Granuloma Formation and Host Defense in Chronic Mycobacterium tuberculosis Infection Requires PYCARD/ASC but Not NLRP3 or Caspase-1

The NLR gene family mediates host immunity to various acute pathogenic stimuli, but its role in chronic infection is not known. This paper addressed the role of NLRP3 (NALP3), its adaptor protein PYCARD (ASC), and caspase-1 during infection with Mycobacterium tuberculosis (Mtb). Mtb infection of macrophages in culture induced IL-1β secretion, and this requires the inflammasome components PYCARD, caspase-1, and NLRP3. However, in vivo Mtb aerosol infection of Nlrp3−/−, Casp-1−/−, and WT mice showed no differences in pulmonary IL-1β production, bacterial burden, or long-term survival. In contrast, a significant role was observed for Pycard in host protection during chronic Mtb infection, as shown by an abrupt decrease in survival of Pycard−/− mice. Decreased survival of Pycard−/− animals was associated with defective granuloma formation. These data demonstrate that PYCARD exerts a novel inflammasome-independent role during chronic Mtb infection by containing the bacteria in granulomas.


Introduction
Mycobacterium tuberculosis (Mtb) is the causative agent of tuberculosis, a disease affecting one-third of the world's population and killing 1.7 million people each year [1]. Mtb is spread by aerosol droplets from persons with active infection. Upon inhalation, Mtb travels to the lung where it infects resident alveolar macrophages [2]. This initial infection leads to an innate immune response, which includes stimulation of Toll-like receptors (TLRs) that recognize pathogens and are located on the plasma membrane and within endosomes of host cells. Mtb is specifically recognized by TLRs 2, 4, and 9 [3]. TLR activation upregulates transcription of proinflammatory cytokines interleukin-1b (IL-1b), tumor necrosis factor alpha (TNFa), and interleukin-6 (IL-6), which are essential for the recruitment of immune cells to the site of infection and controlling Mtb infection [4,5,6].
In addition to TLR recognition, a newly discovered class of intracellular danger sensing proteins, the nucleotide binding domain, leucine rich repeats-containing family proteins known as NLRs, sense pathogens and pathogen products in the cell cytoplasm [7]. With more than twenty members, the NLRs function in host protection against a broad range of danger signals. Several NLRs function in immunity through the formation of a mutli-protein complex known as an inflammasome [8]. When activated by a specific danger signal, the inflammasome forms and results in recruitment and processing of pro-caspase-1, which in turn processes IL-1b and IL-18 to their active forms for secretion from macrophages.
NLRP3 is the most characterized of all the NLR inflammasome-forming proteins due to its abundant expression in macrophages and activation in response to the largest number of identified stimuli. In humans, a gain of function mutation in NLRP3 is associated with hyperinflammatory hereditary periodic fever syndromes with symptoms ranging from mild rash to severe joint swelling [9].
Apoptotic speck-like protein containing a CARD domain (ASC, also PYCARD) was originally identified as a speck-forming protein during apoptosis of HL-60 cells [10]. PYCARD has also been recognized as an adaptor protein that interacts with NLR proteins forming a protein inflammasome structure resulting in caspase-1 processing and subsequent IL-1b and IL-18 activation [11]. PYCARD has been identified as an adaptor protein for NLRP3 and NLRP1, and it is functionally required for the NAIP5 and NLRC4 inflammasomes [12]. Each inflammasome responds to a specific set of stimuli, although there is some redundancy between NLRs. Together, PYCARD, NLRP3, and caspase-1 are essential for macrophage IL-1b maturation in response to a broad range of stimuli including bacteria [13,14] and viruses [15,16,17,18]. In addition to these inflammatory functions, PYCARD association with NLR proteins is required for pyroptotic and pyronecrotic cell death [19,20]. Although the host's innate immune response to Mtb infection is critical for the initial defense against bacteria, the adaptive immune response is ultimately required for containment of the infection in the chronic stage of disease. Adaptive immunity to Mtb infection is characterized by the appearance of antigen specific CD4 + T-cells that secrete interferon-gamma (IFN-c), which is responsible for activating macrophages to kill intracellular bacteria [21]. CD8+ T-cells are also important for controlling bacteria during the chronic phase of Mtb infection [22]. Chronic Mtb infection is controlled by granuloma formation which contains, but does not eliminate bacteria [23]. Granulomas consist of a central core of Mtb-infected macrophages surrounded by successive waves of activated macrophages, giant multinucleated cells, epithelioid cells, lymphocytes, fibroblasts, and dendritic cells. A subset of granulomas undergoes central caseous necrosis due to proteinaceous dead cell mass. Mice form slightly different granulomas that do not form caseous necrotic centers, but otherwise possess the same cell types and similar granuloma organization to humans [24].
