Reassessing the Evolutionary Importance of Inflammasomes

Sensors in the innate immune system survey the extracellular/vacuolar space or the cytosol. TLRs and C-type lectin receptors survey the extracellular/vacuolar compartment. In contrast, RIG-I–like receptors, Nod-like receptors (NLRs), AIM2-like receptors (ALRs), and certain TRIM family proteins survey the cytosolic space. Among these cytosolic sensors, the canonical inflammasomes activate caspase-1. The exception to this paradigm of upstream sensors activating downstream signaling molecules is the so-called “noncanonical inflammasome” pathway, in which murine caspase-11 (and its human orthologs caspase-4 and -5) has the cytosolic sensor and effector functions built into the same protein (1, 2).

Caspase-1 activation leads to processing and secretion of the inflammatory cytokines IL-1β and IL-18, as well as to a form of programmed lytic cell death called pyroptosis. Inflammasomes detect a wide range of cytosolic contaminants. AIM2 detects cytosolic DNA (3). NLRP3 responds to diverse stimuli that seem to trigger catastrophic cellular events (3). In a relatively unique mechanism, NLRC4 is activated by an upstream helper NLR in the NAIP family that is the direct sensor for one of three bacterial proteins: T3SS rod, T3SS needle, or flagellin. These components aberrantly enter the host cytosol through T3SS, presumably while the bacteria is injecting effector proteins to alter host cell function(s) (4). NLRP1b in the mouse responds to the anthrax lethal toxin by serving as a “lure” for this pathogen protease (5). Finally, pyrin (also called TRIM20) senses perturbation of Rho GTPases by bacterial toxins (6). In contrast to these canonical inflammasomes, caspase-11/4/5 detect cytosolic LPS by direct interaction between LPS and their CARD domains (2). These noncanonical inflammasomes trigger pyroptosis but do not process IL-1β directly.

Inflammasomes are evolutionarily maintained in vertebrates (7), although the repertoire may be expanded or contracted in various species. Inflammasomes are proposed to aid in defense against a variety of infections. However, inflammasome activation can also drive septic shock, leading to death (8, 9). This would exert a selective pressure to lose inflammasomes. With this in mind, does the current published evidence support a positive selective pressure that explains the maintenance of inflammasomes over evolutionary time?

In considering the positive selective pressure to maintain inflammasomes in defense against infection, many infectious studies were performed using knockout (KO) mice, typically on the C57BL/6 background. Although inflammasome-deficient mice have increased susceptibility to many pathogens, the effect of inflammasomes on experimental survival studies is typically incremental. Salmonella enterica serovar Typhimurium (S. typhimurium) was one of the best and earliest examples. S. typhimurium replicates somewhat faster in Casp1^−/−Casp11^−/− mice, resulting in significantly increased burdens at any given time point postinfection; for example, at day 5 postinfection, Casp1^−/−Casp11^−/− mice have 10-fold higher burdens than wild-type (WT) mice (10). This translates into the WT mice surviving 2 d longer than Casp1^−/−Casp11^−/− mice (11); however, it is important to remember that the WT mice still succumb, and the ultimate outcome of the infection remains unchanged. Our interpretation of these results is that inflammasomes alone cannot eradicate S. typhimurium but merely slow its kinetics, which could be beneficial in that the animal may survive long enough to develop an adaptive immune response. In further support of this hypothesis, mice infected with Listeria monocytogenes that survive to days 7–8 postinfection are able to mount a CTL response that combats and clears the infection. It is well established that this CTL response eliminates L. monocytogenes (12). Therefore, slowing the course of infection could be the positive selective pressure that maintains inflammasomes over evolutionary time.

Conversely, at first glance, some infectious studies would suggest that inflammasomes have a critical role in defense. For example, we reported that Burkholderia pseudomallei is lethal in Casp1^−/−Casp11^−/− mice, whereas WT mice survive the infection (13). On its surface, such a result seems to have incredible power, leading to the interpretation that inflammasomes fully prevent lethal infection. However, interpretation of this result is confounded by the fact that we chose the dose specifically because it was not lethal in WT mice (i.e., we defined the dose as being sublethal). When a deeper analysis of the dose response is performed, the results lead to a more nuanced interpretation. In this regard, Fabio Re’s group (14) examined three doses of B. pseudomallei infection. They demonstrated that as few as 25 CFU resulted in 100% lethality in Casp1^−/−Casp11^−/− mice, whereas all of the WT mice survived, consistent with our results. However, when Re’s group increased the dose, they found that 100 CFU caused an intermediate lethality in WT mice, and 200 CFU was sufficient to cause 100% lethality in WT mice (14). Therefore, inflammasome detection of B. pseudomallei resulted in an ∼8-fold shift in the lethal infectious dose. Although this is not an insignificant degree of protection, the general conclusion could be that both WT and Casp1^−/−Casp11^−/− mice succumb to extremely low-dose B. pseudomallei challenge. Thus, our interpretation is that, although experiments performed at doses just under the lethal dose can appear fully penetrant, they may actually reflect a more subtle phenotype.

