weapon (7). B. pseudomallei uses a T3SS to escape the phagosome and to replicate in the cytosol. NLRC4 and NLRP3 both detect B. pseudomallei, which promotes IL-1β secretion from murine bone marrow–derived macrophages (BMMs) (8) (Fig. 1A). Despite encoding many of the same virulence factors as B. pseudomallei, including T3SS and T6SS, the closely related B. thailandensis is far less virulent (9). We therefore hypothesized that NLPR3 and NLRC4 also detect B. thailandensis, and indeed, NLRP3 and NLRC4 accounted for all IL-1β secretion in response to B. thailandensis (Fig. 1B). We next determined whether inflammasome activation is critical to survival after B. thailandensis challenge using caspase-1–deficient mice. Kayagaki et al. recently showed that all existing caspase-1–deficient mice also lack caspase-11 because of the backcrossing of a mutant Casp11 allele from 129 into C57BL/6 mice (10). Inflammasome detection was critical for resistance to B. thailandensis, as Casp1–/–Casp11–/– animals succumbed to the infection (Fig. 1C and fig. S2A). In contrast, wild-type C57BL/6 mice survived high-dose intraperitoneal (i.p.) or intranasal (i.n.) challenge (Fig. 1C and fig. S2A). To our surprise, Nlrc4–/–Asc–/– mice that are deficient in all known canonical inflammasomes were also resistant (Fig. 1D and fig. S2B). This indicated that an unknown signaling pathway provides protection via either caspase-1 or -11 (see pathway schematic in fig. S1). Resistance to B. thailandensis was at least partially independent of IL-1β and IL-18, depending on the route of infection (Fig. 1E and fig. S2C), which suggested that both cytokines and pyroptosis can contribute to protection. We therefore examined pyroptosis in vitro and found that cytotoxicity in response to B. thailandensis was impaired in Casp1–/–Casp11–/– BMMs (Fig. 1F). Consistent with our in vivo data, pyroptosis in vitro did not require Nlrc4 or Asc (Fig. 1F). B. pseudomallei similarly triggered pyroptosis in Nlrc4–/–Asc–/– macrophages (Fig. 1G). These results indicate that a pyroptosis-inducing pathway distinct from all known canonical inflammasomes detects B. thailandensis and protects against lethal infection.

Fig. 1

Burkholderia detection and protection conferred by Casp1/11 is independent of all known canonical inflammasomes. LPS-primed BMMs were infected with B. pseudomallei (A and G) or B. thailandensis (B and F) for 4 hours. (A and B) IL-1β secretion was determined by enzyme-linked immunosorbent assay (ELISA), or (F and G) cytotoxicity was determined by lactate dehydrogenase (LDH) release assay. (C, D, and E) Survival curves of wild-type C57BL/6 or the indicated knockout mice infected with 2 × 107 colony-forming units (cfu) i.p. with B. thailandensis. Days p.i., days postinfection. Data are representative of at least three (A), (B), (F), and (G) or of two (D) and (E) experiments. (C) Data are pooled from three experiments. For number of mice in each panel, see table S2. Statistically significant differences with respect to controls are indicated (Student's t test or log-rank test for survival; *P ≤ 0.05).

Inflammasomes discriminate pathogens from nonpathogens by detecting contamination or perturbation of the cytosolic compartment (11). The B. thailandensis T3SS facilitates bacterial access to the cytosol and was required for induction of pyroptosis, whereas the virulence-associated T6SS was dispensable (Fig. 2A). We therefore hypothesized that macrophages detect vacuolar lysis or release of bacteria into the cytosol.

Fig. 2

Diverse cytosolic bacteria activate pyroptosis independent of NLRC4, NLRP3, and ASC. (A) LPS-primed BMMs were infected for 4 hours with either B. thailandensis or the indicated mutants, and cytotoxicity was determined. BMMs were infected for 8 hours with (B) S. typhimurium or S. typhimuriumsifA or (D) S. typhimuriumsifA or S. typhimuriumsifA flgB, and cytotoxicity was determined. WT, wild type. LPS-primed BMMs were infected for 8 hours with (C) S. typhimurium or S. typhimuriumsifA or (E) S. typhimuriumsifA or S. typhimuriumsifA flgB, and IL-1β secretion was determined. (F and G) Cytotoxicity in wild-type or Asc–/– BMMs infected for 4 hours with L. pneumophila, L. pneumophilaflaA, or L. pneumophilasdhAflaA. Cytotoxicity was determined by LDH release and IL-1β secretion by ELISA. Data are representative of at least three experiments. Statistically significant differences with respect to controls are indicated (Student's t test; *P ≤ 0.05; n.s., P > 0.05).