Pro-inflammatory cytokine regulation can be critical to longterm survival of Mtb infection. In vivo assessments of IL-1a/b 2/2 , IL-1R 2/2 , and IL-18 2/2 mice have shown that these cytokines play a role in limiting bacterial lung burden, regulating other cytokines, nitric oxide production, and forming organized granulomas [25,26,27]. Likewise, mice deficient in pro-inflammatory cytokines IL-6 and TNFa have increased mortality during Mtb infection [4,5]. TNFa is important for granuloma formation and maintenance [28]. Therefore, these cytokines are not only important in the innate immune response to Mtb, but also in host defense during chronic Mtb infection.
High interest in the role of NLRs in host immunity has led to the study of inflammasome complexes in response to many pathogens. The research thus far has almost exclusively focused on the acute effects of pathogens and other NLR stimuli. Consequently, the in vivo role of the NLR inflammasome during chronic infection has not been studied. Mtb infection exemplifies a chronic infection of paramount public health interest. Mtb infects macrophages where it must thwart the host immune response to survive and replicate. Nlrp3 inflammasome proteins are expressed in macrophages; thus, we hypothesized that Mtb infection would induce inflammasome activation. Here we show that Mtb induced IL-1b secretion in human and mouse macrophages in vitro and this process was dependent on PYCARD, caspase-1, and NLRP3, but not NLCR4. In vivo, murine Pycard helps protect the host from death during chronic Mtb infection while the effects of Casp-1 and Nlrp3 were negligible. The inability of Pycard 2/2 mice to form organized granulomas and the reduced presence of lung dendritic cells indicates a breakdown in host defense against Mtb. Thus, we identify PYCARD as a critical protein involved in host response to Mtb infection in an inflammasome-independent manner.

Results
Virulent and attenuated Mtb require PYCARD, NLRP3, and caspase-1 for IL-1b secretion by cultured human cells During human Mtb infection, bacteria travel to the alveolar spaces where they infect and replicate inside macrophages. Secretion of IL-1b by macrophages requires pro-IL-1b processing by caspase-1. Caspase-1 interaction with several NLR forming inflammasomes results in caspase-1 processing, a prerequisite for IL-1b activation [12]. To determine if host detection of Mtb involves inflammasome activation, we used a used a panel of human monocytic THP-1 cell lines with reduced expression of inflammasome genes due to shRNA targeting sequences (Table S1, Figure S1). During infection with the virulent Mtb strain H37Rv, THP-1 cells with shPYCARD or shNLRP3 secreted significantly less IL-1b than their scrambled controls ( Figure 1A). This indicates that Mtb activates the NLRP3 inflammasome. IL-18 is an inflammatory cytokine which induces cell-mediated immunity and causes T-cells to secrete IFN-c [29]. Similar to IL-1b, it also requires caspase-1 cleavage for activation. IL-18 secretion from Mtb-infected THP-1 cells was dependent on PYCARD and NLRP3 ( Figure 1B). These data demonstrated that both IL-1b and IL-18 were processed during Mtb infection and that each required the NLRP3 inflammasome formation for activation in the THP-1 human monocytic cell line.
As seen with H37Rv, the attenuated Mtb H37Ra strain induced IL-1b secretion from THP-1 cells in a PYCARD-and NLRP3dependent manner. Induction of IL-1b by H37Ra was dependent on the NLRP3 inflammasome ( Figure 1C). We consistently observed more IL-1b secretion in THP-1 cells infected with H37Ra, although the difference was not statistically significant. Thus, the NLRP3 inflammasome is activated by both virulent and attenuated Mtb. Chemically inhibiting caspase-1 with caspase-1specific inhibitor Y-VAD significantly reduced IL-1b secreted by THP-1 cells, confirming the necessity of caspase-1 cleavage for IL-1b secretion ( Figure 1D).
