Structure-function analysis of Nlrp1b reveals a link. between metabolism and inflammation: energy stress. triggers inflammasome activation

Transcription

1 Structure-function analysis of Nlrp1b reveals a link between metabolism and inflammation: energy stress triggers inflammasome activation by Kuo-Chieh Liao A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Laboratory Medicine and Pathobiology University of Toronto Copyright by Kuo-Chieh Liao (2012)

2 Structure-function analysis of Nlrp1b reveals a link between metabolism and inflammation: energy stress triggers inflammasome activation Kuo-Chieh Liao Doctor of Philosophy Department of Laboratory Medicine and Pathobiology University of Toronto 2012 Abstract Immune cells have evolved an array of pattern recognition receptors (PRRs) to initiate immune responses upon detecting pathogen associated molecular patterns (PAMPs) and danger associated molecular patterns (DAMPs). Nlrp1b, a cytosolic PRR, forms a protein complex termed an inflammasome that activates pro-caspase-1, which leads to processing of inflammatory cytokines, such as pro-il-1β, and to cell death. Nlrp1b has been shown to mediate death of murine macrophages by anthrax lethal toxin. Lethal toxin is a binary toxin that comprises lethal factor (LF), a zinc metalloprotease, and protective antigen (PA), which translocates LF into the cytosol. In this thesis, the murine Nlrp1b inflammasome was reconstituted in a human fibroblast cell line and shown to be responsive to the enzymatic activity of LF. ii

3 Activation of Nlrp1b by LF was allele specific in the reconstituted system since an allele derived from macrophages that are killed by the toxin facilitated processing of pro-caspase-1, whereas an allele from macrophages that are resistant to the toxin did not mediate processing of pro-caspase-1. That the reconstituted system recapitulates the cellular events in intoxicated macrophages was further supported by the observation that a proteasome inhibitor that inhibits macrophage intoxication also prevented pro-caspase-1 activation in the fibroblasts. Next, I identified energy stress as a danger signal that activates the Nlrp1b inflammasome: reduction of intracellular ATP by 2-deoxy-glucose and sodium azide caused Nlrp1b-dependent processing of pro-il-1β. AMP-activated protein kinase (AMPK), a major energy sensor in cells, facilitated Nlrp1b activation after ATP-depletion whereas it was not involved in lethal toxin-stimulated Nlrp1b signaling. However, signals derived from lethal toxin and ATP-depletion seemed to converge upstream of Nlrp1b activation because pharmacological inhibition of the proteasome blocked both triggers. Lastly, mutations that disrupted the nucleotide-binding site of Nlrp1b caused Nlrp1b activation, which is consistent with the finding that Nlrp1b was activated when the cytosolic ATP level decreased. iii

4 Acknowledgements It is my pleasure to thank all the people who have supported and inspired me during my doctoral study. First and foremost, I would like to thank my supervisor Dr. Jeremy Mogridge. If not for his guidance, patience, and open-mindness in discussing science, the work presented in this thesis would not be possible. I am grateful for this enriching experience in my development as a PhD student. I am also appreciative of his financial support during the past 6 years. In addition to Jeremy, I would like to extend my gratitude to past and present members in Jeremy s laboratory: Sarah, Mandy, Vincent, Mia, Bradley, Vineet, Jana, Kristopher, and Melissa. I would especially thank to Sarah and Vincent for technical help and advice and Vineet and Bradley for their helpful discussion about my projects. My sincere thanks to members of my advisory committee: Dr. Stephen Girardin and Dr. Jun Liu for their time in reviewing my work and providing constructive criticism. I also greatly appreciate the support from my parents and friends during my graduate studies. iv

5 Table of contents 1.0 Introduction The NLR family Inflammasomes: platforms of pro-caspase-1 activation, pro-il-1β and pro-il-18 processing, and pyroptosis induction Pro-caspase-1 activation Pro-IL-1β and pro-il-18 processing Pyroptosis induction AIM2 inflammasome AIM2 recognizes cytosolic double-strand DNA AIM2 inflammasome is activated upon detection of two bacterial pathogens, Francisella tularensis and Listeria monocytogenes NLRC4 inflammasome Bacterial flagellins and secretion system components activate the NLRC4 inflammasome Domain structure of NLRC4 and its activation mechanism Role of NLRC4 inflammasome in vivo NLRP3 inflammasome Domain structure of NLRP Microbial Sensing by NLRP3 inflammasome Non-microbial sensing by NLRP3 inflammasome Molecular mechanism of NLRP3 activation NLRP3-associated genetic disorders NLRP1 inflammasome Human NLRP Domain structure of murine Nlrp1b Nlrp1b controls murine macrophage susceptibility to anthrax lethal toxin Role of Nlrp1b inflammasome in vivo Links between inflammation and metabolism Thesis Rationale Material and methods Cell culture and reagents Plasmid construction, cdna cloning, and site-directed mutagenesis IL-1β and ATP level assays LDH release assays Detection of TAP-tagged proteins GST fusion protein purification and in vitro binding assay v

6 2.7 Co-immunoprecipitation assay Knock down of endogenous AMPK Expression of Nlrp1b inflammasome components in human fibroblasts confers susceptibility to anthrax lethal toxin Summary Results Activation of the Nlrp1b inflammasome in HT1080 cells Characterization of LeTx-induced cell death and IL-1β release The CARD domain, but not the NACHT or LRR domains, is required for inflammasome activity Amino-terminal truncation mutants of Nlrp1b are constitutively active Nlrp1b and Nlrp1b interact with pro-caspase-1, but only Nlrp1b self-associates Discussion Energy stress activates the NLRP1b inflammasome Summary Results ATP depletion activates Nlrp1b AMPK promotes inflammasome activation Inhibition of the proteasome impairs inflammasome activation Mutation of the Nlrp1b Walker A motif causes constitutive activation Discussion Discussion Autoproteolysis within the FIIND domain is important for Nlrp1b activity H 2 O 2 activates Nlrp1b allele 1, allele 3 but not allele LeTx and ATP depletion are two distinct signals that activate Nlrp1b NLRP3 and Nlrp1b inflammasomes Future directions Endogenous Nlrp1b in macrophages Identification of the NACHT domain-associated proteins Human NLRP1 inflammasome References vi

7 List of abbreviations 2DG 2-deoxy-glucose AIM2 absent in melanoma 2 AMPK AMP-activated protein kinase APAF-1 apoptotic protease activating factor 1 ASC apoptosis-related speck-like protein ATP adenosine triphosphate CAPS cryopyrin-associated periodic syndromes CARD caspase recruitment domain CINCA chronic infantile neurologic cutaneous and articular syndrome CLR C-type lectin receptor CTP cytidine triphosphate DAMP danger associated molecular pattern dsdna double-stranded DNA FACS familial cold autoinflammatory syndrome FIIND function to find domain GST glutathione S-transferase GTP guanosine triphosphate IAPP islet amyloid polypeptide LDH lactate dehydrogenase LeTx anthrax lethal toxin LF lethal factor LLO listeriolysin O LRR leucine rich repeat mcmv murine cytomegalovirus MDP muramyl dipeptide MSU monosodium urate MWS Muckle-Wells syndrome NAC N-acetyl-cysteine NACHT NAIP, CIITA, HET-E and TP1 NAIP NLR family, apoptosis inhibitory protein NaN 3 sodium azide NLR Nod-like receptor NLRC Nod-like receptor with a CARD domain vii

8 NLRP NO NOMID NTPase PA PAMP PRR PYD RLR ROS SLE STAND STS TAP TLR TUNEL VDAC UTP Nod-like receptor with a PYD domain nitric oxide neonatal-onset multisystem inflammatory disease nucleoside triphosphatase protective antigen pathogen associated molecular pattern pattern recognition receptor pyrin RIG-I-like receptor reactive oxygen species systemic lupus erythematous signal transduction ATPases with numerous domain staurosporine tandem affinity purification Toll-like receptor terminal deoxynucleotidyl transferase dutp nick end labeling voltage-dependent anion channel uridine triphosphate viii

9 List of Figures Figure 1.1 Inflammasomes... 4 Figure 1.2 Activation model of Nlrp1b inflammasome Figure 3.1 Reconstitution of the Nlrp1b inflammasome in HT1080 cells Figure 3.2 Characterization of LeTx-induced cell death and IL-1β release Figure 3.3 Deletion analysis of Nlrp1b Figure 3.4 Amino-terminal truncation mutants of Nlrp1b are constitutively active Figure 3.5 Nlrp1b and Nlrp1b interact with pro-caspase-1, but only Nlrp1b self-associates Figure 4.1 ATP depletion activates the Nlrp1b inflammasome Figure 4.2 AMPK facilitates the Nlrp1b inflammasome activation Figure 4.3 Role of reactive oxygen species and proteasome activity in ATP-depletion induced Nlrp1b activation Figure 4.4 Walker A motif mutant of Nlrp1b is constitutively active Figure 5.1 The FIIND domain interacts with the NACHT domain Figure 5.2 H 2 O 2 activates Nlrp1b allele 1 and allele 3 but not allele ix

10 CHAPTER Introduction Innate immune cells are constantly being engaged by a wide range of pathogenic and non-pathogenic microbes. In order to mount immediate immune responses, these cells, such as macrophages, have evolved an array of germline-encoded pattern-recognition receptors (PRRs) to recognize a limited number of well-conserved microbial signatures called pathogen-associated molecular patterns, or PAMPs. PRRs include two membrane bound receptor families and two cytosolic receptor families. Membrane-bound Toll-like receptors (TLRs) and C-type lectin receptors (CLRs) sense extracellular microbial products and, once activated, induce pro-inflammatory gene expression through different pathways including NF-κB signaling. Cytosolic RIG-I-like receptors (RLRs) induce an antiviral response when viral RNA is detected whereas the other cytosolic receptors, Nod-like receptors (NLRs), activate cytokine gene transcription or process cytokines post-translationally upon detection of microbial products (Benko et al., 2008; Martinon et al., 2009). 1.1 The NLR family The NLR family is composed of 22 genes in humans and at least 34 genes have been 1

11 identified in the mouse genome (Bauernfeind et al., 2011a). All NLRs share a tripartite structural organization consisting of a C-terminal leucine-rich repeat (LRR) domain, a central NACHT domain, and a N-terminal effector domain. The LRR domain is generally believed to be involved in ligand sensing and autoregulation of NLRs. This domain consists of several tandem repeats of a amino acid motif rich in the hydrophobic amino acid leucine. These repeats are organized in a way that contributes to the formation of a horseshoe-shaped domain, which makes the concave side suitable for protein-protein interactions (Bella et al., 2008). To date, it has been demonstrated that the LRR domain of some TLRs directly interacts with PAMPs. However, little evidence of direct binding to PAMPs has been shown for the NLRs. Interestingly, several LRR domain-deleted NLRs exhibit constitutive activity, suggesting that the LRR domain is involved in intramolecular auto-inhibition. This inhibitory interaction is likely to be released when the LRR domain detects ligands. The NACHT domain belongs to the recently defined STAND family of NTPases (Danot et al., 2009). One well-studied example of the STAND family of proteins is APAF-1. Upon mitochondria-derived cytochrome c detection, APAF-1 initiates an ATP-dependent protein complex assembly process. This ring-like protein complex, termed the apoptosome, triggers caspase-9 activation, inducing the apoptotic cell 2

12 death pathway (Acehan et al., 2002). Likewise, the NACHT domain of NLRs mediates oligomerization to form a higher order protein complex. Furthermore, this self-association process depends on the nucleotide binding in the Walker A region and a Mg 2+ binding Walker B region within the NACHT domain. Mutations in these regions disrupt the NACHT-NACHT interaction (Duncan et al., 2007). NLRs differ in the N-terminal effector domains that are essential for activating downstream signaling pathways. The two most common domains are the PYD domain and the CARD domain. These domains are composed of six helices and have distinct surface patches that make them able to form multimers through homotypic interactions. In addition to the PYD domain and the CARD domain, other effector domains are also present in NLRs, which classify NLRs into five subfamilies: NLRA (NLRs with an acidic activation domain), NLRB (NLRs with a BIR domain), NLRC (NLRs with a CARD domain), NLRPs (NLRs with a PYD domain), and NLRX (NLRs with a N-terminal domain without strong homology to others). Among all NLRs mentioned above, some receptors, upon PAMPs detection, initiate protein complex assembly and, consequently, cause pro-inflammatory cytokine processing and secretion. In my thesis, I will focus on these protein complexes, termed the inflammasomes (Fig.1.1). 3

13 Figure 1.1 Inflammasomes. AIM2 and NLRP3 inflammasomes require ASC to recruit pro-caspase-1 for activation. The role of ASC in NLRC4 and NLRP1 inflammasome remains somewhat controversial. Some studies suggest that ASC is not an essential component but facilitates inflammasome activation. 1.2 Inflammasomes: platforms of pro-caspase-1 activation, pro-il-1β and pro-il-18 processing, and pyroptosis induction Pro-caspase-1 activation Caspase-1 is a member of the pro-inflammatory caspases, a subgroup of the caspase family that also includes human capase-4, human caspase-5, and murine caspase-11 4

14 (Li and Yuan, 2008; Martinon and Tschopp, 2007). Caspase-1 was first identified as a result of attempts to purify the enzyme responsible for the processing of pro-il-1β, a cytokine that needs to be cleaved to its mature form to induce immune responses (Wilson et al., 1994). Similar to other caspases, the proximity model for activation seems likely to be true for caspase-1 (Boatright et al., 2003). This 45 kda aspartate-specific cysteine protease is first synthesized as an inactive zymogen. Autoproteolysis occurs when molecules of pro-caspase-1 are brought into proximity, releasing the N-terminal CARD domain, a central p20 fragment and the C-terminal p10 fragment. Two p10 and two p20 fragments then form an active caspase-1 tetramer. Once activated, caspase-1 is able to process inflammatory cytokines and to trigger a cell death pathway. Interestingly, although the enzymatic activity of caspase-1 is essential for its function, proteolysis of pro-caspase-1 seems to be involved in cytokine processing but not in cell death induction. A recent report suggests that a caspase-1 mutant that was not able to undergo proteolysis failed to process inflammatory cytokines; however, this mutant was still capable of inducing cell death in response to bacterial infection (Broz et al., 2010). This finding suggests that there might be several distinct populations of caspase-1, having different cellular physiological functions and being tightly regulated by being restricted to different cellular compartments. 5

15 1.2.2 Pro-IL-1β and pro-il-18 processing Bioactive Interleukin-1β is best known as an endogenous pyrogen that induces fever as well as other immune responses, such as stimulating T cell activation and promoting infiltration of immune cells to infection sites for clearance of pathogens (Dinarello, 2009). Despite its beneficial effects of mounting immune responses, overproduction of IL-1β has been implicated in several auto-inflammatory diseases. Thus, its expression, processing, and secretion are tightly regulated to avoid detrimental side effects. First, IL-1β mrna level is low unless its transcription is induced by TLR, RLR, or CLR ligands, like LPS. Once the 33 kda pro-il-1β protein is synthesized, this biologically inactive proform needs to be cleaved at aspartate 117 by active caspase-1 to generate mature 17 kda IL-1β. This processing event requires inflammasome assembly upon ligand detection to activate caspase-1, which is tightly regulated as well. Lastly, active IL-1β needs to exit cells to fulfill its physiological functions. Since IL-1β lacks a secretion signal peptide, how this protein is secreted remains ill defined. Several models have been proposed, such as a requirement for phospholipase C and secretory lysosomes (Andrei et al., 2004), and the shedding of plasma membrane microvesicles or multivesicular bodies containing exosomes (MacKenzie et al., 2001; Qu et al., 2007). Of note, it is unlikely that IL-1β is passively released from membrane integrity-compromised cells because an 6