In order to establish their intracellular vacuolar growth niche, Salmonella typhimurium and Legionella pneumophila use T3SS and T4SS, respectively, to translocate effector proteins that work in concert to maintain the stability of these altered, bacteria-containing vacuoles (1214). Loss of the S. typhimurium SifA or L. pneumophila SdhA effectors causes rupture of the vacuole and release of bacteria into the cytosol (1517). S. typhimurium uses two distinct T3SSs encoded by the Salmonella pathogenicity island 1 (SPI1) and SPI2; these two T3SSs translocate distinct batteries of effectors, such as SifA by SPI2 (18). Although S. typhimurium–expressing SPI1 and flagellin are readily detected by NLRC4 (19, 20), bacteria grown under conditions that mimic the vacuolar environment express SPI2 and repress flagellin, which minimizes canonical inflammasome detection (1, 11, 21). Infection of BMMs with S. typhimurium that lacked sifA, however, significantly increased IL-1β secretion and pyroptosis (Fig. 2, B and C). IL-1β secretion was dependent on canonical inflammasomes (Fig. 2B), whereas pyroptosis was still observed in Nlrc4–/–Asc–/– and Nlrp3–/–Nlrc4–/– macrophages (Fig. 2C). Furthermore, the NLRC4 inflammasome agonist flagellin was not required for these responses (Fig. 2, D and E). Thus, macrophages detect S. typhimurium when it aberrantly enters the cytosol and activates pyroptosis independently of all known canonical inflammasomes.

L. pneumophila also translocates flagellin through its T4SS. Thus, L. pneumophila mutants lacking flagellin (∆flaA) evaded NLRC4 detection (Fig. 2F) (2). In contrast, L. pneumophilaflaA ∆sdhA mutants induce caspase-1 activation (16, 17), IL-1β secretion (17), and pyroptosis (Fig. 2F) (17). The AIM2-ASC canonical inflammasome has been implicated in L. pneumophilaflaAsdhA–induced IL-1β secretion, likely by detecting DNA released from bacteria lysing in the cytosol; however, the role of AIM2-ASC in pyroptosis was not examined (17). Analogous to S. typhimuriumsifA, L. pneumophilaflaA ∆sdhA induced pyroptosis in the absence of flagellin and ASC (Fig. 2G), which rules out all canonical inflammasomes in triggering pyroptosis under these infection conditions. These data demonstrate that diverse bacteria are detected in the cytosol.

Because IL-1β secretion required the canonical inflammasomes, whereas pyroptosis did not, we hypothesized that cell death is triggered by a distinct mechanism mediated by caspase-11. Like caspase-1, caspase-11 is an inflammatory caspase that can directly trigger pyroptosis (fig. S1). Caspase-11 can also promote IL-1β secretion dependent upon NLRP3, ASC, and caspase-1 (10, 2224). Because caspase-1 is activated by recruitment to an oligomerized platform known as the inflammasome, Kayagaki et al. hypothesized that a similar oligomeric structure would activate caspase-11, which they termed the noncanonical inflammasome (10). Although the cholera toxin B subunit and many different Gram-negative bacteria can trigger caspase-11 activation in vitro (10, 2224), the nature of the physiologic stimulus that activates caspase-11 during infection remains uncertain.