Production of IL-1b in Mtb infected primary mouse macrophages is Pycard, Nlrp3, and Casp-1 dependent To confirm our THP-1 data that the inflammasome is necessary for IL-1b processing, we infected primary mouse macrophages from gene depletion mice. We infected either bone-marrow derived (BMDM) or thioglycolated elicited macrophages with Mtb. Mtbinduced IL-1b secretion in wild type macrophages, confirming the results we obtained with THP-1 cells. Mtb-infected BMDM from Pycard 2/2 , Nlrp3 2/2 , and Casp-1 2/2 mice had significantly reduced IL-1b secretion, indicating that the Nlrp3-inflammasome is necessary for Mtb-induced IL-1b secretion in macrophages. Nlrc4 2/2 BMDM secreted IL-1b at the same or higher levels compared to wild type, indicating this inflammasome is not involved in the host response to Mtb ( Figure 2A). This demonstrates congruent findings using THP-1 cells and primary mouse macrophages.
Induction of the NLRP3 inflammasome was not limited to virulent Mtb H37Rv infection, but also occurred with attenuated Mtb H37Ra ( Figure 2B). Wild type BMDM infected with Mtb H37Ra secreted abundant amounts of IL-1b while Pycard 2/2 and Nlrp3 2/2 macrophages did not. Our findings were repeated in primary thioglycolate elicited macrophages. These highly activated macrophages had a significant reduction in IL-1b secretion in the absence of Pycard and Nlrp3 ( Figure 2C), reinforcing our data from human THP-1 cells and BMDM infected with virulent and attenuated Mtb. IL-18 secretion from both BMDM and thioglycolate elicited macrophages was below the level of detection for all samples tested. In cultured human monocytes and primary mouse macrophages Mtb induced IL-1b and IL-18 secretion in the absence of cell priming. This indicates that Mtb upregulates both transcription and processing of these proinflammatory cytokines in vitro. Taken together, our data show that the ability of Mtb to induce IL-1b secretion was PYCARD, NLRP3, and caspase-1 dependent in human THP-1 cells as well as primary mouse macrophages. In contrast, NLRC4 did not affect IL-1b production, indicating that Mtb induces a host response through the NLRP3, but not the NLRC4, inflammasome.

Pycard but not Nlrp3 protects the host against virulent Mtb infection
Lack of IL-1b in vivo has been previously shown to be important for response to Mtb infection resulting in increased lung bacterial burden, differential regulation of cytokines, and defects in granuloma formation [6,26]. Given our data from infecting cultured cells, we hypothesized that mice lacking the Pycard or Nlrp3 would be unable to process IL-1b resulting in an inability to control Mtb in vivo. To test this hypothesis, wild type, Pycard 2/2 , and Nlrp3 2/2 animals were infected with aerosolized Mtb H37Rv.  Each mouse received between 250-350 cfu per lung. During the first four-and-a-half months, survival of wild type and Pycard 2/2 mice was similar. However, Pycard 2/2 mice had a survival defect thereafter, and they died precipitously between 143 and 155 days post-infection with a mean survival of 148 days ( Figure 3A). By comparison, wild type mice infected with Mtb had a mean survival of 200 days. These results were confirmed in a two additional experiments (not shown). Death of Pycard 2/2 mice during Mtb infection was not due to advanced age as we routinely keep Pycard 2/2 breeder pairs for up to a year with no unexplained deaths. This strong survival phenotype shows that Pycard is important for host defense against virulent Mtb infection in vivo.
PYCARD is an adaptor protein that is required for several NLR proteins to form inflammasomes. Since our in vitro data indicated NLRP3 was necessary for Mtb induction of IL-1b in macrophages, we also investigated survival of Nlrp3 2/2 mice during aerosolized Mtb infection. Although NLRP3 is important for IL-1b secretion from cultured macrophages, there was no difference in survival of Nlrp3 2/2 mice in vivo ( Figure 3B). Overall, Nlrp3 2/2 mice had a mean survival of 209 days compared to 212 days for wild type mice. This data was also confirmed in a second experiment (not shown). In a separate experiment, we tested the ability of another closely related NLR gene, Nlrc4, to protect the host from Mtb. Nlrc4 2/2 mice had a similar survival profile to paired wild type controls, with a mean survival of 190 days compared to 198 days for wild type controls ( Figure 3C). These data indicate that elimination of Pycard, but not Nlrp3 nor Nlrc4, resulted in a reduced ability of the host to defend against Mtb infection.