16 osomoprotectant that prevents cell lysis does not inhibit mature IL-1β secretion in inflammasome-activated cells (see Chapter 3). IL-18 was first discovered as an IFN-γ-inducing cytokine in endotoxemic mice (Puren et al., 1999). Unlike IL-1β, IL-18 does not require a priming step for its expression: it is constitutively expressed in cells. However, caspase-1 cleavage is still essential to process IL-18 from its 24 kda pro-form to 18 kda bioactive form Pyroptosis induction Cell death, including apoptosis and pyroptosis, has long been regarded as a host strategy to contain infection (Bergsbaken et al., 2009; Labbe and Saleh, 2008). For example, the engulfment of dying macrophages by dendritic cells additionally promotes antigen presentation to T cells. In addition, it is also possible that nutrient deprivation within dying cells may restrict bacterial replication and further prevent dissemination. Pyroptosis, by definition, is a cell death pathway that depends on caspase-1. This cell death pathway has been observed in macrophages infected by bacterial pathogens, such as Legionella pneumophila and Francisella tularensis (Fernandes-Alnemri et al., 2010; Zamboni et al., 2006). In addition to the requirement of caspase-1, pyroptosis and apoptosis differ in at least two other aspects. First, during pyroptosis, cell membrane integrity is partially 7

17 compromised. As a result, water flows in causing a measurable increase in cell size. Subsequently, cell rupture leads to the release of cellular contents, such as lactate dehydrogenase (LDH), the activity of which can be measured experimentally. Concomitantly, inflammatory cytokines are released from pyroptotic cells and cause inflammation. In contrast, initially apoptosis does not result in the release of cellular contents and fails to induce inflammation. Second, despite the fact that DNA damage can be both observed in pyroptotic cells and apoptotic cells, as demonstrated by TUNEL assay, the nucleus remains intact in pyroptotic cells but not in apoptotic cells (Bergsbaken et al., 2009) AIM2 inflammasome AIM2 recognizes cytosolic double-strand DNA AIM2 (absent in melanoma 2) was recently identified as a non-nlr inflammasome scaffold protein that detects cytosolic DNA (Burckstummer et al., 2009; Fernandes-Alnemri et al., 2009; Hornung et al., 2009; Roberts et al., 2009). It has an amino-terminal PYD domain and a carboxy-terminal HIN domain. The PYD domain allows AIM2 to recruit pro-caspase-1 through an adaptor molecule ASC 8

18 (apoptosis-related speck-like protein). Upon double-stranded DNA (dsdna) detection, AIM2 activates pro-caspase-1, inducing pro-il-1β processing and secretion, which has been demonstrated in a cell-based reconstitution system and in macrophages. In addition, a higher order protein complex was observed in the presence of high dsdna concentration in chemical crosslinking experiments, indicating the assembly of AIM2 inflammasome upon DNA recognition (Fernandes-Alnemri et al., 2009). Interestingly, DNA from viral origin was also able to trigger AIM2 inflammasome activation (Rathinam et al., 2010). AIM2 deficient macrophages failed to cause pro-il-1β processing and secretion upon viral infection, such as vaccinia virus and murine cytomegalovirus (mcmv). Furthermore, AIM2 has also been reported to respond to bacterial pathogens, potentially through directly interacting with their genetic material (Belhocine and Monack, 2012) AIM2 inflammasome is activated upon detection of two bacterial pathogens, Francisella tularensis and Listeria monocytogenes Francisella tularensis is the causative agent of tularemia, of which the pneumonic form can be lethal if untreated. After being phagocytosed by macrophages, these bacteria escape the phagosome prior to phagosome/lysosome fusion and subsequently 9

19 replicate in the cytosol. Recent reports suggested that the AIM2 inflammasome was activated in Francisella tularensis-infected macrophages because caspase-1 activation and IL-1β secretion were absent in AIM-2 deficient cells (Fernandes-Alnemri et al., 2010; Jones et al., 2010). Of note, in in vivo experiments, AIM2 deficient mice are more susceptible to Francisella tularensis infection than wild type mice. Moreover, bacterial burdens were higher in tissues from AIM2 deficient mice, compared with that from wild type mice. These results suggested that the AIM2 inflammasome is critical for the host to contain Francisella tularensis infection. As dsdna triggers AIM2 activation, it has been speculated that cytosolic AIM2 was stimulated by Francisella tularensis DNA and the genetic material was passively released during bacterolysis. In other words, bacterial mutants which have compromised structural integrity would more readily release DNA after they are engulfed by macrophages; as a result, these mutants would trigger elevated AIM2 inflammasome activation in macrophages. Indeed, Francisella tularensis mutants, which exhibited aberrant morphology during growth in minimal media, were more susceptible to lysis in macrophages and triggered elevated AIM2 inflammasome-mediated cytotoxicity (Peng et al., 2011). Consistent with this idea, AIM2 was shown to colocalize with Francisella tularensis DNA in macrophages by fluorescence microscopy (Jones et al., 2010). Collectively, these results suggested that 10

20 AIM2 was activated by Francisella tularensis DNA during bacterial infection. The second bacterial pathogen recognized by AIM2 is Listeria monocytogenes. To date, three inflammasomes have been reported to be activated by Listeria monocytogenes, possibly via different bacterial components or products (Warren et al., 2008). As expected, AIM2 detected Listeria monocytogenes DNA and triggered caspase-1 activation (Warren et al., 2010). Like Francisella tularensis, Listeria monocytogene mutants that are more susceptible to bacteriolysis caused elevated cell death and IL-1β secretion in an AIM2 dependent manner in macrophages (Sauer et al., 2010). In addition to microbial sensing, AIM2 has been implicated to be involved in systemic lupus erythematous (SLE). Many patients of this disease form immune complexes to dsdna by means of autoreactive antibodies (Sun et al., 2000). Furthermore, elevated IL-1β level is observed. Nonetheless, more studies are needed to further validate the role of AIM2 in SLE. 11

21 1.2.5 NLRC4 inflammasome Bacterial flagellins and secretion system components activate the NLRC4 inflammasome Several bacteria activate NLRC4, also known as IPAF or CLAN, including Legionella pneumophila (Amer et al., 2006; Lightfield et al., 2008), Salmonella typhimurium (Franchi et al., 2006; Miao et al., 2006), and Pseudomonas aeruginosa (Franchi et al., 2007; Miao et al., 2008). Legionella pneumophila is the causative agent of Legionnaire s disease. This bacterium facilitates the translocation of effector proteins into the cytosol of macrophages via the type IV Dot/Icm secretion system. By doing so, this bacterium is able to establish a specialized vacuole that does not fuse with lysosomes, and is therefore suitable for replication. Nonetheless, macrophages have evolved a way to restrict Legionella pneumophila replication. This anti-bacterial pathway is mediated through the NLRC4-caspase-1 axis. NLRC4 or caspase-1 deficient macrophages are more permissive to Legionella pneumophila than wild type macrophages as more bacteria were recovered from macrophages that do not express NLRC4 or caspase-1 (Lightfield et al., 2008). As expected, caspase-1 cleavage, IL-1β 12

22 secretion, and LDH release was dramatically reduced in NLRC4 or caspase-1 deficient macrophages compared to that of wild type macrophages. Subsequently, flagellin was identified to be the ligand that activates the NLRC4 inflammasome during Legionella pneumophila infection (Lightfield et al., 2008). Polymerized flagellins are structural components of flagella, a structure anchored to the bacterial cell wall that enables bacterial mobility. Legionella pneumophila mutants lacking flagellin expression did not cause pyroptosis in macrophages. Neither caspase-1 cleavage nor pro-il-1β secretion was observed. Of note, this flagellin-dependent NLRC4 activation requires a functional secretion system in Legionella pneumophila because a Legionella pneumophila mutant without a functional secretion system failed to trigger pro-il-1β processing and secretion in murine macrophages. Since flagellins have a well-defined function in bacterial motility, they are not likely to be the effector proteins that are translocated into cytosol to manipulate host cell signaling. Instead, flagellins may accidentally be translocated into the cytosol via type IV Dot/Icm secretion and induce NLRC4 activation. In any event, this suggests that the presence of flagellin in the cytoplasm is critical for NLRC4 inflammasome activation. Indeed, delivery of purified flagellin into the cytosol was capable of activating caspase-1 in a NLRC4-dependent manner. (Miao et al., 2006) Additionally, transfection of plasmids encoding flagellin and 13

23 NLRC4 inflammasome components lead to an increase in cell death in a HEK293T reconstituted system (Kofoed and Vance, 2011). Thus, these data indicate flagellin protein is the ligand that activates NLRC4. Like Legionella pneumophila, Salmonella typhimurium and Pseudomonas aeruginosa also caused LDH release and IL-1β secretion in a flagellin and NLRC4 dependent manner. Reduced IL-1β secretion was detected in the medium of macrophages lacking NLRC4 expression compared to that of wild type macrophages during bacterial infection. Moreover, a functional secretion system is also required to trigger this host response. However, intriguingly, a higher load of non-flagellated bacteria mutant with an intact secretion system still caused a significant amount of IL-1β secretion in wild type macrophages (Franchi et al., 2006; Sutterwala et al., 2007). This result suggests that NLRC4 might detect some bacterial products other than flagellin. Indeed, NLRC4 activates caspase-1 in response to bacterial secretion system components (Miao et al., 2010b). For example, a type III secretion system rod protein from Salmonella typhimurium, PrgJ, was able to activate NLRC4. Upon PrgJ recognition, caspase-1 proteolysis was detected in wild type macrophages but not in NLRC4 deficient macrophages. In addition, this innate immune response is not restricted to this bacterial species. Several other homologs of PrgJ were also shown to 14

24 be able to induce robust NLRC4 dependent IL-1β secretion, such as MxiI (Shigella flexneri), and PscI (Pseudomonas aeruginosa). Interestingly, PrgJ and flagellin share an amino acid motif critical for caspase-1 activation. Deletion mutations and point mutations in the carboxy-terminal region of PrgJ and flagellin abrogated caspase-1 activation (Lightfield et al., 2008; Miao et al., 2010b). This finding seems to support the idea that PRRs have evolved to recognize a limited number of invariable regions in microbial products, rather than a wide range of distinct signals Domain structure of NLRC4 and its activation mechanism NLRC4 has an amino terminal CARD domain, a central NACHT domain and a carboxy terminal LRR domain. Consistent with the LRR domain-mediated auto-inhibition model, deletion of the LRR domain caused NLRC4 activation as caspase-1 dependent cell death was observed in the absence of stimulation (Kofoed and Vance, 2011). In addition, deletion of the LRR domain triggered inflammasome assembly as an oligomer with molecular weight around 1,000 kda was detected in gradient native gel by immunoblotting. As expected, this assembly process requires ATP binding within the NACHT domain. Mutations in the NACHT domain that impair nucleotide binding abrogated the assembly of this higher order protein complex. Collectively, these results support the idea that auto-inhibition is mediated 15

25 by the LRR-NACHT interaction. When autoinhibition is released, either upon ligand detection or LRR domain deletion, the NACHT domain initiates nucleotide-binding dependent inflammasome assembly. Like several other NLR proteins, the NLRC4 has been demonstrated to interact with pro-caspase-1 (Damiano et al., 2004). Intriguingly, the NLRC4 CARD domain was also co-immunoprecipitated with ASC (Srinivasula et al., 2002), implicating that ASC might also have a role in NLRC4 inflammasome assembly and activation. In fact, ASC has been shown to promote NLRC4-mediated pro-il-1β processing and secretion, demonstrated by the markedly reduced IL-1β secretion from ASC-deficient macrophages (Case et al., 2009). One possible explanation of this result is that ASC contributes to the formation of a more stable NLRC4 inflammasome protein complex. Paradoxically, in spite of the role of ASC in facilitating pro-il-1β processing and secretion, ASC is not likely to promote pyroptotic cell death induced by NLRC4 activation. Upon NLRC4 agonist stimulation, similar amount of LDH was released from wild type and ASC deficient macrophages. A recent report even indicated that higher LDH activity was detected in the medium of ASC-deficient macrophages treated with NLRC4 agnoists compared to that from wild type macrophages (Case and Roy, 2011). Therefore, ASC seems to facilitate NLRC4-mediated pro-il-1β processing and secretion but negatively regulate pyroptotic cell death, raising the 16

26 possibility that ASC might be involved in differentiating NLRC4 downstream signaling. Although the NLRC4 inflammasome have been reported to respond to flagellin and bacterial secretion system components, the activation mechanism was not determined until recently. Two groups provided evidence that NLRC4 interacts with flagellins and bacterial secretion system components via distinct NAIPs (Kofoed and Vance, 2011; Zhao et al., 2011). NAIP is a homolog of NLRC4. Humans have one known NAIP; mice possess several paralogues. In C57BL/6 mice, four NAIPs (NAIP1, 2, 5 and 6) are expressed. NAIP5 shares the highest sequence similarity with NAIP6 (approximately 90%). Thus, it is generally believed that these two proteins may have similar functions. Indeed, NAIP5 and NAIP6 were both co-immunoprecipitated with flagellins. Although this interaction was independent of NLRC4, NLRC4 promoted the formation of a higher order protein complex, implicating that the assembly of NLRC4 inflammasome includes NLRC4, NAIP5 or NAIP6 and flagellins. In the non-macrophage reconstituted system, NLRC4 and either NAIP5 or NAIP6, were required for pro-caspase-1 proteolysis in response to flagellins. Intriguingly, flagellins were not able to physically associate with NAIP2 in two-hybrid assays or in co-immunoprecipitation assays. Instead, NAIP2 was found to 17

27 interact with secretion apparatus rod components, such as PrgJ from Salmonella typhimurium. NAIP2 is involved in NLRC4 activation responding to PrgJ as knock down of NAIP2 expression by shrna in murine macrophages diminished pro-caspase-1 cleavage triggered by PrgJ-stimulated NLRC4 activation. NAIPs, therefore, determine NLRC4 inflammasome specificity by interacting with distinct bacterial products, which subsequently facilitates inflammasome assembly and activation Role of NLRC4 inflammasome in vivo In vivo experiments further provide evidence indicating that NLRC4 plays a critical role in containing bacterial infection. For example, bacterial burdens were higher in NLRC4 deficient mice compared to that of wild type mice when they were injected intraperitoneally with a Pseudomonas aeruginosa strain that is able to cause pro-caspase-1 proteolysis in a NLRC4-dependent manner in macrophages (Sutterwala et al., 2007). Likewise, a Salmonella typhimurium strain that activates caspase-1 through NLRC4 was not lethal to mice. This strain is ectopically expressing flagellin. All the mice survived 6 days after infection and these bacteria were cleared more efficiently as less bacteria were recovered (Miao et al., 2010a). Reactive oxygen species (ROS) generated by innate immune cells has been 18

28 proposed to mediate this antibacterial process. In order to determine whether innate immune cells kill bacteria through ROS generation, Aderem and colleagues used Ncf-1 deficient mice that failed to generate ROS via NADPH oxidase. They found that less NLRC4 activating bacteria were recovered from wild type mice compared to that from Ncf-1 deficient mice (Miao et al., 2010a). Interestingly, another line of evidence showed that cytosolic flagellin promotes NO secretion, possibly through inducible nitric-oxide synthase upregulation, in a caspase-1 dependent manner (Buzzo et al., 2010). Collectively, these results suggest that ROS generation may play a critical for clearing bacteria via NLRC4-caspase-1 pathway NLRP3 inflammasome Domain structure of NLRP3 NLRP3 (also known as NALP3, cryopyrin, and PYPAF1) is the most extensively studied inflammasome scaffold protein. It is composed of a N-terminal PYD domain, a central NACHT domain and a C-terminal LRR domain. Since it lacks a CARD domain, it is likely that ASC would be an essential component in NLRP3 inflammasome to mediate pro-caspase-1 interaction. As expected, ASC has been 19

