Inhibition of caspase-1 or gasdermin-D enable caspase-8 activation in the Naip5/NLRC4/ASC inflammasome

Legionella pneumophila is a Gram-negative, flagellated bacterium that survives in phagocytes and causes Legionnaires’ disease. Upon infection of mammalian macrophages, cytosolic flagellin triggers the activation of Naip/NLRC4 inflammasome, which culminates in pyroptosis and restriction of bacterial replication. Although NLRC4 and caspase-1 participate in the same inflammasome, Nlrc4-/- mice and their macrophages are more permissive to L. pneumophila replication compared with Casp1/11-/-. This feature supports the existence of a pathway that is NLRC4-dependent and caspase-1/11-independent. Here, we demonstrate that caspase-8 is recruited to the Naip5/NLRC4/ASC inflammasome in response to flagellin-positive bacteria. Accordingly, caspase-8 is activated in Casp1/11-/- macrophages in a process dependent on flagellin, Naip5, NLRC4 and ASC. Silencing caspase-8 in Casp1/11-/- cells culminated in macrophages that were as susceptible as Nlrc4-/- for the restriction of L. pneumophila replication. Accordingly, macrophages and mice deficient in Asc/Casp1/11-/- were more susceptible than Casp1/11-/- and as susceptible as Nlrc4-/- for the restriction of infection. Mechanistically, we found that caspase-8 activation triggers gasdermin-D-independent pore formation and cell death. Interestingly, caspase-8 is recruited to the Naip5/NLRC4/ASC inflammasome in wild-type macrophages, but it is only activated when caspase-1 or gasdermin-D is inhibited. Our data suggest that caspase-8 activation in the Naip5/NLRC4/ASC inflammasome enable induction of cell death when caspase-1 or gasdermin-D is suppressed.

Introduction Legionella pneumophila is the causative agent of Legionnaires' disease. It was identified for the first time in 1976, after an atypical pneumonia affected the participants of the American Legion Convention in Philadelphia, United States [1]. After isolation, L. pneumophila were characterized as Gram-negative, flagellated, intracellular facultative bacteria [2,3]. The species of Legionella were found mainly in freshwater and soil environments, including lakes and irrigation systems [4]. Infection of humans occurs upon inhalation of water droplets derived from these environments containing Legionella [5]. After inhalation, L. pneumophila can subvert the normal vesicle traffic within alveolar macrophages and form LCV (Legionella-containing vacuoles), a process that is mediated by the injection of hundreds of bacterial effectors through a type IV secretion system called Dot/Icm [6][7][8][9]. During its evolution, L. pneumophila were selected based on their replication in protozoa but not in humans, which are accidental hosts [10]. Consequently, L. pneumophila can be recognized by many innate immune receptors in mammalian cells, including proteins from the family of the nucleotide-binding domain and leucine-rich repeat-containing proteins (NLRs). These characteristics make L. pneumophila an excellent model for the study of innate immunity, including intracellular signaling pathways and inflammasomes.
The major inflammasome that leads to the restriction of Legionella replication in macrophages is Naip5/NLRC4. This pathway was discovered in mouse cells upon observations that macrophages from the A/J mouse strain, but not cells from other mice strains, are susceptible to L. pneumophila replication [11]. The resistance was mapped to the Lgn1 locus, which encodes several copies of Naip genes, including Naip5 (Birc1e), which is the gene responsible for resistance [12][13][14][15][16]. Successful lines of investigation culminated in the demonstration that Naip5 recognizes bacterial flagellin and interacts with NLRC4 for caspase-1 activation and the restriction of bacterial replication [17][18][19][20]. This platform was named the Naip5/NLRC4 inflammasome and triggers pore formation and pyroptosis, which has been considered one of the most important mechanisms for the restriction of intracellular pathogen replication via inflammasomes [21][22][23][24]. Host cell death via pyroptosis eliminates intracellular parasite replication and traps intracellular microbes in pyroptotic cells, facilitating microbial destruction by additional phagocytes [23,[25][26][27][28][29]. Pyroptosis occurs concomitantly with the secretion of inflammatory cytokines such as IL-1β and IL-18, a process that requires the adaptor molecule ASC and the formation of NLRC4/ASC puncta [20,30]. ASC also functions as an adaptor protein for other inflammasomes, including AIM2 and NLRP3, which triggers the processing of caspase-1 and caspase-8 [21,[31][32][33][34]. Of note, Naip5/NLRC4 appears to be the only inflammasome required for the restriction of L. pneumophila replication. Macrophages that are deficient in NLRP3 or AIM2 can efficiently restrict L. pneumophila replication [20,21,23,35]. However, the participation of ASC in the resistance of L. pneumophila infection is controversial. In murine macrophages, ASC is dispensable for the induction of pyroptosis and the restriction of bacterial replication [20,21]. By contrast, experiments performed with human monocytes indicate that ASC silencing leads to an increase in bacterial replication [36,37]. Thus, the role of ASC in the restriction of L. pneumophila replication is still unclear.