Caspase-11 activation requires priming through a Toll-like receptor 4 (TLR4)–TRIF–STAT1 pathway (10, 2224) (TRIF is a TLR adaptor protein; STAT1 is an interferon receptor signaling protein). Consistent with this, Tlr4–/– and Trif–/– macrophages did not undergo pyroptosis after S. typhimurium sifA infection, whereas cell death was observed in macrophages deficient in the other TLR4 adaptor, Myd88 (Fig. 3A). This dependence could be overcome by priming the macrophages with interferon-γ (IFN-γ) (Fig. 3A), which signals through STAT1. IFN-γ or lipopolysaccharide (LPS) priming significantly increased the sensitivity of macrophages to S. typhimurium sifA (Fig. 3A and fig. S3A). These priming effects correlated with increased caspase-11 expression (fig. S3, B and C) but could also be mediated by enhancing aberrant vacuolar rupture. We used retroviruses to complement Casp1–/–Casp11–/– macrophages with either Casp1 or Casp11 to determine which was involved. Caspase-11 alone promoted pyroptosis without IL-1β secretion after B. thailandensis infection, whereas caspase-1 enabled both responses (Fig. 3B). This is consistent with B. thailandensis detection through NLRC4- and/or NLRP3-activating caspase-1 (8) and an additional pathway activating caspase-11. In contrast, the responses to S. typhimuriumsifA or L. pneumophilaflaAsdhA acted through caspase-11 but not caspase-1 (Fig. 3, C and D). We further confirmed that caspase-11 was responsible for the cell death observed in Nlrc4–/–Asc–/– macrophages, using microRNA-adapted short hairpin RNA (shRNAmir) (Fig. 3, E and F, and fig. S3E). Finally, Casp11–/– BMMs revealed that caspase-11 was required for pyroptosis after B. thailandensis, S. typhimuriumsifA, and L. pneumophilaflaAsdhA (Fig. 3, G to I). Although a previous report suggested that NLRC4 signals through caspase-11 to alter phagosomal trafficking (25), we saw no evidence that NLRC4 contributes to caspase-11–dependent cell death (Figs. 1F and 2D and fig. S4). Pyroptosis initiated by caspase-11 was morphologically similar to pyroptosis triggered by caspase-1 (fig. S5, A and B). Therefore, macrophages activate caspase-11 in response to cytosolic B. thailandensis, S. typhimurium, or L. pneumophila (fig. S1).

Fig. 3

Caspase-11 mediates pyroptosis after infection by cytosolic bacteria. Macrophage cytotoxicity and IL-1β secretion were determined after infection with S. typhimurium ΔsifA (8 hours), L. pneumophila ΔflaA ΔsdhA (4 hours), or B. thailandensis (4 hours). (A) C57BL/6, Casp1–/–Casp11–/–, Tlr4–/–, Trif–/–, and Myd88–/– BMMs infected with S. typhimuriumsifA with or without IFN-γ priming before infection. (B and C) Retroviral transduction was used to complement Casp1 or Casp11 in Casp1–/–Casp11–/– immortalized BMMs. Macrophages were primed with LPS (B) or IFN-γ (C), and responses to B. thailandensis (B) or S. typhimurium ΔsifA (C) infection were examined. (D) Control or complemented Casp1–/–Casp11–/– BMMs infected with L. pneumophila ΔflaA ΔsdhA. (E and F) Retroviral transduction was used to introduce control or Casp11-targeting of shRNAmir into Nlrc4–/–Asc–/– immortalized BMMs. Macrophages were primed overnight with LPS (E) or IFN-γ (F) and then infected as indicated. (G to I) C57BL/6, Casp1–/–Casp11–/–, and Casp11–/– BMMs infected with B. thailandensis (G), S. typhimurium ΔsifA (H), or L. pneumophila ΔflaA ΔsdhA (I). Data are representative of at least three (A to C), (E), (G), and (H) or of two (D), (F), and (I) independent experiments. Statistically significant differences with respect to controls are indicated (Student's t test; *P ≤ 0.05). nd, none detected.

S. typhimuriumsifA is attenuated (15), which has previously been attributed to the role of SifA in coordinating intracellular trafficking of the Salmonella-containing vacuole. We hypothesized that this attenuation was actually due to innate immune detection though caspase-11. Indeed, S. typhimuriumsifA was mildly attenuated in C57BL/6 mice as expected, but this was not replicated in Casp1–/–Casp11–/– mice (Fig. 4, A and B). We next determined the relative clearance of S. typhimuriumsifA during co-infection with wild-type S. typhimurium, a more quantitative measure of virulence than lethal challenge. We recovered 16 times as much wild-type S. typhimurium as S. typhimuriumsifA from C57BL/6 mice; however, from Casp11–/– mice, we only recovered four times as many wild-type bacteria (Fig. 4C). This indicates that caspase-11 clears S. typhimuriumsifA in vivo; in contrast, wild-type S. typhimurium effectively evades caspase-11 (23) by remaining within the vacuole. The remaining S. typhimuriumsifA attenuation likely reflects the role of sifA as a virulence factor promoting intracellular replication. Moreover, all known canonical inflammasomes were dispensable for S. typhimuriumsifA clearance, as were IL-1β and IL-18 (Fig. 4D), which implicated pyroptosis as the mechanism of clearance. Clearance of bacteria after pyroptosis is mediated by neutrophils through generation of reactive oxygen (21). Consistent with this, NADPH oxidase–deficient p47phox–/– mice were also defective for clearance of S. typhimuriumsifA (Fig. 4D). However, TLR4 and IFN-γ were not required (Fig. 4E), which suggests that there is redundant priming of caspase-11 pathways in vivo. Therefore, caspase-11 protects mice from S. typhimuriumsifA, and because IL-1β and IL-18 are not required, pyroptosis is likely to be the mechanism of bacterial clearance in this case.