29 shown to physically associate with NLRP3 and pro-caspase-1 (Manji et al., 2002; Srinivasula et al., 2002). In a reconstituted system, pro-il-1β processing was not observed unless plasmids encoding ASC were transfected into HEK293K cells with NLRP3, pro-caspase-1, and pro-il-1β plasmids (Agostini et al., 2004). Like other NLR proteins, the NACHT domain in NLRP3 is also believed to bind ATP and promote inflammasome assembly. In fact, the biochemical evidence of this process has been provided (Duncan et al., 2007). The NACHT domain of NLRP3 preferentially bound ATP over GTP, CTP, and UTP. Mutations in the Walker A region, the nucleotide binding motif, abolished ATP binding in the NACHT domain. In addition, the NLRP3 Walker A mutant was not able to self-associate or interact with ASC in co-immunoprecipitation assays. In line with these results, Walker A mutations in NLRP3 diminished IL-1β secretion in transduced human monocytic THP-1 cells. Collectively, these results suggest that nucleotide binding, specifically ATP binding, in the NACHT domain is essential for NLRP3 inflammasome assembly and activation Microbial Sensing by NLRP3 inflammasome Bacterial pore-forming toxins have been reported to activate the NLRP3 inflammasome. These toxins are released by bacteria in a soluble form and 20

30 subsequently polymerize into a ring-like structure forming a pore in the host cell membrane. As a result, ionic imbalance may occur across the membrane. For example, a Listeria monocytogene mutant that lacks the pore-forming toxin listeriolysin O (LLO) failed to cause IL-1β secretion in THP-1 cells and murine macrophages (Meixenberger et al., 2010). Additionally, purified LLO protein was able to trigger caspase-1 activation in a NLRP3 dependent manner. Another example of a pore-forming toxin that activates NLRP3 is α-hemolysin from Staphylococcus aureus (Craven et al., 2009). Purified wild type α-hemolysin molecules caused pro-caspase-1 proteolysis whereas non-cytolytic mutant protein was defective in doing so. In line with these results, α-hemolysin has been reported to induce K + efflux, suggesting that pore formation may result in ionic imbalance, leading to NLRP3 activation (Walev et al., 1993). In addition to bacterial toxins, NLRP3 is activated as a host response to contain fungal and viral infection as well. For example, Candida albicans, an opportunistic fungal pathogen causing life-threatening infections in immunocompromised patients, was shown to induce robust IL-1β secretion in a NLRP3-dependent manner in murine macrophages (Hise et al., 2009). Importantly, in vivo experiments proved that NLRP3 is crucial for host survival as NLRP3-deficient mice were hypersusceptible to Candida albicans infection. All NLRP3-deficient mice died 6 days after infection 21

31 while more than 50% of wild-type mice survived (Gross et al., 2009). NLRP3 was also activated in macrophages during viral infection. Although the evidence of NLRP3-ligand interaction has yet to be demonstrated, it is generally believed that NLRP3 detects viral RNA because analogs of dsrna (poly(i:c)) and ssrna triggered NLRP3- and ASC- dependent IL-1β secretion in THP-1 cells (Allen et al., 2009). Intriguingly, Influenza virus A requires the M2 ion channel to activate NLRP3 as mutations in M2 protein trans-membrane domain abolished pro-il-1β processing and secretion (Ichinohe et al., 2010). In agreement with this result, transduction of murine macrophages with lentivirus expressing M2 proteins was sufficient to induce robust IL-1β secretion. Therefore, consistent with the pore-forming bacterial toxins, this result suggests that intracellular ionic alteration, potentially caused by membrane perforation, triggers NLRP3 inflammasome activation Non-microbial sensing by NLRP3 inflammasome Exogenous irritants, such as asbestos and silica, recently joined the NLRP3 activating group (Dostert et al., 2008; Hornung et al., 2008). Asbestos and silica induced robust IL-1β secretion in murine macrophages in a dose-dependent way. Down-regulation of NLRP3 or ASC expression markedly reduced IL-1β secretion, indicating that these 22

32 proteins play essential roles in mediating this process. Interestingly, NLRP3 activation requires the uptake of these crystalline substances by phagocytosis. Cytochalasin D, a well-characterized phagocytosis inhibitor that impairs actin filament assembly, blocked pro-il-1β processing in asbestos-stimulated macrophages. In contrast, NLRP3 activation induced by bacterial pore-forming toxin is independent of this engulfing process by macrophages, raising the possibility that there might be multiple pathways triggering NLRP3 activation. In addition to microbial products and exogenous irritants, danger-associated molecular patterns (DAMPs) also induce inflammatory cytokine processing and secretion through NLRP3 inflammasome, emphasizing that the innate immune system has not simply evolved to discriminate self from non-self molecules. For example, extracellular ATP was able to induce a robust IL-1β secretion in a NLRP3 dependent manner in murine macrophages (Mariathasan et al., 2006). Although extracellular ATP concentration is low under normal physiological conditions, probably due to the ubiquitous presence of ATPase, endogenous ATP released from damaged or dying cells could cause a localized increase in ATP concentration, consequently stimulating innate immune responses in adjacent cells. In agreement with this hypothesis, the release of ATP was observed from monocytes stimulated with different DAMPs (Piccini et al., 2008). Additionally, mitochondria-derived ATP from damaged cells 23

33 contributed to trigger pro-caspase-1 proteolysis in a NLRP3 and ASC dependent way (Iyer et al., 2009). Collectively, these results suggest that self danger signals, such as ATP, elicit innate immune responses and potentially facilitate pathogen clearance. Although it is well accepted that short-term inflammatory responses are beneficial for the host to contain microbial infection, prolonged inflammation may cause tissue damage. For example, the acute and painful inflammatory response present in patients with gout. The aetiological agent of gout is monosodium urate (MSU). This crystalline substance is formed when uric acid, a normal catabolic product of purine metabolism, is released into the extracellular environment from damaged cells. Given the observation that patients with gout have a higher level of proinflammatory cytokines, it is tempting to speculate that MSU might contribute to this elevated inflammatory symptom. In fact, it has been demonstrated the MSU crystals activated the NLRP3 inflammasome and consequently induced IL-1β secretion (Martinon et al., 2006). Based on this finding, a recent report indicated that IL-1β inhibition actually has a beneficial effect in gouty inflammation. An IL-1 receptor antagonist rapidly relieved inflammatory symptoms of gout in a test of ten patients (So et al., 2007). Although further studies are required to confirm the efficacy of this drug, pharmacological inhibition of IL-1β signaling appears to be a promising therapy to treat gout. 24

34 Molecular mechanism of NLRP3 activation As over-production of inflammatory cytokines is detrimental, NLRP3 activation needs to be tightly regulated. Indeed, NLRP3 activation requires two signals. The first one is the priming step and the second one is the activation step. Initially, the priming step by LPS treatment was thought to induce pro-il-1β production, and the subsequent processing of pro-il-1β to mature IL-1β is the most common readout to assess inflammasome activity. However, no pro-caspase-1 proteolysis was observed in ATP-stimulated macrophages without LPS priming, raising the possibility that NLRP3 might be regulated transcriptionally as well (Hu et al., 2010). Two recent reports corroborated this hypothesis (Bauernfeind et al., 2009; Franchi et al., 2009). NLRP3 expression was induced in the LPS-primed cells and this induction is essential for NLRP3 activation. This idea was further supported by the finding that overexpression of NLRP3 could circumvent the necessity of priming as pro-caspase-1 cleavage was observed readily in cells treated with NLRP3 agonists. Mechanistically, analysis of LPS-mediated downstream signaling revealed that NF-κB activation controls NLRP3 expression as pharmacological inhibition of the NF-κB pathway prevented NLRP3 transcription induction. Collectively, these data indicate that NLRP3 expression is tightly controlled by signals culminating in the activation of 25

35 NF-κB. The second step is the activation step which allows NLRP3 to initiate inflammasome assembly and induce pro-il-1β processing. This step is still under debate and three different models have been proposed. However, these models are not mutually exclusive and each could help to explain the activation step of the NLRP3 inflammasome. The first model is the potassium efflux model. As a result of pore formation in the membrane, which could be done by pore-forming bacterial toxins, potassium efflux occurs and activates NLRP3. This model is supported by the reports demonstrating that high potassium concentration in the medium reduced IL-1β secretion in response to NLRP3 agonists (Petrilli et al., 2007). Of note, this inhibitory effect is specific to potassium as high sodium concentration in the medium had no effect on pro-il-1β processing and secretion induced by NLRP3 activation. Interestingly, although the requirement of potassium efflux was considered to be specific to NLRP3 activation, recent reports indicated that high extracellular potassium concentration appeared to block AIM2 and NLRC4 activation as well, suggesting that potassium efflux might be a common signal triggering inflammasome activation (Arlehamn et al., 2010; Fernandes-Alnemri et al., 2010). The second model is the lysosome destabilization model. Crystalline substances, 26

36 such as silica, may induce lysosome destabilization upon phagocytosis, leading to the translocation of proteolytically active lysosomal proteases into the cytosol. Consequently, NLRP3 is activated due to the degradation of a potential inhibitory binding protein. In agreement with this model, lysosome rupture and the release of proteolytically lysosomal contents were demonstrated by fluorescence microscopy in stimulated macrophages (Hornung et al., 2008). In addition, pharmacological inhibition of cathepsin B, a lysosomal protease, blocked IL-1β secretion from macrophages treated with NLRP3 crystalline agonists. Nonetheless, high concentrations of cholesterol crystals still induced a robust IL-1β secretion from cathepsin B deficient macrophages compared to that from wild type macrophages, implicating other lysosomal components in the activation of NLRP3 (Duewell et al., 2010). The third model is the reactive oxygen species (ROS) model, which is under intense debate. It has been demonstrated that NLRP3 agonists induced ROS production in macrophages (Cruz et al., 2007). Furthermore, N-acetyl cysteine, a non-specific ROS scavenger, reduced pro-caspase-1 processing and IL-1β secretion in macrophages in response to extracellular ATP and silica (Dostert et al., 2008). Interestingly, a recent study suggests an important role for mitochondria in facilitating ROS generation, which in turn promotes NLRP3 activation (Zhou et al., 2011). Knock 27

37 down of the voltage-dependent anion channel (VDAC) in the outer mitochondrial membrane caused a significant reduction in ROS production. In line with this result, there was less pro-caspase-1 cleavage and pro-il-1β processing and secretion detected in these VDAC down-regulated cells compared to that of wild type cells. Collectively, these results support the ROS model of NLRP3 activation and suggest that ROS might derive from mitochondrial respiration. However, one very recent report suggests that ROS are not involved in the activation step. Instead, it is involved in the priming step, inducing NLRP3 expression (Bauernfeind et al., 2011b). NAC inhibited LPS-induced NLRP3 mrna transcription in a dose-dependent manner in macrophages. In any event, more studies are needed to clarify the role of ROS in NLRP3 signaling NLRP3-associated genetic disorders Genetic mutations in NLRP3, previously known as cryopyrin, have been identified to be associated with several auto-inflammatory diseases (Aksentijevich et al., 2002; Hoffman et al., 2001). These diseases are named cryopyrin-associated periodic syndromes (CAPS), including familial cold autoinflammatory syndrome (FCAS), Muckle-Wells syndrome (MWS), and chronic infantile neurologic cutaneous and articular (CINCA) syndrome or neonatal-onset multisystem inflammatory disease 28

38 (NOMID). The symptoms of these diseases, such as fevers and rashes, are likely due to the overproduction of pro-inflammatory cytokines, specifically IL-1β. Consistent with this idea, elevated IL-1β levels were detected in the serum in a mouse model of CAPS (Brydges et al., 2009). Additionally, primary cells isolated from CAPS patients were hypersensitive, demonstrated by the secretion of IL-1β upon LPS treatment (Gattorno et al., 2007). In support of the role of IL-1β in these diseases, the IL-1β receptor antagonist, Anakinra, has been shown to be effective at reducing symptoms clinically (Hawkins et al., 2003). Among all the CAPS-related mutations identified to date, most of them are located in the NACHT domain, implicating that these mutations might be involved in inflammasome assembly (Neven et al., 2004). In fact, one group showed that CAPS-related NLRP3 mutants exhibited increased oligomerization with ASC in an overexpression system (Yu et al., 2006). Intriguingly, these CAPS-related NLRP3 mutants were less soluble than wild type NLRP3 during protein purification, suggesting that CAPS-related NLRP3 mutants might form a larger protein complex than wild type NLRP3. This structural feature could possibly render NLRP3 more prone to change to the active conformation and account for its hypersensitivity upon stimulation. 29

39 1.3 NLRP1 inflammasome Human NLRP1 Human NLRP1 is widely expressed and has the highest expression level in T cells and Langerhans cells (Kummer et al., 2007). This pattern is consistent with the particular involvement of NLRP1 in skin autoimmunity. Mutations in the gene encoding for human NLRP1 have been associated with generalized vitiligo, a multifactorial auto-inflammatory disease characterized by loss of skin pigmentation (Jin et al., 2007a; Jin et al., 2007b). NLRP1 is composed of five domains: an amino-terminal PYD domain, followed by a NACHT domain, a LRR domain, a FIIND domain and a carboxy-terminal CARD domain. The PYD domain has been demonstrated to bind ASC, which bridges the interaction with pro-caspase-1. Intriguingly, the CARD domain seems able to interact with either pro-caspase-1 or pro-caspase-5. In co-immunoprecipitation assays, the CARD domain physically associated with pro-caspase-5 (Martinon et al., 2002). Deletion of the CARD domain impaired the NLRP1-pro-caspase-1 interaction (Faustin et al., 2007). Consistent with the proposed auto-inhibition model, deletion of 30

40 the NACHT domain caused NLRP1 activation, which was demonstrated by cell death induction (Hlaing et al., 2001). Recently, one report suggested that the FIIND domain undergoes autoproteolysis. Although it is unclear whether this autoproteolysis regulates human NLRP1 activity, this cleavage event is likely to be independent of cellular proteases because in vitro translated CARD8, a protein composed of one FIIND domain and a CARD domain, was also processed into two fragments (D'Osualdo et al., 2011). The trigger of human NLRP1 activation remains poorly understood. Disruption of cell integrity is likely to activate NLRP1 in human monocytic THP-1 cells because cell lysis in a hypotonic buffer caused pro-caspase-1 processing (Martinon et al., 2002). This pro-caspase-1 proteolysis was dependent on NLRP1 and ASC since immunodepletion of either NLRP1 or ASC markedly reduced pro-caspase-1 processing in response to cell integrity disruption. Additionally, caspase-5 seemed to facilitate caspase-1-mediated pro-il-1β processing in an overexpression system. Collectively, these results suggest that the NLRP1 inflammasome comprises NLRP1, ASC, pro-caspase-1 and pro-caspase-5. An in vitro reconstituted assay indicated that NLRP1 and pro-caspase-1 are sufficient to assemble a functional NLRP1 inflammasome responding to muramyl dipeptide (MDP), a peptidoglycan constituent from bacteria (Faustin et al., 2007). In 31

41 the presence of MDP, purified NLRP1 activated pro-caspase-1, demonstrated by an increase of relative fluorescence units from the cleavage of fluorogenic caspase-1 substrates (Faustin et al., 2007). Though ASC is not essential in this experimental setting, it could promote caspase-1 activity upon MDP detection by NLRP1. Two anti-apoptotic proteins, Bcl-2 and Bcl-XL, were suggested to negatively regulate NLRP1 activity through physical association (Bruey et al., 2007). This direct interaction was demonstrated by pull down assays using purified proteins. The inhibitory effect of Bcl-2 and Bcl-XL on NLRP1 activity was demonstrated in a cell-based reconstitution system. Overexpression of either Bcl-2 or Bcl-XL diminished pro-il-1β processing and secretion. Furthermore, knock down of Bcl-2 and Bcl-XL expression in THP-1 cells increased IL-1β secretion in response to MDP. Deletion analysis of Bcl-2 revealed that a 20 amino acid peptide was sufficient to inhibit NLRP1 activity, suggesting a possible way to develop cell-permeable peptide therapeutics for auto-inflammatory diseases (Faustin et al., 2009) Domain structure of murine Nlrp1b Nlrp1b differs in the domain structure from human NLRP1 as Nlrp1b does not have the PYD domain, suggesting that ASC might not be an essential component in Nlrp1b inflammasome. Indeed, in a cell-based reconstituted system, ASC was not required 32