Pycard and Nlrp3 do not significantly affect Mtb bacterial burden
Prior in vivo studies show that Mtb has two phases of infection [30]. In the first three weeks following aerosol infection, Mtb grows logarithmically in the lungs. After three weeks of infection, an effective Th1 adaptive immune response is established and the exponential growth of Mtb ends. Mtb bacterial burden in the lungs, spleen, and liver persist at the same level for the remainder of the infection [31]. We assessed bacterial burden in the lungs of Pycard 2/2 and Nlrp3 2/2 mice in two separate infections, with each compared to paired C57BL/6 controls. Animals lacking Pycard and Nlrp3 have similar levels of bacteria in the lungs during the growth and persistence phases of Mtb infection compared to wild type ( Figure 3D, G). The last time points were taken as Pycard 2/2 and Nlrp3 2/2 mice neared death. We saw no statistical difference between the bacterial burden of lungs from Pycard 2/2 , Nlrp3 2/2 , or wild type mice. The bacterial burden in the liver and spleen of the Pycard 2/2 and Nlrp3 2/2 mice was also similar to that found in wild type mice, indicating that Pycard and Nlrp3 do not affect bacterial dissemination ( Figure 3E-F, H-I). Thus, a change in Pro-inflammatory cytokines are produced at similar levels in the lungs of wild type, Pycard 2/2 , and Nlrp3 2/2 mice following Mtb H37Rv aerosol infection Our earlier studies led us to hypothesize that less IL-1b would be produced in the lungs of Pycard 2/2 and Nlrp3 2/2 mice compared to wild type because the gene deletion mice would not be able to cleave pro-IL-1b to its active form in response to Mtb infection. To test this, cytokines were measured from tissue free, homogenized lung extract by ELISA. Surprisingly, we found all three groups of animals produced IL-1b in similar amounts. IL-1b levels increased during the logarithmic growth phase of infection and were still high during the persistence phase of infection ( Figure 4A). By week 16, levels of IL-1b decreased slightly. There was no significant reduction in IL-1b secretion between Pycard 2/2 and Nrp3 2/2 mice compared to wild type at any of the time points we measured. Because of the mechanical nature of lung homogenization, cells are broken open in the process which could result in release of both pro and cleaved IL-1b in the lung homogenate extracts. We performed IL-1b western blots in addition to ELISAs to ensure we were measuring the amount of cleaved IL-1b produced in the lungs of Mtb-infected animals in vivo. At one week post-infection, as expected, only pro-IL1b was present ( Figure 4B, top panel). At 16 weeks post-infection, Pycard 2/2 and Nlrp3 2/2 animals produced comparable if not enhanced levels of cleaved IL-1b compared to wild type mice ( Figure 4B, bottom panel). The amount of IL-1b was reproducible between mice and correlates with the ELISA data. These data show that differences in IL-1b production cannot explain differences in the survival rate of Pycard 2/2 compared to WT or Nlrp3 2/2 animals. We also measured pro-inflammatory cytokines IL-6 and TNFa. TNFa is of particular interest as it is important for granuloma formation and the control of infection in Mtb-infected animals and humans [5,28]. IL-6 has also been shown to be an important host response against Mtb [4]. Furthermore, a previous report showed that human PYCARD is required for TNFa and IL-6 expression upon bacterial infection [32]. Both cytokines followed similar trends as IL-1b with total cytokine levels rising during early infection and then dropping off significantly by week 16 (Figure 4C and D). Nlrp3 2/2 mice had a modest but statistically significantly increase in IL-6 and TNF-a cytokines at 2 weeks (p = 0.026 and p = 0.028, respectively) and 5 weeks post-infection (p = 0.004, p = 0.001, respectively), but all differences were diminished by 16 weeks post-infection. There were no significant differences in lung cytokine levels between Pycard 2/2 and wild type mice at any time points analyzed. Proinflammatory cytokines in serum and bronchoalveolar lavage fluid were below the level of detection. These data indicate that reduced TNFa and IL-6 are not correlated with increased death in Mtb-infected Pycard 2/2 mice.