Fig. 4

Caspase-11 protects against cytosolic bacteria in vivo. (A and B) S. typhimurium or S. typhimuriumsifA were injected i.p. into C57BL/6 (1000 cfu) or Casp1–/–Casp11–/– mice (250 cfu), and survival was monitored. (C to E) The indicated mice were infected with 5 × 104 cfu of both wild-type S. typhimurium and S. typhimurium ∆sifA marked with ampicillin or kanamycin resistance, respectively. Bacterial loads from three to four mice per genotype were determined 48 hours later, and the competitive index was calculated [CI = log(S. typhimuriumsifA/S. typhimurium)]. A CI of −1 corresponds to 10 cfu of S. typhimurium for every 1 cfu of S. typhimuriumsifA. (F and G) C57BL/6, Casp1–/–Casp11–/–, or Casp11–/– mice were infected with (F) 2 × 107 cfu mouse passaged B. thailandensis i.p. or (G) 100 cfu B. pseudomallei i.n. (A), (B), (F), and (G) Data are pooled from two independent experiments. (C to E) Representative of two experiments. For the number of mice in each panel, see table S2. Statistically significant differences with respect to controls are indicated (Student's t test or log-rank test for survival; *P ≤ 0.05; n.s., P > 0.05).

We next examined the susceptibility of Casp11–/– mice to the naturally cytosolic pathogens B. thailandensis and B. pseudomallei. Although C57BL/6 mice are resistant to B. thailandensis infection, Casp11–/– mice succumbed (Fig. 4F). Likewise, Casp11–/– mice succumbed to B. pseudomallei infection, whereas C57BL/6 mice survived (Fig. 4G). Because Nlrc4–/– mice are also susceptible to B. pseudomallei infection (8), we conclude that both caspase-1 and caspase-11 play critical roles in limiting B. pseudomallei infection.

Collectively, these data demonstrate, for the first time, that caspase-11 protects animals from lethal infection by bacteria that have the ability to invade the cytosol. This could be critical for defense against ubiquitous environmental bacteria such as B. thailandensis that encode virulence factors but that have not evolved to evade caspase-11 detection. It will be interesting to determine whether caspase-11 is activated in response to the process of vacuolar rupture or the presence of bacteria within the cytosol. Caspase-11 also responds to vacuolar bacteria under delayed kinetics, but such responses have not been shown to provide protection from infection in vivo (10, 2224). LPS-induced septic shock is mediated by caspase-11 (10), which suggests that caspase-11 can be activated by other mechanisms besides cytosol-localized bacteria. Thus, we propose that caspase-11 provides protection against pathogens, but is dysregulated during overwhelming infection and so contributes to septic shock and mortality. It will be interesting to determine whether caspase-11 triggers eicosanoid secretion, as is seen for caspase-1, and whether these mediators contribute to septic shock (26). The identity of the hypothesized noncanonical inflammasome(s) that activate caspase-11 and the precise nature of the activating signal will shed more light on the mechanisms by which caspase-11 can both promote innate immunity and exacerbate immunopathology. These insights may lead to novel therapies to treat infection and sepsis.

Supplementary Materials

Materials and Methods

Figs. S1 to S5

Tables S1 and S2

References (2748)

References and Notes

  1. Acknowledgments: The authors thank V. Dixit for sharing mice (a material transfer agreement is on file) and M. Heise for sharing mice and S. Miller, J. Mougous, and H. Schweizer for sharing bacterial strains. We also thank D. Rodriguez and Z. Zhou for managing mouse colonies. The data presented in this manuscript are tabulated in the main paper and in the supplementary materials. This work was supported by NIH grants AI097518 (E.A.M.), AI057141 (E.A.M. and A.A.), AI065359 (P.A.C.), AI075039 (R.E.V.), AI080749 (R.E.V.), and AI063302 (R.E.V.); Investigator Awards from the Burroughs Wellcome Fund and Cancer Research Institute (R.E.V.); and an NSF graduate fellowship (M.F.F.).
View Abstract

Navigate This Article