We have previously demonstrated the existence of a pathway that is dependent on flagellin and NLRC4 but independent of caspase-1 [38]. Here, we used macrophages and Casp1/11 -/mice to systematically assess this pathway. By searching for additional components that operate in the NLRC4 inflammasome independently of caspase-1/11, we found that caspase-8 interacts with NLRC4 in a process that is dependent on ASC. This pathway effectively accounts for resistance to infection in macrophages and in vivo when caspase-1 is absent. In wild-type cells, caspase-8 is recruited to the Naip5/NLRC4/ASC/caspase-1 inflammasome, but is not activated. Caspase-8 activation in this platform only occurs when caspase-1 or gasdermin-D is inhibited, suggesting that this pathway may be important when pyroptosis is inhibited.

11-dependent
We have previously demonstrated that activation of the flagellin/NLRC4 inflammasome triggers caspase-1-dependent and independent responses to restrict Legionella replication in macrophages and in mouse lungs [38]. However, the caspase-1-independent mechanisms underlying this pathway are unknown. To further characterize this pathway, we performed growth curves using high and very low multiplicity of infections (MOIs) in bone marrowderived macrophages (BMDMs). Macrophages were infected with wild-type L. pneumophila in the JR32 background (WT Lp) and the isogenic mutants flaAand fliI -. FliI is an ATPase that is required for the secretion of flagellin through the flagellar apparatus [39]. Consequently, fliImutants express flagellin but are non-motile and non-flagellated, making them an appropriate control for flaAmutants for investigations related to the role of flagellin. We found that BMDMs from C57BL/6 and Asc -/mice fully restrict the replication of WT Lp and fliIbacteria at low and high MOIs. In contrast, Nlrc4 -/cells are permissive and Casp1/11 -/cells are partially restrictive (S1 Fig). Bacterial mutants for flagellin bypass NLRC4-mediated growth restriction and replicate in all macrophages as previously described [17][18][19]40]. These data support previous reports showing that ASC is not required for the restriction of L. pneumophila replication in the presence of caspase-1/11 [20,21]. In addition, these data further support our previous assertion that flagellin triggers an uncharacterized pathway that is dependent on NLRC4 and independent of caspase-1 and caspase-11 [38]. We decided to use BMDMs from Casp1/11 -/mice to further investigate this NLRC4-dependent and caspase-1/ 11-independent pathway.
Flagellated L. pneumophila triggers NLRC4 puncta that associate with caspase-8 in a process that is ASC-dependent The transduction of BMDMs with a retrovirus encoding NLRC4 fused to GFP allows the visualization of NLRC4 puncta in the cytoplasm of macrophages infected with flagellated bacteria [30]. Here, we used this retroviral system to investigate the formation of the NLRC4 inflammasome in the absence of caspase-1/11. BMDMs from Casp1/11 -/mice were transduced with NLRC4-GFP and infected with WT Lp, fliIand flaAat different MOIs and time points. We found that WT Lp and fliItriggered the formation of NLRC4 puncta in the absence of caspase-1/11 ( Fig 1A). The formation of NLRC4 puncta was influenced by the MOI and significantly diminished in response to flaAbacteria ( Fig 1A). Next, we evaluated the requirement of ASC for the formation of NLRC4 puncta in the absence of caspase-1/11. We constructed a mouse that was deficient in ASC and caspase-1/11 and found that whereas the formation of NLRC4 puncta occurred in the absence of caspase-1/11, ASC was essential for formation of the NLRC4 puncta ( Fig 1B). These data are in agreement with previous findings indicating that ASC is critical for the nucleation of several inflammasomes, including AIM2, NLRP3 and NLRC4 [30,32,33,[41][42][43][44][45][46]. Our results confirm that ASC is essential to NLRC4 puncta formation formed in the absence of caspase-1/11. Next, we used this NLRC4-GFP system to identify additional components of the NLRC4 inflammasome that operates in the absence of caspase-1/11. Non-inflammatory caspases have been previously shown to participate in the assembly of inflammasomes and to interact with ASC, including caspase-3, caspase-7 and caspase-8 [32][33][34]37,44,[46][47][48][49][50][51]. Thus, we transduced Casp1/11 -/macrophages with a retrovirus encoding NLRC4-GFP and evaluated the colocalization of NLRC4 with these caspases. In this experiment, we used the pan-caspase inhibitor Z-VAD to block caspase activation and to visualize puncta formation. We did not detect significant numbers of NLRC4 or ASC puncta containing caspase-3 and caspase-7 (S2 Fig). In contrast, caspase-8 and ASC was present in more than 90% of the NLRC4 puncta ( Fig 1C and S2 Fig). These data are in agreement with our findings indicating that ASC is required for NLRC4 puncta formation, accordingly, endogenous ASC colocalizes with NLRC4 and caspase-8 in the same puncta ( Fig 1D). To evaluate the participation of caspase-8 in the NLRC4 inflammasome, we transduced BMDMs from Casp1/11 -/mice with a retrovirus encoding ASC fused to GFP (ASC-GFP) and analyzed ASC puncta colocalization with caspase-8. We found that ASC puncta formed readily after the infection and that this process occurred in response to WT Lp and fliIbut not flaA - (Fig 2A). After 8 hours of infection, the formation of ASC puncta was partially dependent on flagellin (Fig 2A). We stained caspase-8 in macrophages transduced with retrovirus encoding ASC-GFP and found that caspase-8 colocalized with ASC puncta in response to infection with flagellated bacteria (Fig 2B). Collectively, these results indicate that flagellin triggers the assembly of an inflammasome composed of NLRC4 and ASC, which colocalizes with caspase-8.