42 for Nlrp1b activation (see Chapter 3). Moreover, ASC-deficient RAW macrophage-like cells undergo pyroptosis upon stimulation (Newman et al., 2009). Except the lack of a PYD domain, Nlrp1b has a N-terminal NACHT domain, a central LRR domain, followed by a FIIND domain, and a C-terminal CARD domain. Interestingly, autoproteolysis was also observed in the FIIND domain of Nlrp1b (see Chapter 3). Two fragments were detected when Nlrp1b was overexpressed. This autoproteolysis is not likely due to the artifact of overexpression since two fragments were also detected in endogenous Nlrp1b in macrophages. In fact, this cleavage event is important for Nlrp1b activity (Brad Frew, unpublished data) Nlrp1b controls murine macrophage susceptibility to anthrax lethal toxin Murine Nlrp1 has three paralogues: Nlrp1a, Nlrp1b, and Nlrp1c. Like Nlrp1b, Nlrp1a has all four domains whereas Nlrp1c lacks the CARD domain. Although the functions of Nlrp1a and Nlrp1c are currently unclear, Nlrp1b was shown to mediate macrophage cell death in response to anthrax lethal toxin (LeTx), a bacterial toxin that contains one zinc metalloprotease, lethal factor (LF) and one protein for LF entry, protective antigen (PA) (Boyden and Dietrich, 2006). Anthrax lethal toxin action begins when PA binds to one of the two known receptors on cells and is proteolytically activated by members of the furin protease 33

43 family. The protease removes a 20 kda domain from the N-terminus, leaving PA63 bound to the receptor. Receptor-bound PA63 can self-associate to form a ring-shaped heptamer, called the prepore. The prepore is capable of binding LF and the resulting complexes are delivered to the endosomal compartments. There, under the influence of low ph, the prepore forms a transmembrane pore that delivers LF to the cytosol. Though it is known that MAPKKs are direct targets of LF in the cytosol, it is unknown whether there are other substrates of LF (Collier and Young, 2003). Nlrp1b has five alleles and the expression of each allele is determined by the genetic background of the mice strain (Boyden and Dietrich, 2006). For example, macrophages derived from BALB/C mice harbor Nlrp1b allele 1 whereas macrophages from C57BL/6 mice express Nlrp1b allele 2. These alleles differ in their responsiveness to LeTx. Nlrp1b allele 1 and 5, collectively called sensitive alleles, triggered pro-caspase-1 processing and caused macrophage death in response to LeTx; however, Nlrp1b allele 2, 3 and 4, called resistant alleles, failed to do so when macrophages were treated with LeTx. It is still unclear why Nlrp1b allele 2, 3 and 4 do not respond to LeTx. One possible explanation for the defect in activating caspase-1 by Nlrp1b allele 4 in response to LeTx is that allele 4 lacks the CARD domain, which makes it incapable of binding to pro-caspase-1 for activation. Although it is well known that LeTx induced Nlrp1b-mediated macrophage 34

44 death, little is known about the mechanism of Nlrp1b activation. A recent report suggested that Nlrp1b is membrane associated, which was demonstrated by subcellular fractionation assays (Nour et al., 2009). In addition, size exclusion chromatography revealed, upon LeTx stimulation, the formation of high molecular weight oligomers that include Nlrp1b and pro-caspase-1 but not ASC. Like other inflammasomes, the role of potassium flux and lysosome destabilization has been tested in Nlrp1b activation (Averette et al., 2009). High extracellular potassium concentration abolished pro-caspase-1 processing, indicating that potassium efflux is involved in Nlrp1b activation (Fink et al., 2008). Furthermore, a cathepsin B inhibitor protected cells from LeTx-indcued cytotoxicity. Although knock down of cathepsin B expression had little effect on cell protection, this could be due to inefficient knock down of cathepsin B by sirna (Newman et al., 2009). Taken together, these results suggest that potassium efflux and lysosome destabilization are likely to contribute to Nlrp1b inflammasome assembly and activation. MG-132, a proteasome inhibitor, has been shown to protect macrophages from LeTx-induced cell death (Squires et al., 2007; Tang and Leppla, 1999). Recently, as expected, MG-132 was demonstrated to block pro-caspase-1 processing and IL-1β secretion in LeTx-stimulated macrophages, raising the possibility that Nlrp1b might be triggered by the proteasome-mediated degradation of an inhibitory binding protein 35

45 (Squires et al., 2007) Role of Nlrp1b inflammasome in vivo Initially, LeTx-induced cytotoxicity in macrophages harboring sensitive Nlrp1b alleles was conceived as a strategy employed by Bacillus anthracis to evade immune surveillance and promote bacterial dissemination. However, recent in vivo experiments suggested that LeTx-induced macrophage lysis mediated by sensitive Nlrp1b alleles actually contributes to host survival (Moayeri et al., 2010; Terra et al., 2010). C57BL/6 mice, of which macrophages are resistant to LeTx due to the expression of Nlrp1b allele 2, succumbed to spore or vegetative bacteria infection after 6 days. In contrast, transgenic C57BL/6 mice that express Nlrp1b allele 1 survived during the period of the experiment. Consistent with this finding, caspase-1 deficient mice expressing Nlrp1b allele 1 were susceptible to spore infection. Collectively, these results suggested the critical role of the Nlrp1b-caspase-1 signaling axis to contain Bacillus anthracis infection and promote host survival. 1.4 Links between inflammation and metabolism Immune cells that respond to infection initiate energy-demanding processes, often in 36

46 inflamed tissues that are low in oxygen and glucose. These conditions cause energetic stress that results in a metabolic switch from oxidative phosphorylation to glycolysis. Although glycolysis is a less efficient means to generate ATP, it is rapid and capable of meeting the needs of activated cells. The glycolytic pathway is also upregulated by Toll-like receptor signaling, indicating that immune cells that detect pathogens prepare to fight infection by altering their metabolism. For example, dendritic cells stimulated multiple TLR agonists up-regulate Glucose transporter1 expression, accompanied by an increase in lactate production and a decrease in mitochondrial oxygen consumption, all hallmarks of glycolysis (Krawczyk et al., 2010). Studies on the NLRP3 inflammasome have revealed additional links between metabolism and innate immunity (Tannahill and O'Neill, 2011). For example, inflammatory cytokines, such as IL-1β, have been implicated in inducing insulin resistance in type 2 diabetes (T2D) (Feve and Bastard, 2009). However, how IL-1β is generated during disease progression remains poorly understood. A recent study suggested that palmitate, one of the most abundant saturated fatty acids that is substantially elevated following a high-fat diet, induced robust IL-1β secretion from murine macrophages via a NLRP3 dependent pathway (Wen et al., 2011). This result is likely to be relevant to T2D because a high-fat diet is believed to be an important 37

47 factor contributing to T2D. Mechanistically, AMP-activated protein kinase (AMPK) has been demonstrated to regulate NLRP3 activation in response to palmitate. AMPK is a major energy sensor in cells and play an essential role in fatty acid metabolism (Carling et al., 2011; Steinberg and Kemp, 2009). It has been speculated that AMPK regulates NLRP3 activity through altering ROS generation because AMPK suppresses ROS production by inhibiting expression and function of NADPH oxidase (Wang et al., 2010). Indeed, AMPK activation reduced ROS production induced by palmitate and, consequently, diminished NLRP3 activity (Wen et al., 2011). Intriguingly, NLRP3-mediated IL-1β secretion is not restricted to macrophages or dendritic cells. High glucose concentration was capable of activating NLRP3 and triggering pro-il-1β processing in islets isolated from mice (Zhou et al., 2010). The low level of IL-1β secreted from islets was probably due to its lower expression of NLRP3 inflammasome components. Furthermore, another mechanism of enhanced IL-1β production in T2D was also revealed recently. Oligomers of islet amyloid polypeptide (IAPP), a peptide hormone involved in glucose metabolism and gut function, activated NLRP3 and subsequently caused robust pro-il-1β processing and secretion (Masters et al., 2010). Unlike palmitate-induced NLRP3 activation, phagocytosis is required for this IAPP aggregate 38

48 to activate NLRP3 since Cytochalasin D diminished IAPP aggregate-induced IL-1β secretion. Thus, NLRP3 seems to be activated during diabetes progression, leading to a sustained IL-1β production, which in turn contributes to insulin resistance by affecting intracellular insulin-signaling pathways. 39

49 1.5 Thesis Rationale Nlrp1b has been shown to mediate LeTx-induced macrophage death. Although it remains unclear how Nlrp1b is activated by LeTx, one model of Nlrp1b activation has been proposed (Fig. 1.2). In resting cells, the intramolecular NACHT-LRR interaction keeps Nlrp1b in an inactive form. Once the LRR domain detects the LeTx activity, this auto-inhibitory interaction is released. Then, Nlrp1b is able to form oligomers via NACHT-NACHT interactions. The CARD domain can recruit pro-caspase-1, which undergoes autoproteolysis upon being brought into proximity with other molecules of pro-caspase-1. Consequently, active caspase-1 induces pyroptosis and triggers pro-il-1β processing. In order to provide biochemical evidence of this activation model to study the Nlrp1b inflammasome, I set out to reconstitute the Nlrp1b inflammasome in a cell-based system. In this system, I will perform structure function analysis of Nlrp1b to gain some insight into the Nlrp1b activation mechanism. In addition, this analysis might provide some clues of other ligands that are also recognized by Nlrp1b, because it is unlikely that Nlrp1b has just evolved to detect such a specific bacterial protease activity. Nlrp1b should be able to respond to other PAMPs or DAMPs as well. 40

50 Figure 1.2 Activation model of the Nlrp1b inflammasome. Initially, the intramolecular interaction between the NACHT domain and the LRR domain keeps Nlrp1b in an inactive form. When the LRR domain detects the LeTx activity, this autoinhibitory interaction is released. Then, the NACHT domain is free to self-associate; the CARD domain is able to recruit pro-caspase-1. Pro-caspase-1 undergoes proteolysis when multiple molecules of pro-caspase-1 are brought into proximity. As a result, caspase-1 induces cell death and trigger pro-il-1β processing. 41

51 CHAPTER Material and methods 2.1 Cell culture and reagents HT1080 cells (ATCC) were cultured in Dulbecco's modified Eagle medium supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin. PEG 200, PEG 2000, and PEG 6000 (Sigma-Aldrich) were added to the culture medium at a final concentration of 15 mm. PA, LF, and LF-E687A were purified as described previously and applied to cells at a final concentration of 10-8 M (Kassam et al., 2005). The proteasome inhibitor, MG-132 (Calbiochem), was used at 10 µm. Sodium azide, 2-deoxyglucose (2DG), N-acetyl-cysteine (NAC) and staurosporine (STS) were purchased from Sigma-Aldrich and used at indicated concentrations. 2.2 Plasmid construction, cdna cloning, and site-directed mutagenesis The Nlrp1b allele 1 gene was amplified by using the forward primer 5 -CGC GGA TCC TAT GGA AGA ATC CCC ACC CAA G-3 and the reverse primer 5 -CGC CTC GAG TCA TGA TCC CAA AGA GAC CCC ACC TG-3 from cdna derived from RAW264.7 cells. The PCR product was digested with BamHI and XhoI, and 42

52 then ligated into pntap-a (Stratagene ). The Nlrp1b allele 3 gene was cloned in two steps. A 5 fragment was amplified using the forward primer 5 -CGC GGA TCC TAT GGA AGA ATC CCC ACC CAA G-3 and reverse primer 5 -CGC CTC GAG TCA TGA TCC CAA AGA GAC CCC ACC TG-3 from cdna derived from TIB-47 cells. The PCR product was digested with the restriction enzymes BamHI and XhoI, and then ligated into pntap-a. The second fragment of the Nlrp1b allele 3 gene was amplified by using the forward primer 5 -CGC CTC GAG GAA GTC ACC CTT CAC CTC TAC and the reverse primer 5 -CGC GGG CCC TCA TGA TCC CAA AGA GAC CCC ACC-3. The PCR product was digested with XhoI and ApaI, and then ligated into the plasmid containing the 5 fragment of the Nlr1pb allele 3 gene. The pro-il-1β gene was amplified by using the forward primer 5 -CGC GAA TTC ATG GCA ACT GTT CCT GAA CTC-3 and the reverse primer 5 -CGC CTC GAG GGA AGA CAC GGA TTC CAT GGT G-3. The PCR product was digested with EcoRI and XhoI, and then ligated into pcdna3-ha (Go et al., 2009). pcdna3-t7 was generated by inserting the T7 tag into pcdna3 between the ApaI and NheI restriction sites. Pro-caspase-1 was amplified by using the forward primer 5 -CGC GGA TTC TAT GGC TGA CAA GAT CCT GAG G-3 and the reverse primer 5 -CGC CTC GAG ATG TCC CGG GAA GAG GTA G-3. The PCR 43

53 product was digested with BamHI and XhoI, and then ligated into pcdna3-t7. QuickChange site-directed mutagenesis (Stratagene) was performed according to the manufacturer s instructions to introduce the C284A mutation into pcdna3-pro-caspase-1-t7 using the oligonucleotide 5 -GAT CAT TAT TCA GGC AGC GCG TGG AGA GAA ACA AGG-3 and its complement. Nlrp1b truncation plasmids were constructed by amplifying fragments from pntap-nlrp1b allele 1. The reverse primer 5 -CGC CTC GAG TCA TGA TCC CAA AGA GAC CCC ACC TG-3 was used with the following forward primers to amplify the designated fragments: Nlrp1b CGC GGA TCC TGA GGA TAG TGA GGA AAG ACA C-3, Nlrp1b CGC GGA TCC TGA CCT GTC CTC TCT CAG TGC C-3, Nlrp1b CGC GGA TCC TTT CCA ACT CTT CTC TGA GAT CTA C-3, and Nlrp1b CGC GGA TCC TCT GCA CTT CAT GGA CCA GCA TC-3. Nlrp1b was amplified by using the forward primer 5 -CGC GGA TCC TAT GGA AGA ATC CCC ACC CAA G-3 and the reverse primer 5 -CGC CTC GAG TCA CAA GGA AGG GGC ATC TTT GAG-3. The PCR products were digested with the restriction enzymes BamHI and XhoI, and the resulting products were ligated into pntap-a. To construct the Nlrp1b vector, a SalI site was introduced into pntap-nlrp1b allele 1 by using the oligonucleotide 5 -CTT AAA TTC ACC GTC 44