We next considered whether this interpretation is consistent with other published data in infectious models. In contrast to Re’s work (14), most studies only use one infectious dose. Although this limits our ability to precisely evaluate the shift in lethal dose, we can estimate the minimum change supported by the data (rubric is described in Table I legend). Because we wished to contrast various infectious models, we realized the difficulty of comparing pathogen burdens, pathology, or cytokine responses at various time points. In contrast, the lethal challenge provides the same readout (survival) for an array of pathogens, routes, and doses; the survival assay also integrates temporal effects. Therefore, we attempted to identify and collate all lethal challenge studies that were performed in inflammasome-KO mice to illustrate the relative importance of inflammasomes in defense against numerous pathogens (Table I). Unless otherwise indicated (Table I, “Notes” column), all KO mice are Casp1^−/−Casp11^−/− mice. It should be noted that, although many survival studies were performed with bacteria, fewer were published with viral, fungal, and parasitic infection (discussed further below) (15–18). After collation of published survival studies (Table I), it was striking that most changes in lethal dose were estimated to be in the 1–5-fold range.

Therefore, the combined results of all infection survival studies in inflammasome-deficient mice support our interpretation that inflammasomes delay the course of infection but do not have a dominant role in defense. Conversely, what if inflammasomes could provide rapid and potent defense against infection? This would exert a strong selective pressure for pathogens to evolve inflammasome-evasion strategies, minimizing the usefulness of inflammasomes. The difference between these two interpretations has a profound influence on how we think about the importance of inflammasomes.

Indeed, many pathogens evade inflammasome detection. For example, S. typhimurium induces rapid and profound caspase-1 activation in vitro. However, S. typhimurium encodes two T3SSs (SPI-1 and SPI-2), of which only SPI-1 is detected by NLRC4. Therefore, upon host cell entry and during the systemic phase of infection in vivo, S. typhimurium suppresses SPI-1 expression in favor of the “silent” SPI-2 system. Further, S. typhimurium represses flagellin expression (19). The significance of these strategies in vivo cannot be understated, because S. typhimurium engineered to express flagellin or SPI-1 rod protein during systemic infection are completely attenuated in WT mice but are fully virulent in inflammasome-deficient mice (4, 20). L. monocytogenes also represses flagellin and is attenuated when engineered to persistently express it (21, 22). By virtue of being a Gram-positive bacterium, L. monocytogenes also naturally evades caspase-11, simply because it lacks LPS. In contrast, cytosol-invasive Francisella spp. have evolved to actively modify their LPS structure to evade caspase-11 detection (9). Another mechanism to evade cytosolic LPS detection by caspase-11 is by replicating in the vacuole, as is the case with numerous vacuolar pathogens, including S. typhimurium (13).

In contrast, Yersinia spp. encode the effector YopM, which is capable of inhibiting caspase-1 by directly binding to the active site. This is critically important in vivo, because Y. pseudotuberculosis yopM mutants are attenuated in WT mice but retain full virulence in Casp1^−/−Casp11^−/− mice (23). Similarly, Shigella flexneri encode the OspC3 T3SS effector that inhibits human caspase-4 (24), thus preventing detection of its LPS as the bacterium invades the cytosol. Direct inhibition of inflammasome components is not limited to bacteria. Indeed, poxviruses encode crmA, a serpin that can inhibit several caspases (both inflammatory and apoptotic) and was identified as a caspase-1 inhibitor before the discovery of pyroptosis (25). Further, Kaposi’s sarcoma–associated herpesvirus encodes an NLR analog (the tegument protein Orf63) that inhibits NLRP1 and NLRP3 activation of caspase-1 (26).

Another possible strategy for pathogens to cope with inflammasomes is to avoid the consequences downstream of inflammasome detection. B. pseudomallei may be an example of this strategy by resisting the neutrophil killing (27) that occurs after the bacteria are ejected into the extracellular space by pyroptosis (4). Staphylococcus aureus may be another example, because they are highly resistant to killing by neutrophils and have numerous toxins that inhibit neutrophil chemotaxis (28).