The double-stranded DNA sensor AIM2 is known to recruit ASC to trigger puncta formation in response to infection, leading to caspase-1 activation and IL-1β and IL-18 release [31,[52][53][54][55]. The role of AIM2 inflammasome in the recognition of L. pneumophila has been demonstrated using sdhAdeficient bacteria. In the absence of SdhA, bacteria do not maintain vacuole integrity and localize in the macrophage cytoplasm, triggering activation of the AIM2 inflammasome [56,57]. In addition, the AIM2 inflammasome has been shown to trigger caspase-8 activation independently of caspase-1 [33,34,58]. Thus, we investigated whether AIM2 is present in the NLRC4/ASC/caspase-8 inflammasome that is formed in response to flagellinpositive L. pneumophila. We stained AIM2 in macrophages transduced with retrovirus encoding NLRC4-GFP and found no AIM2 in the NLRC4 puncta (S3A Fig). Moreover, we generated Aim2/Casp1/11 -/mice and found that AIM2 was dispensable for the formation of NLRC4 puncta in response to flagellin-positive bacteria (S3B Fig). The NLRC4/ASC/caspase-8 inflammasome accounts for flagellindependent restriction of Legionella replication in macrophages in the absence of caspase-1/11 Our data are consistent with the hypothesis that caspase-8 is a part of the inflammasome composed of NLRC4 and ASC. Thus, we investigated whether caspase-8 is activated during infection. We found that caspase-8 was strongly activated in Casp1/11 -/-BMDMs in response to fliIbut not flaAbacteria ( Fig 3A). In agreement with the requirement of ASC for the assembly of the NLRC4/ASC/caspase-8 inflammasome, we found that caspase-8 activation was abolished in Asc/Casp1/11 -/cells ( Fig 3A). Caspase-8 activation occurred normally in Aim2/Casp1/11 -/cells, indicating that AIM2 was not involved in the activation of caspase-8 through the flagellin/NLRC4/ASC inflammasome (S4 Fig). We also evaluated caspase-8 activation by western blot analysis by measuring the cleavage of p55 and the production of p18 isoforms. We found that flagellated bacteria triggered caspase-8 activation in Casp1/11 -/but not in Asc/ Casp1/11 -/cells. This phenomenon was evident by the reduction in p55 and increased production of p18 in Casp1/11 -/-BMDMs infected with fliIbut not flaAbacteria ( Fig 3B).
Next, we evaluated the participation of caspase-8 in caspase-1/11-independent restriction of L. pneumophila replication in macrophages, a process that was dependent on flagellin and NLRC4. Endogenous caspase-8 was silenced in Casp1/11 -/-BMDMs using two independent retrovirus encoding shRNA to target caspase-8. A non-target sequence was used as a control (NT). By western blotting, we detected reduced caspase-8 expression in Casp1/11 -/transduced cells. The shRNA Casp8 Seq1 was more efficient than Seq2 for silencing caspase-8 as determined by western blot (Fig 4A and S5 Fig) and RT-PCR ( Fig 4B). Importantly, complete silencing of caspase-8 cannot be achieved because caspase-8 expression is required for macrophage survival [59,60]. Nonetheless, using the described silencing conditions, we did not detect signs of cell death or LDH in the supernatant of the transduced macrophages. To evaluate the efficiency of caspase-8 silencing, we quantified caspase-8-containing puncta formation and caspase-8 activation in macrophages infected with flagellated L. pneumophila. We found that the frequency of puncta containing caspase-8 and caspase-8 activation was reduced in caspase-  (Fig 4C and 4D). Next, we evaluated the effect of caspase-8 for the restriction of L. pneumophila replication in Casp1/11 -/-BMDMs. We found that silencing caspase-8 culminated in increased replication of fliIbut not flaAbacteria (Fig 4E and 4F). These data indicated that caspase-8 contributed to the restriction of bacterial replication in a process that was dependent on flagellin, supporting the hypothesis that caspase-8 functionally participates in responses that are NLRC4/ASC-dependent and caspase-1/11-independent.