54 GAC CTG GAG GGG TTG-3 and its complement. A second SalI site was introduced by using the oligonucleotide 5 -AGC ATC CTG TGG GTC GAC CTG TCC TCT CTC-3 and its complement. This plasmid was digested with SalI and resolved by agarose gel electrophoresis. Gel extraction was performed and the plasmid lacking the SalI fragment was ligated together. Nlrp1b fragments for GST fusion protein constructs were amplified with the forward primer 5 -CGC GGA TCC TTC CAA CTC TTC TCT GAG ATC TAC-3 and reverse primer 5 -CGC CTC GAG TCA TGA TCC CAA AGA GAC CCC ACC TG-3 for Nlrp1b ; and the forward primer 5 -CGC GGA TCC CTG CAC TTC ATG GAC CAG CAT C-3 and reverse primer 5 -CGC CTC GAG TCA TGA TCC CAA AGA GAC CCC ACC TG-3 for Nlrp1b The PCR products were digested with restriction enzymes BamHI and XhoI, and the resulting products were ligated into pgex4t-1. A six histidine-tag was inserted into pcdna3-ha between the HindIII and BamHI restriction sites to generate pcdna3-his-ha. Nlrp1b was amplified by using the forward primer 5 -CGC GGA TCC TTC CAA CTC TTC TCT GAG ATC TAC-3 and reverse primer 5 -CGC CTC GAG TGA TCC CAA AGA GAC CCC AC-3. Nlrp1b was amplified by using the forward primer 5 -CGC GGA TCC CTG CAC TTC ATG GAC CAG CAT C-3 and the reverse primer 5 -CGC 45

55 CTC GAG TGA TCC CAA AGA GAC CCC AC-3. The PCR products were digested with the restriction enzymes BamHI and XhoI, and the resulting products were ligated into pcdna3-his-ha. pcdna3-his-nlrp1b HA and pcdna3-his-nlrp1b HA were digested with the restriction enzymes HindIII and XhoI. The digestion products were resolved by electrophoresis. Gel extraction was performed to isolate the inserts, which were then ligated into pcdna3-t7 to generate pcdna3-his-nlrp1b T7 and pcdna3-his-nlrp1b T7. QuickChange site-directed mutagenesis (Stratagene) was performed according to the manufacturer s instructions to generate p-ntap-nlrp1b-wa mutant, containing G137A, K138A, and S139A. These three alanine mutations were introduced to pntap-nlrp1b by using the oligonucleotide 5 - GAA GGG GCT GCT GGG ATT GCG GCG GCA ACA CTG GCC AGG CTG GTG AAG-3 and its complement. The AMPKα1 was amplified by using the forward primer 5 - CGC GCG GCC GCA ATG GCG ACA GCC GAG AAG CAG -3 and the reverse primer 5 - CGC CGC GAG TTG TGC AAG AAT TTT AAT TAG -3 from an Open Biosystems plasmid, Clone ID: LIFESEQ The PCR product was digested with NotI and XhoI, and then ligated into pcdna3-his*6-flag*3. The pcdna3-his*6-ampkα1 T174A-FLAG*3 was made by using the oligonucleotide 5 - GGT GAA TTT TTA AGA GCA AGT TGT GGC TCA CC -3 and its complement. 46

56 The Nlrp3 gene was amplified by using the forward primer 5 - CGC AAG CTT ATG ACG AGT GTC CGT TGC AAG -3 and the reverse primer 5 - CGC CTC GAG TCA CCA GGA AAT CTC GAA GAC TAT AG -3 from an Open Biosystems plasmid, clone ID: The PCR product was digested with Hind III and XhoI, and then ligated into pntap-b (Stratagene ). The pntap-nlrp3-wa mutant, containing G227A, K228A, and T229A, was made by using the oligonucleotide 5 - GTG TTC CAG GGA GCA GCA GGC ATC GCG GCA GCC ATC CTA GCC AGG AAG ATT ATG -3 and its complement. Additionally, a R258W mutation was introduced into pntap-nlrp3 by using the ologionucleotide 5 - CTA TTT GTT CTT TAT CCA CTG CTG GGA GGT GAG CCT CAG GAC G-3 and its complement. The ASC gene was amplified from an Open Biosystems plasmid, clone ID: , by using the forward primer 5 - CGC GGA TCC ATG GGG CGG GCA CGA GAT GCC-3 and the reverse primer 5 - CGC GCG GCC GCT CAG CTC TGC TCC AGG TCC ATC AC-3. The PCR product was digested with BamHI and NotI, and then ligated into pcdna3-flag. 2.3 IL-1β and ATP level assays One million HT1080 cells were seeded on a 10 cm dish the day before transfection. 47

57 On the day of transfection, 1 µg each of pntap-nlrp1b, pcdna3-pro-caspase-1-t7, and pcdna3-pro-il-1β-ha were transfected using 9 µl of 1 mg/ml polyethyleneimine ph 7.2. Approximately 24 h after transfection, cells were treated with LF (10-8 M) and PA (10-8 M) or 50 mm 2DG and 10 mm NaN 3 in 5 ml medium for 3 h. The cell supernatant was mixed with 1 µl of α-ha antibody (Sigma-Aldrich H9658) overnight, followed by the addition of 100 µl of protein A sepharose (GE Healthcare) and a 2 h incubation. Proteins were eluted from the protein A sepharose beads with SDS loading dye and subjected to immunoblotting using a polyclonal HA antibody (Santa Cruz sc805). Cell pellets were harvested and then lysed with 300 µl of EBC buffer (0.5% NP-40, 20 mm Tris ph 8, 150 mm NaCl, 1 mm PMSF) for 60 min. Equivalent amounts of cell lysate protein (~ 30 µg) were subjected to SDS-PAGE and immunoblotted with α-ha (Santa Cruz sc805) and α-β-actin (Sigma-Aldrich A5441) antibodies. Intracellular ATP levels were measured by a CellTiter-Glo Luminescent Cell Viability Assay (Promega G7571) in accordance with the manufacturer s instructions. 48

58 2.4 LDH release assays Release of cytoplasmic LDH into the cell supernatant was measured by CytoTox 96 Non-Radioactive Cytotoxicity Assay (Promega G-1780) according to the manufacturer s instructions. The percentage of LDH released was calculated as 100 (experimental LDH - spontaneous LDH)/(maximum LDH spontaneous LDH). 2.5 Detection of TAP-tagged proteins One 10-cm dish of HT1080 cells were transfected with 1µg plasmids encoding designated TAP-tagged proteins, 1 µg plasmids encoding pro-caspase-1-t7, and 1 µg plasmids encoding pro-il-1β-ha. Approximately 24 h after transfection, cell pellets from each plate were lysed with 300 µl EBC buffer at 4 ºC for 1 h. Cell lysates from 3 plates were incubated with 25 µl streptavidin agarose resin (Thermo Scientific 20349) for ~2 h. Beads were washed 3 times with 1 ml EBC buffer. Proteins were eluted with SDS and analyzed by immunoblotting with α Calmodulin binding peptide antibody (Upstate ) or α-nlrp1b antibody. 2.6 GST fusion protein purification and in vitro binding assay GST, GST-Nlrp1b and GST-Nlrp1b proteins were bound to glutathione 49

59 sepharose at a concentration of ~0.4 mg/ml according to manufacturer s instructions (GE Healthcare). Lysates were prepared from one plate of HT1080 cells transfected with designated plasmids by sonication or incubation with 300 µl EBC buffer at 4 C for 1 h. For the pro-caspase-1-t7 binding assay, 40 µl of beads were incubated at 4 C for ~2 h with 400 µl of cell lysate containing pro-caspase-1-c284a-t7. For the oligomerization binding assay, 40 µl of beads were incubated with 600 µl cell lysate containing either His-Nlrp1b HA or His-Nlrp1b HA. Beads were washed 3 times with 1 ml of cold EBC buffer. Proteins were eluted with SDS and analyzed by immunoblotting with α-caspase-1 p10 (M20) (Santa Cruz sc514) or α-ha antibodies (Santa Cruz sc805). 2.7 Co-immunoprecipitation assay Two plates of HT1080 cells were transfected with pcdna3-his-nlrp1b T7 and pcdna3-his-nlrp1b HA or with pcdna3-his-nlrp1b HA and pcdna3-his-nlrp1b T7. Cells were lysed in 600 µl EBC buffer by sonication and the lysates were clarified by centrifugation. Lysates were incubated with 1 µl of α-ha antibody (Sigma-Aldrich H9658) or 1 µl of control α-gfp antibody (Covance MMS-118R) for 2 h, followed by the addition of 50 µl of protein A sepharose (GE Healthcare) and a 2 h incubation. Complexes were resolved by 50

60 SDS-PAGE and immunoblotted using an α-t7 antibody (Novagen 69522). 2.8 Knock down of endogenous AMPK Half a million HT1080 cells were seeded on a 10 cm dish the day before transfection. On the day of transfection, 15 µl of control sirna (10 µm) (Santa cruz sc-37007) or AMPK sirna (10 µm) (Santa cruz sc-45312) were transfected using 15 µl Lipofectamine RNAiMAX Transfection Reagent (Invitrogen ). Approximately 48 h after transfection, cell lysates were harvested and cytosolic proteins were extracted by using NE-PER Nuclear and Cytoplasmic Extraction Reagent Kit (Fisher scientific PI78833) in accordance with the manufacturer s instructions. Equivalent amounts of protein (~60 µg) were subjected to SDS-PAGE and immunoblotted with α-phospho-ampk (Santa Cruz sc ), α-total-ampk (Santa cruz sc-74461), and α-β-actin (Sigma-Aldrich A5441) antibodies. 51

61 CHAPTER Expression of Nlrp1b inflammasome components in human fibroblasts confers susceptibility to anthrax lethal toxin The content of this chapter is reprinted, in part, from Infection and Immunity, Vol. 77, No. 10, Kuo-Chieh Liao and Jeremy Mogridge, Expression of Nlrp1b inflammasome components in human fibroblasts confers susceptibility to anthrax lethal toxin, pages , with permission from American Society for Microbiology. Contributions: I performed all the experiments presented in this chapter. 3.1 Summary Anthrax lethal toxin causes macrophages and dendritic cells from some mouse strains to undergo caspase-1-dependent cell death. Central to this process is the NOD-like receptor Nlrp1b (Nalp1b), which detects intoxication and then self-associates to form a complex termed an inflammasome that is capable of activating the pro-caspase-1 zymogen. The nature of the signal detected directly by Nlrp1b is not known and the mechanisms of inflammasome assembly are poorly understood. Here I demonstrate 52

62 that transfection of human fibroblasts with plasmids encoding murine Nlrp1b and pro-caspase-1 was sufficient to confer susceptibility of the cells to lethal toxin-mediated death. As has been observed in murine macrophages, the enzymatic activities of lethal toxin and the proteasome were both required for the activation of the Nlrp1b inflammasome and this activation led to pro-interleukin-1β processing. Release of interleukin-1β from cells was not dependent on cell lysis as its secretion was not affected by an osmoprotectant that prevented the appearance of lactate dehydrogenase in the culture medium. I generated constitutively active mutants of Nlrp1b by making amino-terminal deletions to the protein and observed that the ability to activate pro-caspase-1 was dependent on the CARD domain, which bound pro-caspase-1, and a region adjacent to the CARD domain that promoted self-association. My results demonstrate that lethal toxin can activate Nlrp1b in a non-myeloid cell line and are consistent with work that suggests activation induces proximity of pro-caspase-1. 53

63 3.2 Results Activation of the Nlrp1b inflammasome in HT1080 cells I sought to determine whether LeTx could activate the assembly of the murine Nlrp1b inflammasome in a human non-myeloid cell line. Different combinations of plasmids encoding Nlrp1b (allele 1), pro-caspase-1, and pro-il-1β were transfected into HT1080 fibroblasts. Approximately 24 h after transfection, cells were treated with LeTx (10-8 M LF and 10-8 M PA) for 3 h and cell lysates and supernatants were then probed for HA-tagged IL-1β by immunoblotting. Increased levels of pro-il-1β were observed in the cytosolic lysates of unintoxicated cells that had been transfected with Nlrp1b compared to the lysates of cells that had not been transfected with Nlrp1b (Fig. 3.1A), although the significance of this observation is not clear. The 17 kda mature form of IL-1β was detected in the supernatant of cells transfected with plasmids encoding Nlrp1b and pro-caspase-1 only if the cells had been treated with LeTx (Fig. 3.1A). The level of the 35 kda pro-il-1β in these cells was lower than in the corresponding unintoxicated cells, which is consistent with pro-il-1β being 54

64 processed and secreted. Neither processing of pro-il-1β nor secretion of IL-1β was observed when cells were transfected with plasmids encoding Nlrp1b or pro-caspase-1 alone. These results suggest that LeTx induces the assembly of a functional Nlrp1b inflammasome in transfected fibroblasts. LeTx induces pyroptosis in macrophages expressing Nlrp1b allele 1 or 5, but not in macrophages expressing allele 2, 3 or 4 (Boyden and Dietrich, 2006). To determine whether a resistant allele would be activated by LeTx in HT1080 cells, plasmids containing pro-caspase-1 and pro-il-1β were co-transfected with either a plasmid containing allele 1 or allele 3 of Nlrp1b. The transfected cells were treated with LeTx for 3 h and cell lysates and supernatants were probed for HA-tagged IL-1β. Consistent with the allele specific activation of the inflammasome in macrophages, LeTx caused the processing and secretion of IL-1β in cells that expressed Nlrp1b allele 1, but not allele 3 (Fig 3.1B). The expression levels of Nlrp1b allele 1 and allele 3 were similar in HT1080 cells; the presence of a higher mobility band observed for Nlrp1b allele 1 suggested the presence of a protease-sensitive site in this protein (Fig. 3.1C). This processing may result from expressing the protein in human cells as the higher mobility form was not detected in murine macrophages (data not shown). To confirm that the enzymatic activity of LF is required to activate the Nlrp1b inflammasome, LF-E687A, which lacks catalytic activity because of an active site 55

65 mutation, was used in the assay. As above, pro-il-1β was processed and IL-1β was secreted into the medium in the presence of wild-type LF and PA. When the cells were treated with PA and LF-E687A, PA alone, or LF alone, the level of pro-il-1β remained constant in cells and mature IL-1β was not observed in the medium (Fig. 3.1D). This result indicates that the enzymatic activity of LF is required to activate the Nlrp1b inflammasome. Proteasome activity is involved in LeTx-mediated macrophage death (Squires et al., 2007), so to test whether the proteasome is required to activate the Nlrp1b inflammasome in fibroblasts, transfected cells were co-treated with LeTx and the proteasome inhibitor MG-132. The mature form of IL-1β was detected in the medium of cells treated with LeTx, but not in the medium of cells treated with LeTx and MG-132 (Fig. 3.1E). The level of pro-il-1β in the cell lysates did not diminish in LeTx-treated cells exposed to MG-132, suggesting that proteasome activity is required for inflammasome activation. 56

66 Figure 3.1 Reconstitution of the Nlrp1b inflammasome in HT1080 cells. (A) Combinations of pntap-nlrp1b (1µg), pcdna3-pro-caspase-1-t7 (1µg) and pcdna3-pro-il-1β-ηα (1µg) were transfected into HT1080 cells. Approximately 24 h after transfection, cells were treated with LeTx (10-8 M LF and 10-8 M PA) for 3 h and cell lysates were probed for HA-tagged pro-il-1β and β-actin by immunoblotting (IB); supernatants were immunoprecipitated (IP) with α-ha antibodies and then probed for HA-tagged IL-1β by immunoblotting. (B) Plasmids pcdna3-pro-caspase-1-t7 (1µg) and pcdna3-pro-il-1β-ha (1µg) were co-transfected with either pntap-nlrp1b allele 1 (1µg) or allele 3 (1µg) into HT1080 cells. Cells were treated with LeTx and HA-tagged IL-1β was detected as described above. (C) Cells were transfected either with pntap-nlrp1b allele 1 or allele 3. Approximately 24 h after transfection, cells were lysed and TAP-tagged proteins were precipitated using streptavidin resin and subjected to Western blotting using an antibody against calmodulin binding peptide to detect the TAP tag. (D) Cells were transfected with plasmids pntap-nlrp1b (1µg), pcdna3-pro-caspase-1-t7 (1µg) and pcdna3-pro-il-1β-ha (1µg). Transfected cells were left untreated or were treated with indicated combinations of PA (10-8 M), LF (10-8 M), and LF-E687A (10-8 M). After 3 h, HA-tagged IL-1β was detected as described above. (E) HT1080 cells were transfected with plasmids pntap-nlrp1b (1µg), pcdna3-pro-caspase-1-t7 (1µg) and pcdna3-pro-il-1β-ηα (1µg). Transfected cells were co-treated with LeTx 57