These pathogens are striking examples that demonstrate the importance of evading inflammasome detection (Fig. 1). They show that inflammasomes could provide extremely potent defense against infection in vivo but fail to do so because many pathogens minimize detection.

Vertebrate-adapted pathogens have evolved in the context of selective pressure exerted by the host immune system and have accordingly potent virulence traits. In contrast, for opportunistic pathogens for which humans are dead-end accidental hosts, there is no selective pressure to evade the human immune response. However, such microbes could also encode an impressive array of virulence factors, which can enable extreme pathogenicity. For example, Burkholderia thailandensis encodes cytosol-invasive T3SS but almost never causes infection in people (29). As such, it has been used as an experimental surrogate for its close relative B. pseudomallei (a BSL3 pathogen); in this capacity, our laboratory studied B. thailandensis infection in vivo in mice.

Shockingly, during systemic infection with B. thailandensis there is a 1,000,000-fold change between Casp1^−/−Casp11^−/− and WT mice (Table I). The resistance conferred by inflammasomes is extremely efficient; WT mice fully sterilize even high-dose B. thailandensis infection (2 × 10⁷ CFU) within just 1 d (29). Clearly, inflammasomes can provide potent protection when presented with this specific pathogen. The strength of this phenotype was extremely surprising, rivaled only by the effect of inflammasomes upon pathogens that were engineered to remove inflammasome-evasion strategies. This made us re-evaluate the nature of B. thailandensis; we had simplistically considered it to be a surrogate for B. pseudomallei. However, it is important to keep in mind that B. thailandensis occupies a specific environmental niche as a soil microbe, where its T3SS is presumably used to invade the cytosol of an unknown eukaryotic host (30).

Is B. thailandensis an isolated unique case? With these thoughts in mind, we began to screen for other environmental opportunistic pathogens that primarily cause disease in immunocompromised individuals, hypothesizing that healthy individuals would competently clear them via the activity of inflammasomes.

In this endeavor, we discovered another ubiquitous environmental opportunistic pathogen against which inflammasomes play a similarly striking role, conferring a >50,000-fold shift in the lethal dose (Table I). Chromobacterium violaceum is a Gram-negative bacterium that lives in the sediment of freshwater rivers and lakes. Although C. violaceum encodes two T3SSs that are similar to Salmonella SPI-1 and SPI-2 and can be detected by NLRC4 (31), it virtually never causes disease in immunocompetent people (32).

A confounding factor in the identification of these examples is that there may be redundant host defense pathways that independently confer sterilizing innate immunity. Indeed, it is quite surprising that defense against C. violaceum and B. thailandensis is monoallellic. However, our studies with C. violaceum revealed that pyroptosis is partially redundant with NK cell cytotoxicity, although both are dependent on caspase-1 activation (33). Further, if an opportunistic pathogen does not kill the host within 1 wk, the adaptive-immune response could compensate for the loss of inflammasomes.

As evidenced by the vertebrate-adapted column in Table I, several opportunistic pathogens have been examined in Casp1^−/−Casp11^−/− mice, yet none demonstrated such large changes in the lethal dose as seen with C. violaceum and B. thailandensis. C. violaceum encodes two T3SSs (34) and, concomitantly, defense is conferred by NLRC4 in vivo (33). B. thailandensis invades the host cytosol, and its LPS is detected by caspase-11 (13, 29), which confers defense in vivo. In both cases, it is the specific virulence traits that are detected by the inflammasomes. The evasion strategies used by vertebrate-adapted pathogens are likely not present in these environmental bacteria, because they did not evolve in the presence of potential inflammasome detection. For example, Klebsiella pneumoniae and Vibrio vulnificus are opportunistic pathogens that typically cause disease in immunocompromised patients (35, 36), which fits our model that healthy individuals are protected by inflammasomes. However, neither has a strong in vivo phenotype in inflammasome-deficient mice (Table I). Additionally, neither encodes a T3SS or is cytosol invasive, as is the case for B. thailandensis and C. violaceum. Thus, they may lack inflammasome agonists.

The predominance of evidence suggests that vertebrate-adapted pathogens evolved to minimize the effectiveness of inflammasomes; in vivo phenotypes are typically incremental, and evasion strategies are prevalent. Why, then, are inflammasomes maintained in the human genome? The extreme virulence of B. thailandensis and C. violaceum in inflammasome-deficient mice leads us to propose a new hypothesis regarding the importance of inflammasomes in the immune system. We hypothesize that inflammasomes defend against ubiquitous environmental microbes with specific virulence traits. Further, we propose that the inflammasome-directed defense is so efficient that the infection never progresses to clinically apparent symptoms. It is interesting to note that inflammasome-deficient patients have not been identified. We speculate that a survey of apparently immunocompetent people with susceptibility to specific opportunistic pathogens, such as C. violaceum, could identify patients with inflammasome mutations. For example, in a case series of 106 C. violaceum infections, only 15% were directly attributed to an immunocompromising comorbidity (such as chronic granulomatous disease) (32).