Activation of the NLRC4/ASC/caspase-8 inflammasome triggers pore formation and cell death Activation of caspase-1 inflammasomes induces pyroptosis and contributes to the restriction of infection by flagellated bacteria such as L. pneumophila, Salmonella typhimurium and Burkholderia thailandensis [23]. Accordingly, we have previously demonstrated that L. pneumophila trigger pyroptosis in a process mediated by caspase-1 and caspase-11, which are activated in response to flagellin and LPS, respectively [20,24,35,61]. Thus, we investigated whether activation of the NLRC4/ASC/caspase-8 inflammasome could trigger cell death in Casp1/11 -/macrophages. We assessed membrane permeabilization fluorometrically in real time via the uptake of propidium iodide. Macrophages were infected with fliIor flaA -, and pore formation was monitored in real time for 6 hours. C57BL/6 macrophages triggered robust pore formation in response to infection with fliIand reduced pore formation in response to flaA -( Fig  7A-7C). The pore formation observed in C57BL/6 macrophages infected with flaAmutants was dependent on caspase-11 [61] and will not be addressed herein. Importantly, despite the absence of both caspase-1 and caspase-11, we detected significant pore formation in Casp1/ 11 -/cells infected with fliI - (Fig 7B). This response was not detected in Casp1/11 -/cells infected with flaAor in Asc/Casp1/11 -/macrophages infected with fliIor flaA - (Fig 7B and 7C). These data support the hypothesis that NLRC4/ASC/caspase-8 induces pore formation. Experiments Caspase-8 activation in the Naip5/NLRC4/ASC inflammasome performed using Aim2/Casp1/11 -/macrophages corroborate our previous findings, indicating that AIM2 is not required for the activities of the NLRC4/ASC/caspase-8 inflammasome ( Fig  7B and 7C). Pore formation induced in response to caspase-1 and caspase-11 activation Caspase-8 activation in the Naip5/NLRC4/ASC inflammasome culminates with the induction of macrophage lysis, a process that can be assessed by the presence of LDH in tissue culture supernatants [24,61,62]. Thus, we measured LDH release in the supernatants of macrophages infected with fliIand flaAfor 8 hours. By comparing Casp1/ 11 -/and Asc/Casp1/11 -/cells, we found that macrophage lysis occurred despite the absence of caspase-1/11 (Fig 7D). Cell death was flagellin-dependent because infections with fliIbut not flaAinduced LDH release (Fig 7D). The participation of caspase-8 in pore formation and cell death induced in caspase-1/11-deficient macrophages was evident using both MOI 5 (Fig 7) and MOI 10 ( S9 Fig). Importantly, cell death was not observed in Asc/Casp1/11 -/-, a feature that corroborates the pore formation studies and indicates that the NLRC4/ASC/caspase-8 inflammasome triggers pore formation and lysis of infected cells. We also assessed whether the NLRC4/ASC/caspase-8 inflammasome was important for the activation of inflammatory cytokines. We found that whereas C57BL/6 macrophages readily triggered the production of IL-1β after 24 hours of infection with flagellated bacteria, the Casp1/11 -/or Asc/Casp1/11 -/deficient cells do not trigger a IL-1β production (S10A Fig). The production of IL-12p40 by these cells confirmed that all macrophages were primed and could respond to L. pneumophila infection (S10B Fig). To evaluate the participation of caspase-8 in cell death induced by the NLRC4/ ASC/caspase-8 inflammasome, we silenced endogenous caspase-8 using shRNA. Macrophages that were transduced with retrovirus encoding shRNA did not exhibit pore formation before infection, indicating that transduction itself did not trigger cell death (Fig 7E and 7H). In contrast, pore formation was evident in Casp1/11 -/but not in Asc/Casp1/11 -/macrophages in response to fliIinfection (Fig 7F and 7I). Pore formation induced in Casp1/11 -/was diminished in caspase-8-silenced cells (Fig 7F). In support of the role of flagellin for triggering these responses, we did not detect pore formation in cells infected with flaA - (Fig 7G and 7J). To further evaluate the participation of caspase-8 in pore formation induced by flagellin, we performed pore formation assays using Z-IETD, a cell permeable peptide that binds irreversibly to the catalytic site of caspase-8 [63][64][65]. We found that treatment of Casp1/11 -/macrophages with DMSO or Z-IETD did not cause pore formation in uninfected cells (Fig 7K). However, Z-IETD treatment reduced the pore formation induced by fliI - (Fig 7L) but not by flaA - (Fig  7M). Collectively, these data indicate that flagellin-positive bacteria trigger pore formation and cell death-independent of caspase-1/11 via a process that requires ASC and caspase-8.