67 (10-8 M LF and 10-8 M PA) in the absence or presence of MG-132 (10µM). After 3 h, HA-tagged IL-1β was detected as described above. Blots are representative of three independent experiments Characterization of LeTx-induced cell death and IL-1β release I next sought to determine whether the activation of the Nlrp1b inflammasome by LeTx caused death of HT1080 cells. Cells were transfected with different combinations of plasmids encoding Nlrp1b, pro-caspase-1, and pro-il-1β. Approximately 24 h after transfection, the cells were treated with LeTx for 6 h and then the supernatants were collected and assayed for lactate dehydrogenase (LDH) activity. LDH activity was observed in the supernatants of only those cells transfected with Nlrp1b and pro-caspase-1 and then treated with LeTx (Fig. 3.2A). Co-transfection of pro-il-1β did not increase LDH activity in the supernatant. These data suggest that activation of the Nlrp1b inflammasome leads to both IL-1β secretion and cell death, but that IL-1β does not stimulate cell death. Previously published work has indicated that IL-1β release from LeTx-treated macrophages occurs through cell lysis, rather than through a secretory system (Wickliffe et al., 2008). To address whether IL-1β release coincides with cell death in fibroblasts, we performed time course experiments. Transfected cells were treated with LeTx for up to 6 h and the cell supernatants were assayed for LDH activity (Fig. 58

68 3.2B) and IL-1β protein (Fig. 3.2C). The release of LDH and IL-1β occurred with similar kinetics. I next addressed whether osmoprotection of cells would prevent release of IL-1β by incubating cells with polyethylene glycol (PEG). Addition of PEG200, PEG2000, or PEG6000 to cells did not affect the release of IL-1β (Fig. 3.2D). PEG6000, did however, prevent the release of LDH (Fig. 3.2E). These data indicate that caspase-1 activity leads to the formation of membrane pores with diameters between 2.8 nm (PEG2000) and 5 nm (PEG 6000) (Dacheux et al., 2001) and that IL-1β release does not require cell lysis. 59

69 Figure 3.2 Characterization of LeTx-induced cell death and IL-1β release. (A) Different combinations of plasmids pntap-nlrp1b (1µg), pcdna3-pro-caspase-1-t7 (1µg) and pcdna3-pro-il-1β-ηα (1µg) were transfected into HT1080 cells. Approximately 24 h after transfection, cells were treated with LeTx (10-8 M LF and 10-8 M PA) for 6 h and supernatants were assayed for LDH activity. (B and C) Cells were transfected with plasmids pntap-nlrp1b (1µg), pcdna3-pro-caspase-1-t7 (1µg) and pcdna3-pro-il-1β-ηα (1µg). Transfected cells were left untreated or treated with LeTx (10-8 M LF and 10-8 M PA) for 1, 2, 3, or 6 h. Supernatants were assayed for LDH activity (B); supernatants and cell lysates were probed for HA-tagged IL-1β by immunoblotting (C). (D and E) Plasmids encoding pntap-nlrp1b (1µg), pcdna3-pro-caspase-1-t7 (1µg) and pcdna3-pro-il-1β-ηα (1µg) were transfected into HT1080 cells. Approximately 24 h after transfection, cells were co-treated with LeTx (10-8 M LF and 10-8 M PA) and indicated osmoprotectants (15 mm) for 3 h and supernatants and cell lysates were 60

70 probed for HA-tagged IL-1β by immunoblotting (D); the supernatants were assayed for LDH activity (E). Results shown are representative of three independent experiments. Error bars represent the standard deviation The CARD domain, but not the NACHT or LRR domains, is required for inflammasome activity To determine the functional importance of the Nlrp1b domains, three deletion mutants were made: Nlrp1b (deletion of the amino-terminal region and the NACHT domain), Nlrp1bΔ (LRR domain deletion), and Nlrp1b (CARD domain deletion) (Fig. 3.3A, B). Each construct was co-transfected with plasmids encoding pro-caspase-1 and pro-il-1β. Transfected cells were treated with LeTx for 3 h and pro-il-1β processing and secretion were monitored. In contrast to full-length Nlrp1b, expression of Nlrp1b or Nlrp1bΔ resulted in the secretion of IL-1β in the absence of LeTx (Fig. 3.3C). Nlrp1b was not able to activate caspase-1 as no active form of IL-1β was detected in the supernatant in the presence or absence of LeTx. These data suggest that the CARD domain is necessary for Nlrp1b to activate caspase-1; the NACHT and LRR domains are not necessary for caspase-1 activation and may function in autoinhibition. 61

71 Figure 3.3 Deletion analysis of Nlrp1b. (A) Domain structures of various Nlrp1b deletion constructs. (B) HT1080 cells were transfected with various TAP-tagged Nlrp1b deletion constructs. Approximately 24h after transfection, cells were lysed and TAP-tagged proteins were precipitated using streptavidin resin and immunoblotted using antibody directed against the calmodulin binding peptide segment of the TAP tag. (C) Nlrp1b deletion constructs were co-transfected with plasmids pcdna3-pro-caspase-1-t7 (1µg) and pcdna3-pro-il-1β--ηα (1µg) into HT1080 cells. After 24 h, the cells were left untreated or were treated with LeTx (10-8 M LF and 10-8 M PA) for 3 h. Cell lysates were probed for HA-tagged pro-il-1β and β-actin by immunoblotting; supernatants were immunoprecipitated with α-ha antibodies and then probed for HA-tagged IL-1β by immunoblotting. Blots are representative of three independent experiments. 62

72 3.2.4 Amino-terminal truncation mutants of Nlrp1b are constitutively active I next made a series of Nlrp1b deletion mutants to define the region that can constitutively activate caspase-1 (Fig. 3.4A, B). Expression of either Nlrp1b , Nlrp1b , or Nlrp1b caused the secretion of IL-1β (Fig. 3.4C). In contrast, I did not detect IL-1β in the supernatant of cells tranfected with the Nlrp1b plasmid. This correlated with the observation that expression of Nlrp1b , but not Nlrp1b , led to LDH release (Fig. 3.4D). Thus, a fragment of Nlrp1b containing the CARD domain and 56 amino acids amino-terminal to the CARD domain activates caspase-1, whereas the CARD domain alone does not exhibit any detectable activity. The role of the proteasome in mediating events downstream of inflammasome activation was assessed by treating cells expressing Nlrp1b with MG-132. IL-1β was detected in the medium of Nlrp1b expressing cells 6 h after transfection (Fig. 3.4E); the level of IL-1β was higher by 10 h post-transfection whether or not MG-132 was added to the cells at 6 h post-transfection. In contrast, IL-1β was detected in the supernatants of cells expressing full-length Nalp1b at 10 h if toxin was added at 6 h, but no increase in the IL-1β level was detected if MG-132 was added with the toxin. This lack of involvement of proteasome activity 63

73 downstream of inflammasome activation was confirmed by monitoring the release of LDH the increase in LDH activity detected in the supernatant of Nlrp1b expressing cells between 6 h and 10 h post-transfection was not blocked by MG-132 (Fig. 3.4F). 64

74 Figure 3.4 Amino-terminal truncation mutants of Nlrp1b are constitutively active. (A) Domain structures of Nlrp1b deletion constructs. (B) HT1080 cells were transfected with various Nlrp1b deletion constructs. Approximately 24 h after transfection, cells were lysed and TAP-tagged proteins were precipitated using streptavidin resin and immunoblotted using antibody directed against the calmodulin binding peptide segment of the TAP tag. (C) Cells were transfected with plasmids containing different Nlrp1b fragments, pro-caspase-1 and pro-il-1β. Approximately 24 h after transfection, cell lysates were probed for HA-tagged pro-il-1β and β-actin by immunoblotting; supernatants were immunoprecipitated with α-ha antibodies and then probed for HA-tagged IL-1β by immunoblotting. (D) Cells were transfected with plasmids containing Nlrp1b fragments, pro-caspase-1 and pro-il-1β. 65

75 Approximately 24 h after transfection, supernatants were assayed for LDH release. (E and F) Cells were transfected with plasmids encoding indicated proteins and after 6 h, cells were treated with MG-132 and/or LeTx for an additional 4 h. Supernatants were probed for IL-1β-HA and LDH at either 6 h or 10 h post-transfection, as indicated. Results shown are representative of three independent experiments. Error bars represent the standard deviation Nlrp1b and Nlrp1b interact with pro-caspase-1, but only Nlrp1b self-associates To determine why Nlrp1b , but not Nlrp1b , induced the secretion of IL-1β, we performed an in vitro binding assay to determine whether the two Nlrp1b constructs could interact with pro-caspase-1. Glutathione-S-transferase (GST) fusions of the Nlrp1b fragments were used to precipitate the catalytically inactive pro-caspase-1-c284a-t7 mutant from HT1080 cell lysates. Both GST-Nlrp1b and GST-Nlrp1b bound pro-caspase-1-c284a-t7, while the GST control did not (Fig. 3.5A). Although Nlrp1b precipitated slightly more pro-caspase-1-c284a-t7 than Nlrp1b did, this result suggests that the inability of the CARD domain construct to promote secretion of IL-1β is not due to an inability to bind pro-caspase-1. The proximity model of caspase activation posits that mediator proteins activate molecules of pro-caspases by bringing them together to facilitate a trans proteolytic reaction (Boatright et al., 2003). Thus, I speculated that Nlrp1b , but not Nlrp1b , would self-associate. To test this notion, GST fusion proteins were 66

76 used in binding assays with HT1080 cell extracts containing either His-Nlrp1b HA or His-Nlrp1b HA. His-Nlrp1b HA was precipitated by GST-Nlrp1b , but no detectable amount of His-Nlrp1b HA was pulled down by GST-Nlrp1b (Fig. 3.5B). Co-immunoprecipitation experiments were then performed to confirm that Nlrp1b self-associates and Nlrp1b does not. His-Nlrp1b HA was expressed with His-Nlrp1b T7 in HT1080 cells. The T7-tagged protein was immunoprecipitated along with the HA-tagged Nlrp1b (Fig. 3.5C). In contrast, co-immunoprecipitation was not observed between the Nlrp1b constructs (Fig. 3.5D). Taken together, these experiments indicate that Nlrp1b associates with itself, but Nlrp1b does not. 67

77 Figure 3.5 Nlrp1b and Nlrp1b interact with pro-caspase-1, but only Nlrp1b self-associates. (A) GST, GST-Nlrp1b and GST-Nlrp1b were immobilized on glutathione-sepharose beads and incubated with mammalian cell lysates containing pro-caspase-1-c284a-t7. Precipitated proteins and 5% of the input lysates were subjected to Western blot analysis using α-caspase-1 p10 antibody. (B) GST, GST-Nlrp1b and GST-Nlrp1b were immobilized on glutathione-sepharose beads and incubated with mammalian cell lysates containing either His-Nlrp1b HA or His-Nlrp1b HA. Precipitated proteins and 5% of the input lysates were subjected to Western blot analysis using α-ha antibodies. (C) HT1080 cells were transfected with His-Nlrp1b HA, His-Nlrp1b T7, or both constructs. Cells were lysed 24 h after transfection and proteins were immunoprecipitated using α-ha antibody, followed by immunoblotting with α-t7 antibody. (D) HT1080 cells were transfected with His-Nlrp1b HA, His-Nlrp1b T7, or both constructs. Cells were lysed 24 h after transfection and proteins were immunoprecipitated using α-ha antibody, followed by immunoblotting with α-t7 antibody. Blots are representative of three independent experiments. 68

78 3.3 Discussion Numerous studies indicate that NLRs detect microbial products or endogenous danger signals, but the identification of the molecules that directly activate NLRs has proven to be difficult (Kanneganti et al., 2007; Mariathasan and Monack, 2007). The activation of Nlrp1b by LeTx is dependent on the proteolytic activity of the toxin, but neither the LeTx substrate, nor the Nlrp1b ligand involved in activation is known. Murine dendritic cells and macrophages are the only cell types that have been shown to undergo pyroptosis upon treatment with LeTx, which could be because expression of Nlrp1b is restricted to these cell types, but could also be because factors involved in activation are missing in other cell types. I have shown here that transient expression of Nlrp1b and pro-caspase-1 is sufficient to sensitize human fibroblasts to LeTx-induced pyroptosis. Thus, the activation pathway appears to be conserved in human cells and is not dependent on myeloid cell-specific proteins. That this heterologous system reflects how Nlrp1b is activated in murine macrophages is supported by the observation that proteasome activity is required for inflammasome activation in both cases (Fig. 3.1E and 3.4E and 3.4F) (Squires et al., 2007). The role of the proteasome in this process is not known, but its involvement suggests that Nlrp1b is not a direct target of LF and rather that the proteasome might degrade a negative regulator of Nlrp1b. A second observation that indicates the 69

79 fidelity of the heterologous system is the demonstration of Nlrp1b allele specificity for function. Nlrp1b allele 1, which supports pyroptosis in macrophages, supported pro-il-1β processing in fibroblasts, whereas allele 3, found in LeTx-resistant macrophages, did not. There are a number of amino acid differences between alleles 1 and 3 making it difficult to speculate on why allele 3 is not able to detect LeTx activity and/or to assemble into a functional inflammasome. I note that the CARD domains of the two alleles are identical, indicating that the defect in allele 3 is not due to an inability to bind pro-caspase-1. Activation of caspase-1 not only caused IL-1β secretion, it induced death of the HT1080 cells. Studies using murine macrophages have suggested that IL-1β release results from cell lysis, precluding the release of this cytokine as a cause of death (Wickliffe et al., 2008). My results indicated that IL-1β secretion does not require cell lysis because the osmoprotectant PEG6000 prevented LDH release, but not IL-1β secretion. LDH release was dependent on the transfection of Nlrp1b and pro-caspase-1, but not pro-il-1β, which is consistent with the notion that IL-1β does not mediate cell death. A difference between what has been observed in macrophages and in transfected fibroblasts is the kinetics of cell death macrophages are killed within min of toxin treatment (Squires et al., 2007; Wickliffe et al., 2008), whereas the transfected fibroblasts died over the course of several hours. 70

80 This difference in the kinetics of cell death could be a result of differences in the expression levels of inflammasome components or downstream mediators of cell death, such as Bnip3 (Ha et al., 2007). I demonstrated that the CARD domain of Nlrp1b was essential for inflammasome activity. This finding was not surprising because unlike human Nlrp1, Nlrp1b lacks a pyrin domain at its amino-terminus to recruit pro-caspase-1 through the adaptor ASC. ASC is not likely required for Nlrp1b inflammasome function as the LeTx-sensitive RAW264.7 cell line does not express ASC at detectable levels (Pelegrin et al., 2008) and I found that co-transfection of ASC into HT1080 cells did not stimulate inflammasome activity (data not shown). Deletion of the LRR domain yielded a constitutively active Nlrp1b mutant. This finding is consistent with an autoinhibition model in which an interaction between the LRR domain and the NACHT domain holds Nlrp1b in an inactive conformation that is relieved when the LRR domain detects an activating signal. Deletion of the NACHT domain also yielded a constitutively active mutant, indicating that oligomerization of the NACHT domain is not required for processing of pro-il-1β, at least in the context of a truncated protein. These data agree with a study on human NLRP1 that showed that a truncation mutant that lacked the NACHT domain was able to cause cell death (Hlaing et al., 2001). 71