There are several caveats that temper interpretation of the extreme virulence of B. thailandensis and C. violaceum in Casp1^−/−Casp11^−/− mice. One consideration is that Casp1^−/−Casp11^−/− animals are inbred and typically maintained on a C57BL/6 background that naturally lacks NRAMP1 (37). Mice lacking NRAMP1 have increased susceptibility to numerous intracellular pathogens, including S. typhimurium, L. monocytogenes, and Mycobacterium tuberculosis. This caveat has not gone unnoticed. In 2006, Lara-Tejero et al. (10) investigated the roles of inflammasomes in Nramp1-sufficient mice during S. typhimurium infection. However, regardless of the presence of a functional Nramp1 gene, Casp1^−/−Casp11^−/− mice still showed an incremental susceptibility to S. typhimurium infection in comparison with their WT counterparts.

Further, 129/SvEv mice are Nramp1 sufficient but carry a spontaneous mutation in caspase-11 (38). Aachoui et al. (29) infected 129/SvEv (Casp11^−/−Nramp1^+/+) mice and found that they have extreme susceptibility to B. thailandensis that is comparable to C57BL/6 Casp11^−/− (Nramp1^−/−) mice, whereas BALB/c (Casp11^+/+Nramp1^−/−) mice remained resistant. Thus, the potency of inflammasome defense against this environmental bacterium was not significantly altered by the presence or absence of Nramp1.

In addition to the influence of Nramp1 on different inbred mouse strains, there is the fact that mice are not men; although this may seem to be an obvious caveat, it is one worth expanding upon. There are species barriers that can influence the virulence of pathogens. There are several examples of bacteria (Shigella spp., Salmonella typhi) that fail to establish mouse infections despite infectivity in humans. However, there are far more examples of viruses that have human-specific strains that do not infect mice: human CMV, HIV, hepatitis B virus, hepatitis C virus, EBV, measles, mumps, and many more (39). Therefore, there could be more stringent noninflammasome barriers restricting interspecies dissemination of viruses in comparison with bacteria. Therefore, our hypothesis that inflammasomes defend against environmental microbes might not hold true for viruses.

Certain inflammasomes, but not all, are conserved between mice and humans. NLRC4, NLRP3, caspase-1, and AIM2 are highly conserved over evolutionary time from sharks to humans (based on searches of the Basic Local Alignment Search Tool by V.I. Maltez and E.A. Miao of inflammasome components). Yet, despite the high conservation of the ALR family member AIM2, there are 12 additional ALRs in mice but only 3 more in humans, suggesting active evolution of this gene family (40). Because AIM2/ALRs are often attributed to recognizing viral infections, the diversity within this family over evolutionary time could argue against the hypothesis that they defend against specific environmental pathogens. Similarly, mice encode three NLRP1 genes that are highly polymorphic among inbred mouse strains, and there is only one human NLRP1 (5). This again suggests active evolution of the NLRP1 locus. Thus, our hypothesis might only hold true for certain highly conserved inflammasomes and not for the more divergent or polymorphic inflammasomes.

If vertebrate-adapted pathogens minimize the effectiveness of inflammasomes, why should we continue to study them? First, even suboptimal inflammasome responses may have clinically beneficial effects during infection by vertebrate-adapted pathogens. Second, inflammasomes seem to be important drivers of the pathology of severe sepsis and septic shock (8, 9, 41). There are no immunologically directed therapies for sepsis; supportive treatments are useful but, nevertheless, 28% of patients with sepsis die. Study of the downstream effect(s) of aberrant inflammasome activation could aid in the development of new directed treatments.

If vertebrate-adapted pathogens minimize the effectiveness of inflammasomes, how should we best study them? Opportunistic pathogens have the potential to lead us to novel insights and to reveal therapeutic approaches. Our work with C. violaceum demonstrates the reality of this potential: we discovered a novel link between inflammasome activation and in vivo perforin-mediated defense that was then applicable to L. monocytogenes infection as a cytokine therapy (33). With the rise in antibiotic-resistant microbes, there is a pressing need for new immunotherapeutic approaches to treat infection.

We apologize if we missed any publications in our table. We attempted to be as thorough as possible and only cited studies with live infections and both WT and inflammasome-deficient mice.

The authors have no financial conflicts of interest.