Naip5 is required for NLRC4/ASC/caspase-8 inflammasome activation in response to flagellated Legionella Our data reveal that the NLRC4/ASC/caspase-8 inflammasome is activated in Casp1/11 -/macrophages in response to infection with flagellated Legionella. To evaluate if Naip5 is required for activation of this inflammasome, we used shRNA to silence endogenous Naip5. In Naip5 silenced Casp1/11 -/macrophages, we detected a reduced activation of caspase-8 in response to WT Lp and fliIbacteria (Fig 8A). Naip5 silencing was confirmed by RT-PCR (Fig 8B). We also tested if Naip5 is important for pore formation induced via caspase-8 in Casp1/11 -/macrophages. By evaluating pore formation, we found that Naip5 is important for efficient pore formation in response to WT Lp and fliIbacteria (Fig 8D-8E). As previously reported, no pore formation was detected in response to flaAbacteria (Fig 8F) or in Asc/Casp1/11 -/macrophages (Fig 8G-8J). Finally, we tested if Naip5 is important for restriction of L. pneumophila replication in Casp1/11 -/macrophages. We found that Naip5 is important for restriction of flagellin-positive L. pneumophila replication in Casp1/11 -/macrophages (Fig 8K). Naip5 silencing did not affect the replication of flaAbacteria in Casp1/11 -/macrophages ( Fig 8L). As predicted, Asc/Casp1/11 -/macrophages were permissive to replication of both flaAand fliIbacteria and Naip5 did not influenced this process (Fig 8M and 8N). Taken together, these Caspase-8 activation in the Naip5/NLRC4/ASC inflammasome data indicates that Naip5 participate of the NLRC4/ASC inflammasome that trigger caspase-8 activation in the absence of caspase-1/11. Caspase-8 is recruited to the NLRC4/ASC/caspase-1 inflammasome but it is only activated when caspase-1 or gasdermin-D is suppressed In the experiments shown thus far, we used Casp1/11 -/macrophage as a tool to assess the caspase-8 effects without the interference of caspase-1. However, because caspase-1 is present in natural conditions, we tested if caspase-8 participates in the NLRC4/ASC inflammasome in Caspase-8 activation in the Naip5/NLRC4/ASC inflammasome the presence of caspase-1. First, we infected C57BL/6 macrophages with flagellin-positive L. pneumophila to assess if endogenous caspase-8 colocalizes with the Naip5/NLRC4/ASC inflammasome. In uninfected conditions, we detected no significant puncta formation ( Fig  9A). However, in response to fliIbacteria, caspase-8 colocalizes with ASC ( Fig 9B) and caspase-1 (Fig 9C). We determined that caspase-8 is present in more than 60% puncta containing caspase-1 (Fig 9C). These data indicates that regardless to the presence of caspase-1, the caspase-8 is recruited to the inflammasome during activation.
Caspase-1 activation via the NLRC4 inflammasome is known to trigger activation of gasdermin-D (GSDMD) to induce cell death [67,68]. Thus, we tested if inhibition of GSDMD is sufficient to enable caspase-8 activation in the presence of caspase-1. To achieve this, we inhibited endogenous GSDMD using shRNA and found that despite the presence of caspase-1, caspase-8 is robustly activated when GSDMD is inhibited (Fig 9G). RT-PCR was used to confirm the silencing of the two different sequences of shRNA used (Fig 9H). To further confirm the GSDMD silencing, we performed pore formation assay in C57BL/6 macrophages infected with L. pneumophila. We found that GSDMD silencing inhibited the caspase-1-mediated pore formation induced by WT Lp and fliI -L. pneumophila (S11A-S11D Fig). However, GSDMD did not participate of pore formation induced via caspase-8 that occurs in Casp1/11 -/macrophages (S11E-S11H Fig). Collectively, these data indicates that despite the presence of caspase-1, caspase-8 activation occur in the Naip5/NLRC4/ASC inflammasome when GSDMD is inhibited.