81 I made a series of truncation mutants to delimit the region of Nlrp1b required for constitutive activity. Nlrp1b , which contains the CARD domain and the adjacent 56 amino acids, activated pro-caspase-1, whereas the CARD domain alone did not. Both of these constructs bound pro-caspase-1, demonstrating that a CARD-pro-caspase-1 interaction is not sufficient to activate pro-caspase-1. The 56 amino acids adjacent to the CARD domain promoted the self-association of the truncation mutant in GST pulldown and co-immunoprecipitation experiments. These results suggest that a segment of the FIIND domain induces the close proximity of pro-caspase-1 molecules that promotes trans-proteolysis. A similar region exists adjacent to the CARD domains of human Nlrp1 and CARD8 (TUCAN/CARDINAL) (Agostini et al., 2004; Razmara et al., 2002) and might also enhance self-association of these proteins. 72

82 CHAPTER Energy stress activates the NLRP1b inflammasome Contributions: I performed all the experiments presented in this chapter. 4.1 Summary The efficacy of the innate immune system depends on its ability to mount an appropriate response to diverse infections and damaging agents. Key components of this system are pattern recognition receptors that detect pathogen associated- and damage associated- molecular patterns. Nlrp1b is a pattern recognition receptor that forms a caspase-1 activation platform, known as an inflammasome, in response to sensing the proteolytic activity of anthrax lethal toxin. Here, I demonstrate that Nlrp1b also becomes activated in cells that are subjected to energy stress. The energy-sensor AMPK promoted inflammasome activation, although activation of AMPK in the absence of ATP depletion was not sufficient for casapase-1 mediated pro-il-1β processing. Surprisingly, I found that mutation of the ATP-binding motif of Nlrp1b caused constitutive activation, suggesting that ATP might inhibit the Nlrp1b inflammasome rather than being required for its assembly. My results 73

83 provide a novel link between cellular metabolism and the innate immune system. 4.2 Results ATP depletion activates Nlrp1b To determine whether reduction of cytosolic ATP can activatate the Nlrp1b inflammasome, I used a reconstituted system in which the HT1080 human fibroblast cell line was transfected with Nlrp1b, pro-caspase-1 and pro-il-1β. I have shown previously that LeTx induces Nlrp1b inflammasome assembly as monitored by the cleavage of pro-il-1β by caspase-1 and the subsequent release of processed IL-1β (see Chapter 3). To inhibit the production of ATP, I treated cells with the glycolysis inhibitor 2-deoxyglucose (2DG) and an inhibitor of the mitochondrial electron transport chain, sodium azide. Treatment of cells with these two compounds greatly reduced cytosolic ATP and caused the release of IL-1β (Fig. 4.1A and 4.1B). Two forms of IL-1β were secreted, which correspond in molecular weight to cleavage events at the canonical site, yielding a 17 kda form, and an amino-terminal site, yielding a ~25 kda form. Secretion of IL-1β was observed from cells expressing Nlrp1b allele 1, but not allele 3. The allele specificity for 74

84 inflammasome activation by ATP depletion coincides with that observed in LeTx-mediated inflammasome activation in HT1080 cells (Fig. 4.1A) and in murine macrophages (Boyden and Dietrich, 2006). LeTx also causes release of pro-il-1β into the medium, which was likely through passive release because the release of pro-il-1β, but not processed IL-1β, was inhibited by the addition of the osmotic stabilizer PEG-6000 (data not shown). Of note, as shown in Fig. 4.1B, unlike 2DG and NaN 3 treatment, LeTx did not reduce ATP levels, which suggests that the two stimuli represent distinct Nlrp1b activation signals. To determine the threshold level of cytosolic ATP that triggers inflammasome activation, I grew cells in minimal medium with varying concentrations of either glucose or 2DG. IL-1β was released from cells grown in 1 mm glucose, but not from cells grown in 10 mm glucose. This reduction of glucose in the medium caused a drop in cytosolic ATP levels from ~90% to ~80% of that observed in control cells grown in complete medium. Further decreasing the concentration of glucose in the medium or adding 2DG caused a greater reduction in ATP levels and increased the release of IL-1β. Additionally, hypoxia reduced intracellular ATP level to ~50% in cells grown in minimal medium. As expected, hypoxia induced pro-il-1β processing and secretion from cells expressing Nlrp1b allele 1 but not allele3. 75

85 Figure 4.1 ATP depletion activates the Nlrp1b inflammasome. (A) Plasmids pcdna3-pro-caspase-1-t7 (1µg) and pcdna3-pro-il-1β-ηα (1µg) were co-transfected with either pntap-nlrp1b allele 1 (1µg) or allele 3 (1µg) into HT1080 cells. Approximately 24 h after transfection, cells were treated with LeTx (10-8 M LF and 10-8 M PA) or 50 mm 2DG and 10mM NaN 3 in 5mL medium for 3 h. The cell lysates were probed for HA-tagged pro-il-1β and β-actin by immunoblotting; supernatants were imunoprecipitated with anti-ha antibodies and then probed for HA-tagged IL-1β and pro-il-1β by immunoblotting. (B) HT1080 cells were transfected with 1 µg pcdna3-pro-caspase-1-t7, 1µg pcdna3-pro-il-1β-ha and 1µg pntap-nlrp1b allele 1 or allele 3. Approximately 24 h after transfection, these cells were trypsinized and 1x10 5 cells were seeded in each well in 96-well plate. After 76

86 2h, these transfected cells were treated with LeTx (10-8 M LF and 10-8 M PA) or 50 mm 2DG and 10 mm NaN 3 for 3h, and cell lysates were assayed for intracellular ATP level. (C) Cells were transfected with plasmids encoding pcdna3-pro-caspase-1-t7 (1µg), pcdna3-pro-il-1β-ha (1µg), and pntap-nlrp1b (1µg). These transfected cells were trypsinized and 1x10 5 cells were seeded per well in a 96-well plate. After 2 h, these cells were left in minimal DMEM (20mM HEPES, ph 7.5) containing indicated concentrations of 2DG or glucose for 3 h. The cell lysates were then assayed for intracellular ATP. (D) Plasmids encoding NTAP-Nlrp1b (1µg), pro-caspase-1-t7 (1µg) and pro-il-1β-ha were transfected into HT1080 cells. Approximately 24 h after transfection, cells were left in minimal DMEM (20 mm HEPES, ph 7.5) with indicated concentrations of 2DG or glucose. After 3 h treatment, HA-tagged pro-il-1β and IL-1β were detected as described as above. (E) Cells were transfected with plasmids encoding pcdna3-pro-caspase-1-t7 (1µg), pcdna3-pro-il-1β-ηα (1µg), and pntap-nlrp1b allele 1 or 3(1µg). These transfected cells were trypsinized and 1x10 5 cells were seeded per well in a 96-well plate. After 2 h, these cells were left in minimal DMEM (20mM HEPES, ph 7.5) and incubated under normoxic or hypoxic (1% O 2 ) conditions for 3 h. The cell lysates were then assayed for intracellular ATP. (F) Plasmids encoding NTAP-Nlrp1b allele 1 or 3 (1µg), pro-caspase-1-t7 (1µg) and pro-il-1β-ha were transfected into HT1080 cells. Approximately 24 h after transfection, cells were left in minimal DMEM (20 mm HEPES, ph 7.5) with indicated conditions. After 3 h treatment, HA-tagged pro-il-1β and IL-1β were detected as described as above. The asterisk indicates a ~25 kda HA-tagged pro-il-1β cleaved product. Results shown are representative of at least three independent experiments. Error bars represent the standard deviation AMPK promotes inflammasome activation Depletion of cellular ATP is sensed by the master regulator of metabolism, AMPK (Carling et al., 2011). AMPK is phosphorylated when a high AMP:ATP ratio exists in cells and in turn phosphorylates numerous substrates to increase catabolic and decrease anabolic processes. To determine whether AMPK signaling is involved in the activation of Nlrp1b by ATP depletion or LeTx, we used sirna to reduce AMPK 77

87 levels (Fig. 4.2A). Treatment of cells with a combination of 2DG and sodium azide, but not LeTx, caused the phosphoryation of AMPK in control cells. In cells treated with AMPK sirna, there was a reduced level of phosphorylated AMPK and less IL-1β was secreted compared to control cells. In contrast, similar amounts of IL-1β were secreted by control and AMPK knock-down cells in response to LeTx. In order to confirm that AMPK facilitated ATP depletion-induced Nlrp1b activation, a dominant negative form of AMPKα1 subunit was overexpressed with Nlrp1b, pro-caspase-1 and pro-il-1β. In Figure 4.2B, it is shown that overexpression of AMPKα1 T174A diminished AMPK phosphorylation and reduced IL-1β secretion in response to 2DG and NaN 3 but not LeTx. I next assessed whether signaling by AMPK is sufficient to activate Nlrp1b by treating cells with staurosporine. Staurosporine is a kinase inhibitor that causes activation of AMPK in the absence of ATP depletion (Fig. 4.2C and 4.2D). Staurosporine did not cause the release of IL-1β from cells. 78

88 Figure 4.2 AMPK facilitates Nlrp1b inflammasome activation. (A) HT1080 cells were first transfected with either 15 µl control sirna (10µM) or 15 µl AMPK sirna (10 µm). After 24 h, these cells were again transfected with 1µg pcdna3-pro-caspase-1-t7, 1µg pcdna3-pro-il-1β-ha and 1µg pntap-nlrp1b. Approximately 24 h after the second transfection, cells were treated with LeTx (10-8 M LF and 10-8 M PA) or 50 mm 2DG and 10 mm NaN 3 in 5mL medium for 3h. Cell lysates were probed for phospho-ampk, total-ampk, HA-tagged IL-1β and β-actin by immunoblotting. Supernatants were immunoprecipitated with anti-ha antibodies and probed for HA-tagged IL-1β and pro-il-1β by immunoblotting. (B) Plasmids pntap-nlrp1b (1µg), pcdna3-pro-caspase-1-t7 (1µg) and pcdna3-pro-il-1β-ηα (1µg) were co-transfected with pcdna3-his*6-flag*3 vector (2µg), pcdna3-his*6-ampkα1 WT-FLAG*3 (2µg), or pcdna3-his*6-ampkα1 T174A-FLAG*3 (2µg) into HT1080 cells. Approximately 24 h after transfection, cells were treated with LeTx (10-8 M LF and 10-8 M PA) or 50 79

89 mm 2DG and 10mM NaN 3 in 5mL medium for 3 h. Cell lysates were probed for phospho-ampk, total-ampk, HA-tagged IL-1β and β-actin by immunoblotting; supernatants were imunoprecipitated with anti-ha antibodies and then probed for HA-tagged IL-1β and pro-il-1β by immunoblotting. (C) Plasmids encoding pro-caspase-1-t7 (1µg), pro-il-1β-ha (1µg), and NTAP-Nlrp1b (1µg) were transfected into HT1080 cells. These transfected cells were treated with LeTx (10-8 M LF and 10-8 M PA), 50 mm 2DG and 10 mm NaN 3, 2 µm staurosporine (STS). After 3 h, cell lysates were harvested and probed for phospho-ampk, total-ampk, HA-tagged IL-1β and β-actin by immunoblotting. HA-tagged pro-il-1β and IL-1β in the supernatants were detected as described above. (D) Cells were transfected with 1 µg pcdna3-pro-caspase-1-t7, 1 µg pcdna3-pro-il-1β-ha and 1µg pntap-nlrp1b. Approximately 24 h after transfection, these cells were trypsinized and 1x10 5 cells were seeded per well in a 96-well plate. After 2 h, these transfected cells were treated with LeTx (10-8 M LF and 10-8 M PA), 50 mm 2DG and 10mM NaN 3, or STS (2µM) for 3 h, and cell lysates were assayed for intracellular ATP. The asterisk indicates a ~25 kda HA-tagged pro-il-1β cleaved product. Results shown are representative of three independent experiments. Error bars represent the standard deviation Inhibition of the proteasome impairs inflammasome activation Several studies have suggested that reactive oxygen species (ROS) generated from the mitochondrial electron transport chain are required to trigger NLRP3 assembly downstream of diverse stimuli such as extracellular ATP, particulates, and palmitate (Wen et al., 2011; Zhou et al., 2011). The level of ROS required may be derived from damaged mitochondria or from excessive oxidative phosphorylation. Recently, however, the requirement for ROS has been suggested to be at the priming stage the expression of NLRP3 rather than the activation stage (Bauernfeind et al., 2011b). To address whether activation of the Nlrp1b inflammasome required ROS, I 80

90 pre-treated transfected HT1080 cells with the non-specific ROS scavenger N-actetylcysteine (NAC). I found that NAC did not inhibit IL-1β secretion from cells treated with either LeTx or 2DG (Fig. 4.3A), nor did NAC affect cytosolic ATP levels (Fig. 4.3B). It has been established that proteasome inhibitors block the activation of Nlrp1b by LeTx (Squires et al., 2007; Tang and Leppla, 1999), so I sought to determine whether this was also the case for activation by ATP depletion. The proteasome inhibitor MG-132 blocked the secretion of IL-1β in response to both LeTx and 2DG/sodium azide (Fig. 4.3C). MG-132 did not, however, prevent the depletion of cytosolic ATP by 2DG/sodium azide (Fig. 4.3D), suggesting that proteasome activity is required to mediate Nlrp1b inflammasome activity triggered by both LeTx and ATP depletion. 81

91 Figure 4.3 Role of reactive oxygen species and proteasome activity in ATP-depletion induced Nlrp1b activation. (A) Plasmids pc-dna3-procaspase-1-t7 (1 µg), pcdna3-pro-il-1β-ha (1 µg), and pntap-nlrp1b (1 µg) were transfected into HT1080 cells. Approximately 24 h after 82

92 transfection, cells were left untreated or treated with LeTx (10-8 M LF and 10-8 M PA), 50 mm 2DG and 10 mm NaN 3, or 50 mm 2DG in the absence or presence of 25mM N-acetyl cysteine (NAC). After 3 h, cell lysates were collected and probed for HA-tagged pro-il-1β and β-actin by immunoblotting; supernatants were immunoprecipitated with anti-ha antibodies and probed for HA-tagged pro-il-1β and IL-1β by immunoblotting. (B) HT1080 cells were transfected with 1 µg pcdna3-pro-caspase-1-t7, 1 µg pcdna3-pro-il-1β-ha and 1 µg pntap-nlrp1b. These transfected cells were trypsinized and 1x10 5 cells were seeded per well in a 96-well plate. After 2 h, these cells were left untreated or treated with LeTx (10-8 M LF and 10-8 M PA), 50 mm 2DG and 10 mm NaN 3, or 50 mm 2DG in the absence or presence of NAC (25mM) for 3 h. Cell lysates were assayed for intracellular ATP. (C) Cells were transfected pcdna3-pro-caspase-1-t7 (1µg), pcdna3-pro-il-1β-ha (1µg) and pntap-nlrp1b (1 µg). Approximately 24 h after transfection, cells were treated with LeTx (10-8 M LF and 10-8 M PA) or 50 mm 2DG and 10mM NaN 3 in presence or absence of MG-132 (10 µm). After 3 h, cell lysates were collected and probed for HA-tagged pro-il-1β; HA-tagged pro-il-1β and IL-1β in supernatants were detected as described above. (D) Plasmids encoding pro-caspase-1-t7 (1 µg), pro-il-1β-ha (1 µg), and NTAP-Nlrp1b (1 µg) were transfected into HT1080 cells. These transfected cells were trypsinized and 1x10 5 cells were seeded in each well in 96-well plate. After 2 h, these cells were treated with LeTx (10-8 M LF and 10-8 M PA) or 50 mm 2DG and 10 mm NaN 3 in the absence or presence of MG-132 (10 µm) for 3 h. Cell lysates were assayed for intracellular ATP. The asterisk indicates a 26 kda HA-tagged pro-il-1β cleaved product. Results shown are representative of three independent experiments. Error bars represent the standard deviation Mutation of the Nlrp1b Walker A motif causes constitutive activation The NACHT domains of NLRPs are thought to mediate homo-oligomerization by using energy derived from ATP (Duncan et al., 2007). Because this notion seemed counter-intuitive for a sensor of low ATP levels, I mutated the Walker A motif of the Nlrp1b NACHT domain to assess its involvement in inflammasome activation. In contrast to wild-type Nlrp1b, the Walker A mutant was constitutively active, 83