Discussion
The recognition of Legionella flagellin by the Naip5/NLRC4 inflammasome in macrophages is a major mechanism for the restriction of bacterial replication in mouse cells [17][18][19][20]22,69,70]. It is well accepted in the field that not all NLRC4 functions require caspase-1 [29,38,71,72]. This conclusion is supported by the observation that Nlrc4 -/mice (and their macrophages) are significantly more permissive to L. pneumophila replication than Casp1/11 -/mice [29,38]. Here, we unraveled this caspase-1/11-independent pathway and characterized an inflammasome composed of Naip5, NLRC4, ASC and caspase-8, which operates in the absence of caspase-1 and caspase-11. This inflammasome effectively participates in the mechanisms involved in the restriction of bacterial replication in macrophages and in vivo. Previous biochemistry studies using yeast two-hybrid screening showed that NLRC4 is ubiquitinated by Sug1, a process that facilitates the activation of caspase-8 [51]. Thus, it is possible that Sug1 is also a component of this NLRC4 inflammasome. In addition, previous studies using the Salmonella Caspase-8 activation in the Naip5/NLRC4/ASC inflammasome enterica serovar Typhimurium have indicated that both caspase-8 and caspase-1 are recruited to the ASC puncta in response to infection. However, caspase-8 is involved in the synthesis of pro-IL-1β and is dispensable for Salmonella-induced cell death [32]. These data contrast with published data using L. pneumophila, which indicate that in the absence of caspase-1/11, no inflammatory cytokines are produced [20,21,25,30,35,66]. Accordingly, our data indicates that this Naip5/NLRC4/ASC/Caspase-8 inflammasome is very inefficient to trigger IL-1β maturation when caspase-1 is not present.
Importantly, AIM2 is not part of this NLRC4/ASC/caspase-8 inflammasome. AIM2 is wellknown to trigger caspase-8 activation via ASC [33,34,58]. Our data unequivocally demonstrate that AIM2 is neither a component of this inflammasome, nor is it required for inflammasome functions. AIM2 did not colocalize with the NLRC4/ASC/caspase-8 puncta, it was dispensable for the activation of caspase-8 in response to flagellated L. pneumophila and for the induction of cell death and restriction of L. pneumophila replication. The pyrin domain of ASC can bind to the death domain of caspase-8 [46]. Thus, it is possible that the interactions of the ASC pyrin domain with the caspase-8 dead domain are critical for the recruitment of caspase-8 to the complex. Consistent with this hypothesis, our studies unequivocally show that ASC is required for the assembly and function of this NLRC4 inflammasome. Importantly, the characterization of this NLRC4/ASC/caspase-8 inflammasome accounted to clarify a controversy in the field concerning the participation of ASC in the NLRC4 inflammasome. Studies utilizing biochemistry and cells from gene-deficient mice have demonstrated that ASC is dispensable for NLRC4 functions, including pyroptosis and the restriction of L. pneumophila replication [20,21,29]. However, ASC is essential for caspase-1 cleavage and the processing of inflammatory cytokines in response to flagellated L. pneumophila, a process that is NLRC4-dependent and NLRP3-independent [20,21,25,30]. In addition, ASC participates in the restriction of intracellular replication of L. pneumophila under certain circumstances [36,37]. Our studies provide data that help to consolidate these data in a cohesive model. NLRC4 can operate to form an inflammasome in absence of ASC that triggers pore formation and the restriction of bacterial replication [20,21,25,30]. This platform does not form puncta and is ineffective for triggering caspase-1 cleavage and processing inflammatory cytokines. When ASC is present, NLRC4 inflammasome associates with ASC and recruit caspase-1 and caspase-8 to the puncta. This inflammasome is very efficient to cleave caspase-1 and inflammatory cytokines such as IL-1β. Interestingly, our data indicate that caspase-8 is not activated when caspase-1 is present. However, when caspase-1 is missing or when Gasdermin-D is inhibited, we detected a robust caspase-8 activation. These data suggest that activation of caspase-8 in the Naip5/NLRC4/ASC inflammasome functions as a backup strategy to guarantee cell death when key components of the pyroptotic cell death are inhibited. Interestingly, our data and previously published data indicate that gasdermin-D is dispensable for caspase-8-induced cell death [73]. Therefore, when gasdermin-D is inhibited, caspase-8 engages gasdermin-D-independent cell death. It is possible that caspase-8 induces caspase-3 and caspase-7 to targed gasdermin-E (also known as DFNA5) to induce pore formation and cell death independent of gasdermin-D [74,75]. This may guarantee appropriate responses to pathogens that inhibit canonical components of pyroptotic cell death such as caspase-1 or gasdermin-D.

In vitro infections and CFU determination
For the in vitro infections, the cultures were infected at a multiplicity of infection of 0.015, 5 or 10 followed by centrifugation for 5 minutes at 300 X g at room temperature and incubation at 37˚C in a 5% CO 2 atmosphere. In the colony-forming units (CFU) experiments, cultures infected at a MOI of 10 were washed two times with PBS, and 1 ml of medium was added to each well. For CFU determination, the cultures were lysed in sterile water, and the cell lysates were combined with the cell culture supernatant from the respective wells. Lysates plus supernatants from each well were diluted in water, plated on CYE agar plates, and incubated for 4 days at 37˚C for CFU determination as described previously [28,38].

Caspase-8 activation
To assess caspase-8 activation, BMDMs were infected with fliIor flaAfor 8 hours, and the activity of caspase-8 was measured using the Caspase-8 Glo Assay (Promega) according to manufacturer´s recommendations. To evaluate caspase-8 activation by western blot analysis, 1 X 10 7 cells were infected at a MOI of 10 and lysed 8 hours after infection in RIPA buffer with protease inhibitor, as previously described. The lysates were suspended in 4X Laemmli buffer, boiled for 5 minutes, resolved by 15% SDS PAGE and transferred to 0.22-μm nitrocellulose membranes. The membranes were blocked in Tris-buffered saline (TBS) with 0.01% Tween-20 and 5% non-fat dry milk or for 1 hour. The mouse anti-caspase-8 (Enzo-1G12) and antirabbit peroxidase-conjugated antibody (KPL; 1:3000) were diluted in blocking buffer for the incubations (overnight for anti-cleaved caspase-8 and 1 hour for the secondary antibody). ECL luminol reagent (GE Healthcare) was used for antibody detection.

Real Time q-PCR
Total RNA was extracted from 2 X 10 6 macrophages using total RNA isolation kit (illustra RNAspin, GE Healthcare, UK), according to manufacturer's instructions. After extraction, an aliquot of 2 μl was used to determine the RNA concentration in NanoDrop (Thermo Fisher Scientific) and 1 μg of the extracted RNA was used for the cDNA conversion using the iScriptTM cDNA Synthesis kit (BIO-RAD) in a thermal cycler. The cDNA (10 ng) was used for the quantification of the Caspase 8 gene expression (TaqMan Assay: Casp 8-Mm01255716_m1) by realtime PCR using TaqMan Fast Advanced Master Mix, according to the manufacturer's instruction (Applied Biosystems). Actin beta (Actb) gene (TaqMan Assay: Actb-Mm00607939_s1) was used as a control for normalization of expression levels. The quantification of GSDMD and Naip5 were performed using 25 ng of cDNA and 10 μM of each primer, 1X SYBR Green (Applied Biosystems), and was normalized using the housekeeping gene glyceraldehyde-3phosphate dehydrogenase (Gapdh). The specificity of the PCR products was assessed by melting curve analysis for all samples. The following primers were used: Gapdh, FWD-AGGTCGGTG TGAACGGATTTG, REV-TGTAGACCATGTAGTTGAGGTCA, GSDMD, FWD-TCATGT GTCAACCTGTCAATCAAGGACAT, REV-CATCGACGACATCAGAGACTTTGAAGGA, NAIP5, FWD-GTTGAGATTGGAGAAGACCTCG, REV-CACACGTGAAAGCAACCATGG. The real-time quantitative reaction was performed in the Viia 7 Real-Time PCR System (Applied Biosystems). The results were analyzed using the 2 -ΔΔCT method and are expressed in relation to the reference group. The percentage of silencing knockdown was estimated using the [1][2] x 100] equation [78].

Mice and in vivo infections
Mice used in this study were breed and maintained in institutional animal facilities. Mice used were C57BL/6 (Jax 000664), Nlrc4 -/- [79], Casp1/11 -/- [80], Asc -/- [81], Casp11 -/and Casp1 -/-Casp11 tg [66] were. Double-deficient mice were generated by intercrossing a F1 progeny of the parental strains. All mice were matched by sex and age (all were at least 8 weeks old at the time of infection) and were in a C57BL/6 mouse genetic background. For the in vivo experiments, approximately 5-7 mice per group were used, as indicated in the figures. For in vivo infections, the mice were anesthetized with ketamine and xylazine (50 mg/kg and 10 mg/kg, respectively) by intraperitoneal injection followed by intranasal inoculation with 40 μl of phosphate-buffered saline (PBS) containing 1 X 10 5 bacteria per mouse. For CFU determination, the lungs were harvested and homogenized in 5 ml of sterile water for 30 seconds using a tissue homogenizer (Power Gen 125; Thermo Scientific). Lung homogenates were diluted in sterile water and plated on CYE agar plates for CFU determination as previously described [28,38].

Pore formation assay
Pore formation in BMDMs was quantified based on the permeability to propidium iodide (PI) in damaged cells as previously described [61]. BMDMs were seeded in a black, clear-bottom 96-well plate (1 X 10 5 cells/well). Before infection, the medium was replaced with 10% RPMI without phenol red, 0.038 g/ml NaHCO 3 , 6 μl/ml PI and anti-L. pneumophila (1:1000). Infected BMDMs were maintained at 37˚C, and PI was excited at 538 nm. The fluorescence emission was read at 617 nm every 5 minutes using a plate fluorometer (SpectraMax i3x, Molecular Devices). Total pore formation was determined by lysing cells with Triton X-100.