93 suggesting that ATP hydrolysis is not required to form a functional inflammasome (Fig. 4.4A and 4.4B). I next measured NLRP3 activity by using a constitutively active mutant identified in patients with the auto-inflammatory disorder Muckles-Wells syndrome (Aganna et al., 2002). NLRP3-R258W caused secretion of IL-1β in the presence of the ASC adaptor, but not in its absence (Fig. 4.4C and 4.4D). Mutation of the Walker A motif in NLRP3-R258W impaired its activity, indicating that a functional ATPase domain is required for NLRP3 function (Fig. 4.4E and 4.4F). These data suggest that the NACHT domains of Nlrp1b and NLRP3 have different requirements for ATP. 84

94 Figure 4.4 Walker A motif mutant of Nlrp1b is constitutively active. (A) HT1080 cells were transfected with 1 µg pcdna3-pro-caspase-1-t7, 1 µg pcdna3-pro-il-1β-ha and 1µg pntap-nlrp1b, either wild type (WT) or Walker A (WA) mutant. Approximately 24 h after transfection, cells were lysed and TAP-tagged proteins were precipitated using streptavidin resin and immunoblotted using antibody directed against the calmodulin binding peptide segment of the TAP 85

95 tag. (B) Plasmids encoding pro-caspase-1-t7 (1 µg), pro-il-1β-ha (1 µg) were co-transfected into cells with either 1 µg pntap-nlrp1b WT or 1 µg pntap-nlrp1b-wa. Approximately 24 h after transfection, cell lysates were collected and probed for HA-tagged IL-1β and β-actin by immunoblotting; supernatants were immunoprecipitated with anti-ha antibodies and probed for HA-tagged pro-il-1β and IL-1β by immunoblotting. (C) Cells were transfected with 1 µg pcdna3-pro-caspase-1-t7, 1 µg pcdna3-pro-il-1β-ha and 1 µg indicated NTAP-Nlrp3 plasmids along with 50 ng pcdna3-flag-asc or 50 ng empty vector. Approximately 24 h after transfection, TAP-tagged proteins were detected as described above. (D) Plasmids encoding indicated proteins were co-trasnfected into HT1080 cells with 1 µg pcdna3-pro-caspase-1-t7 and 1 µg pcdna3-pro-il-1β-ha. Approximately 24 h after transfection, cell lysates were harvested and probed for HA-tagged IL-1β and β-actin; supernatants were probed for HA-tagged pro-il-1β and IL-1β as described above. (E) Plasmids pcdna3-flag-asc (50 ng), pcdna3-pro-caspase-1-t7 (1µg), pcdna3-pro-il-1β-ha (1 µg) and various pntap-nlrp3 (1 µg) were transfected into HT1080 cells. TAP-tagged proteins were detected as described above. (F) Cells were transfected with 50 ng pcdna3-flag-asc, 1µg pcdna3-pro-caspase-1-t7, 1 µg pcdna3-pro-il-1β-ha and 1 µg pntap-nlrp3. HA-tagged IL-1β and pro-il-1β were detected as described above. The asterisk indicates a 26 kda HA-tagged pro-il-1β cleaved product. Results shown are representative of three independent experiments. 86

96 4.3 Discussion The efficacy of the innate immune system depends on its ability to detect a diversity of infections. It does so by using pattern recognition receptors to sense conserved microbial structures and to discern between normal events and those that are associated with cellular damage. The damage signals that are detected directly by pattern recognition receptors are unknown, although the events that elicit these signals are being uncovered. NLRP3 is activated by agents that cause leaks in the plasma membrane and by phagocytosed particulates. I have demonstrated here that Nlrp1b is activated by energy deprivation. Activation of Nlrp1b by energy stress and the subsequent release of IL-1β could benefit the host in at least two ways. First, the chemokine activity of IL-1β could recruit neutrophils to inflamed tissue to combat infection. Second, IL-1β might facilitate tissue repair and immune cell function by stimulating glucose uptake (Del Rey et al., 2006). Nlrp1b was first discovered to be activated by anthrax LeTx using a genetic approach to determine why macrophages from some strains of mice are rapidly killed by the toxin while others are not (Boyden and Dietrich, 2006). The macrophages that were killed were found to express a sensitive allele of Nlrp1b (alleles 1 and 5) and to undergo a caspase-1 dependent form of cell death known as pyroptosis, whereas those that were not killed express a resistant allele (alleles 2-4). Although it was 87

97 originally believed that the induction of pyroptosis by LeTx was a virulence strategy used by B. anthracis to subvert the immune response, subsequent studies have shown that mice that express a sensitive allele are more resistant to an anthrax infection because of the beneficial release of IL-1β (Moayeri et al., 2010; Terra et al., 2010). That Nlrp1b is detecting a danger signal produced by LeTx rather than the toxin itself is attractive because the notion that a pattern recognition receptor has evolved to sense a virulence factor from a single pathogen is counter-intuitive. LeTx does not cleave Nlrp1b, although the proteolytic activity is required for inflammasome activation. Furthermore, proteasome inhibitors protect cells from LeTx-induced pyroptosis and prevent Nlrp1b-mediated caspase-1 activation, suggesting a requirement for proteasome function at an intermediate step in the signaling pathway. There is no evidence that inhibition of MAPK signaling is a DAMP detected by Nlrp1b, and notably, pharmacological inhibitors of these pathways do not induce pyroptosis. It is possible, therefore, that cleavage of an unidentified substrate by LeTx is required for Nlrp1b activation. Although LeTx did not cause a detectable reduction in total cytosolic ATP levels, the ability of MG-132 to interfere with Nlrp1b activation in response both to the toxin and to energy deprivation suggests that there is a common event in the activation pathways. One possibility is that LeTx causes a highly localized depletion of ATP 88

98 that was not detected in our assays. Alternatively, LeTx may inactivate an ATP-binding protein, which is likewise inactivated by ATP depletion, to initiate inflammasome activation. My finding that an intact Walker A motif is not required for Nlrp1b inflammasome assembly is in striking contrast to findings for NLRP3 and is consistent with a role for sensing low ATP levels. A recent study has shown that a strain of B. anthracis that produces LeTx activates caspase-1 in macrophages expressing a resistant allele of Nlrp1b (Ali et al., 2011). The authors demonstrated that both the bacterium and the toxin were required to activate caspase-1 and that the danger signal was mediated by extracellular ATP that had leaked from the macrophages. The authors did not directly test which inflammasome was responsible for caspase-1 activation by knocking down NLRP expression, but the data are consistent with NLRP3 activation. Extracellular ATP, a well-characterized trigger of NLRP3 (Mariathasan et al., 2006), binds purinergic receptors to cause the opening of pannexin-1 channels. Membrane openings cause an efflux of potassium that is required for the assembly of the NLRP3 inflammasome. That Nlrp1b detects energy stress adds to a growing body of work that links metabolism and inflammation. This connection is likely a consequence of the requirement of immune cells to function in inflamed tissues that are deprived of oxygen and glucose. Hypoxia-inducible factor 1 α (HIF-1α) is a transcription factor 89

99 that controls the cellular response to low oxygen conditions by upregulating genes involved in glycolysis, angiogenesis and glucose uptake; in myeloid cells, HIF-1α also increases the expression of TNF-α and anti-microbial factors (Peyssonnaux et al., 2005). The importance of HIF-1α for macrophage function was demonstrated by the finding that HIF-1α null macrophages exhibit impaired inflammatory function and a metabolic defect that results in an ~80% reduction in ATP levels (Cramer et al., 2003). Thus, sensors of oxygen and ATP appear to coordinate to mount an effective immune response. 90

100 CHAPTER Discussion 5.1 Autoproteolysis within the FIIND domain is important for Nlrp1b activity In this thesis, a cell-based reconstituted system was established to study the function of Nlrp1b inflammasome. This system is likely to recapitulate most cellular events in LeTx-intoxicated macrophages as pro-il-1β processing and secretion is allele specific, dependent on the enzymatic activity of LF and proteasome activity. The constitutive activity exhibited by the Nlrp1b protein without the NACHT domain suggests that the NACHT domain mediates autoinhibiton to keep Nlrp1b in an inactive form (see Chapter 3). Intriguingly, in co-immunoprecipitation assays, the Nlrp1b NACHT domain interacted with the Nlrp1b FIIND domain (Fig. 5.1A), raising the possibility that this NACHT-FIIND interaction may occur intramolecularly and contribute to autoinhibition. In order to test this hypothesis, I first performed deletion analysis to identify the minimal region of the FIIND domain that interacts with the NACHT domain. In Figure 5.1B, Nlrp1b , comprising the CARD domain and an additional 31 amino acids amino-terminal to the CARD domain, interacted with the NACHT domain whereas Nlrp1b did not. Alanine substitutions of K1113, L1114, and 91

101 I1115 in Nlrp1b abolished its interaction with the NACHT domain while alanine substitutions of H1111 and M1112 did not (Fig. 5.1C). Next, alanine substitutions were introduced at K1113, L1114, and I1115 in full length Nlrp1b (Nlrp1b-AAA). Presumably, if the intramolecular NACHT-FIIND interaction keeps Nlrp1b in an inactive form, these alanine substitutions should disrupt this NACHT-FIIND interaction in full length Nlrp1b and cause Nlrp1b activation. As shown in Figure 5.1D, Nlrp1b-AAA failed to induce pro-il-1β processing and secretion even in the presence of LeTx, suggesting that Nlrp1b-AAA was defective in activating caspase-1. Surprisingly, Nlrp1b-AAA was expressed as a singlet, suggesting that proteolysis might be important for Nlrp1b activity (Fig 5.1E). In Figure 3.1C, two bands were observed when Nlrp1b allele 1 was overexpressed whereas one single band was detected for allele 3. Deletion analysis suggests that this proteolysis may occur within the FIIND domain because Nlrp1b FIIND-CARD protein was detected as a doublet whereas Nlrp1b CARD protein was detected as a singlet (Fig. 3.4B). In fact, in a prior attempt to understand why allele 3 does not respond to LeTx activity, I made several constructs encoding hybrids of Nlrp1 allele 1 and 3. Strikingly, among all these hybrid proteins, proteolysis of Nlrp1b correlated perfectly with its activity. Proteins processed into two fragments were capable of activating pro-caspase-1 while proteins defective in proteolysis failed to 92

102 trigger pro-il-1β processing and secretion. This result, again, suggested a tempting hypothesis that the cleavage event within the FIIND domain might be critical for Nlrp1b activity. In order to test this hypothesis, a TEV cleavage site was introduced into the FIIND domain of a Nlrp1b mutant that is defective in proteolysis (Nlrp1b-TEV) (Brad Frew, unpublished data). This TEV site was recognized and cleaved by a highly specific cysteine protease, TEV. As expected, Nlrp1b-TEV was defective in proteolysis and activating caspase-1. However, when TEV protease was overexpressed with Nlrp1b-TEV, this Nlrp1b-TEV was processed into two fragments and was able to induce robust IL-1β secretion. Moreover, a point mutation within the TEV site that prevents cleavage abolished pro-il-1β processing and secretion, confirming the idea that the cleavage event within the FIIND domain is required for Nlrp1b activation. Although the mechanistic view of this proteolysis in regulating Nlrp1b activity is currently unclear, one possibility is that processed Nlrp1b has higher binding affinity toward pro-caspase-1 and facilitates its activation while unprocessed Nlrp1b is incapable of doing so. 93

103 Figure 5.1 The FIIND domain interacts with the NACHT domain. (A)(B)(C) HT1080 cells were transfected with indicated plasmids. Cells were lysed 24 h after transfection and proteins were immunoprecipitated using α-ha antibody, followed by immunoblotting with α-flag antibody. (D) Cells were transfected pcdna3-pro-caspase-1-t7 (1µg), pcdna3-pro-il-1β-ha (1µg) and pntap-nlrp1b-aaa (1 µg). Approximately 24 h after transfection, cells were treated with LeTx (10-8 M LF and 10-8 M PA). After 3 h, cell lysates were collected and probed for HA-tagged pro-il1β; HA-tagged pro-il-1β and IL-1β in supernatants were detected as described above. (E) HT1080 cells were transfected with 1 µg pcdna3-pro-caspase-1-t7, 1 µg pcdna3-pro-il-1β-ha and 1µg pntap-nlrp1b, either wild type (WT) or AAA mutant. Approximately 24 h after transfection, cells were lysed and TAP-tagged proteins were precipitated using streptavidin resin and immunoblotted using antibody directed against Nlrp1b. 94

104 5.2 H 2 O 2 activates Nlrp1b allele 1, allele 3 but not allele 4 Although Nlrp1b allele 3 did not activate caspase-1 in response to LeTx or ATP depletion, it does not necessarily suggest that allele 3 is incapable of detecting other stimuli. As it has been shown that H 2 O 2 activated the NLRP3 inflammasome, H 2 O 2 was used to test whether it could activate Nlrp1b allele 3. In Figure 5.2A, the expression level of Nlrp1b allele 1, allele 3 and allele 4 were similar. In Figure 5.2B and 5.2C, H 2 O 2 activated Nlrp1b allele 1 and 3 but not 4, causing pro-il-1β processing and secretion, despite ~10-fold more IL-1β was detected in the medium from cells expressing allele 1 than that of cells expressing allele 3. As discussed above, allele 4 is likely to be an inactive protein since it lacks the CARD domain for pro-caspase-1 binding. Of note, H 2 O 2 caused a significant reduction of intracellular ATP (data not shown). Since ATP depletion could not activate Nlrp1b allele 3, it is not likely that H 2 O 2 -induced ATP depletion activated Nlrp1b allele 3. Therefore, H 2 O 2 seems to be the trigger that induces Nlrp1b allele 3 activation independently of ATP-depletion. Although it has been demonstrated that ROS might not be involved in Nlrp1b activation as NAC did not diminish pro-il-1β processing and secretion caused by either LeTx or 2DG and NaN 3 treatment (Fig. 4.3A), it is possible that overexpression 95

105 of Nlrp1b inflammasome components circumvented the necessity of ROS to facilitate its activation. Actually, in this reconstituted system, it would be interesting to compare the H 2 O 2 signal and ATP depletion signal to see whether H 2 O 2 could induce more IL-1β secretion than 2DG and NaN 3 treatment. Figure 5.2 H 2 O 2 activates Nlrp1b allele 1, 3 but not 4. (A) Plasmids pcdna3-pro-caspase-1-t7 (1µg) and pcdna3-pro-il-1β-ha (1µg) were co-transfected with pntap-nlrp1b allele 1 (1µg), pntap-nlrp1b allele 3 (1µg), or pntap-nlrp1b allele 4 (1µg) into HT1080 cells. Approximately 24 h after transfection, cells were lysed and TAP-tagged proteins were precipitated using streptavidin resin and immunoblotted using antibody directed against Nlrp1b. (B) Plasmids encoding pro-caspase-1-t7 (1µg) and pro-il-1β-ha (1µg) were co-transfected into cells with 1µg pntap-nlrp1b allele 1, 1µg pntap-nlrp1b allele 3 or 1µg pntap-nlrp1b allele 4. Approximately 24h after transfection, cells were treated with 1mM H 2 O 2 in 5mL medium for 3h. Then cell lysates were probed for HA-tagged pro-il-1β and β-actin by immunoblotting; supernatants were 96