Lactate dehydrogenase release assay and ELISA
BMDMs were seeded in 24-well plates (5 X 10 5 cells/well). Infections were performed in RPMI1640 medium without phenol red, 15 mM HEPES and 2 g/l NaHCO 3 supplemented with 10% FBS. After 8 hours of infection, the supernatants were collected for analysis of lactate dehydrogenase (LDH) release. Total LDH was determined by lysing the cultures with Triton X-100. LDH was quantified using the CytoTox 96 LDH-release kit (Promega). For cytokine determination, enzyme-linked immunosorbent assay (ELISA) were used. BMDMs were seeded into 24-well plates (5 X 10 5 cells/well) and infected with WT Lp, fliIand flaA -(MOI 10) for 24 hours. BMDMs supernatant was assessed using ELISA kits according to manufac-turer´s recommendations (BD Biosciences).

Ethics statement
The care of the mice was in compliance with the institutional guidelines on ethics in animal experiments; approved by CETEA (Comissão de Ética em Experimentação Animal da Faculdade de Medicina de Ribeirão Preto, approved protocol number 218/2014). CETEA follow the Brazilian national guidelines recommended by CONCEA (Conselho Nacional de Controle em Experimentação o Animal). For euthanasia, the mice were treated with ketamine and xylazine (50 mg/kg and 10 mg/kg, respectively) by intravenous injection.

Statistical analysis
The data were plotted and analyzed using GraphPad Prism 5.0 software. The statistical significance was calculated using the Student's t-test or analysis of variance (ANOVA). Nonparametric test Mann-Whitney U test were used for analysis of in vivo experiments. Differences were considered statistically significant when P was <0.05, as indicated by an asterisk in the figures.  A-B) The percentage of colocalization of caspase-3, caspase-7 and caspase-8 with NLRC4-GFP and ASC-GFP was determined using an epifluorescence microscope. (C-E) BMDMs generated from Casp1/11 -/mice were infected with fliIat a MOI of 10 for 8 h. The cultures were fixed and stained with anti-ASC (green), anti-caspase-3 (red) (C), anti-caspase-7 (red) (D), anti-caspase-8 (red) (E). Cell nuclei were stained with DAPI (cyan). Images were acquired by multiphoton microscopy with a 63x oil immersion objective and analyzed using ImageJ software. The images are the maximal projection of a z project. Scale bar, 10μm. Data show the average ± SD of triplicate wells. Ã , P<0.05, Student´s t test. Data are presented for one representative experiment of two experiments with similar results. (TIF) S3 Fig. AIM2 does not colocalize and is not required for NLRC4-GFP puncta formation. Bone marrow-derived macrophages (BMDMs) generated from Casp1/11 -/and Aim2/Casp1/ 11 -/mice were transduced with retrovirus encoding NLRC4-GFP and infected with wild-type L. pneumophila (WT) or with motility-deficient L. pneumophila mutants expressing flagellin (fliI -) at a MOI of 10 for 8 h. (A) The cultures were stained with anti-AIM2 (red), the cell nuclei were stained with DAPI (cyan) and the NLRC4-GFP puncta is shown in green. The percentage of colocalization of AIM2 with NLRC4-GFP is shown. Images were acquired by multiphoton microscopy with a 63x oil immersion objective and analyzed using ImageJ software.  Fig 4A. Bone marrow-derived macrophages (BMDMs) generated from Casp1/11 -/mice were transduced with a retrovirus encoding shRNA sequences to target caspase-8 (Seq1, Seq2) and a non-target shRNA sequence (NT). The silencing was confirmed by western blot analysis (Fig 4A). Cell lysates were separated by SDS-PAGE, blotted and probed with anti-caspase-8 (pro-caspase-8 p55) and anti-α-actin. Immunoblots were analyzed in Image J software and the caspase-8 p55 to α-actin ratio is shown.  Fig 5E. Bone marrow-derived macrophages (BMDMs) generated from Casp1/11 -/and Asc/Casp1/11 -/mice were transduced with a retrovirus encoding shRNA sequences to target caspase-8 (Seq1, Seq2) and a non-target shRNA sequence (NT). The silencing was confirmed by western blot analysis (Fig 5E). Cell lysates were separated by SDS-PAGE, blotted and probed with anti-caspase-8 (pro-caspase-8 p55) and anti-α-actin. Immunoblots were analyzed in Image J software and the caspase-8 p55 to α-actin ratio is shown.