Caspase-1 is not required for survival, bacterial containment, or the production of proinflammatory cytokines in Mtb infected mice Caspase-1 recruits and cleaves IL-1b to its active form and is common to all NLR inflammasomes. A previous report has shown that at a low infectious unit of influenza virus, viralinduced host response is mediated by an Nlrp3-independent but Pycard/Casp-1-dependent inflammasome [15]. To assess if Caspase-1 and Pycard together mediate host protection in the context of inflammasome activation, we investigated the contribution of Caspase-1 during Mtb infection. Infection of Casp-1 2/2 mice with Mtb produced no difference in survival compared to wild type mice ( Figure 5A). Measurement of bacterial burden showed that Casp-1 2/2 and wild type mice had similar levels of bacteria in the lungs over the course of Mtb infection ( Figure 5B). Casp-1 2/2 and wild type mice also had similar bacterial burden in the liver and spleen, indicating that bacterial dissemination and growth in these organs are comparable between genotypes ( Figure 5C). Overall, these data indicate that Caspase-1 does not play a prominent role in host protection during Mtb infection. Hence, this work describes a novel Pycard-dependent but Casp-1independent form of host immunity.
Caspase-1 cleaves pro-IL-1b to its biologically active form. However, our earlier results demonstrated that cleaved IL-1b levels were similar in Pycard 2/2 , Nlrp3 2/2 , and Nlrc4 2/2 lungs. An analysis of Casp-1 2/2 mice show that rather than diminished IL-1b, they had increased lung IL-1b in their lungs compared to wild type as measured by ELISA ( Figure 5D). We utilized western blots to measure both the pro-and cleaved forms of IL-1b from lung homogenate extracts collected one week and 20 weeks post infection. At one week post infection only low levels of pro-IL-1b were present in the Casp-1 2/2 and wild type lungs ( Figure 5E, top panel). At 20 weeks post infection pro-and cleaved IL-1b were present in both Casp-1 2/2 and wild type mouse lungs with more found in the former ( Figure 5E, bottom panel). These data demonstrate that Casp-1 2/2 mice have compensatory mechanisms of cleaving IL-1b in the absence of caspase-1. Levels of IL-6 and TNF-a also were unchanged between Casp-1 2/2 and wild type lungs ( Figure 5F-G). Overall our in vivo characterization of the Casp-1 2/2 mouse reveals that Caspase-1 does not play a prominent role in host protection against Mtb. This supports the conclusion that protective host immunity to Mtb is independent of the inflammasome.

Pycard 2/2 mice form fewer granulomas
To further explore the mechanism by which Pycard enhances survival in chronically infected mice, we examined the histopathology of Mtb-infected mouse lungs. We examined granuloma structure because of its role in containing bacteria during chronic Mtb infection [23] and because Pycard 2/2 mice exhibit an abrupt drop in survival during the chronic phase of infection. We assessed lung inflammation 16 weeks post infection by measuring the amount of inflamed tissue from each mouse lung (see experimental procedures). Low magnification images showed that the alveolar spaces of wild type, Pycard 2/2 , Nlrp3 2/2 , and Casp-1 2/2 mice were obstructed by an influx of immune cells into the lung. We found the lungs of all our gene depletion and wild type mice to be highly inflamed, ranging from 90-98 percent of the total lung tissue ( Figure 6A-B). In all infected mice, we observed dense areas of immune cells, indicative of granulomatous lesions. Strikingly, the Pycard 2/2 animals formed significantly fewer granulomas per lung compared to wild type, Nlrp3 2/2 , and Casp-1 2/2 lungs ( Figure 6C-D). Examination under higher magnification showed that wild type and Nlrp3 2/2 mice formed similar sized granulomas. Hence, Pycard affected the formation of granulomas, but once they were formed, granuloma size was not affected by this gene (Figure 6E-F). Casp-1 2/2 mice formed larger granulomas, although this did not affect the outcome of infection. To further assess the lung granuloma defect in Pycard 2/2 mice, we stained mouse lung sections with Ziehl Neelsen stain to identify Mtb. Earlier data showed that the overall lung bacterial burden was similar in wild type and Pycard 2/2 mice by bacterial plating of homogenized lungs ( Figure 3F-G). In contrast, analysis of acid fast staining revealed striking differences in bacterial localization in wild type and Pycard 2/2 lungs. Acid fast staining of wild type lungs shows very little Mtb located in non-granulomatous lung tissue. In contrast, Pycard 2/2 mice contain bacteria 406 images demonstrate that Pycard 2/2 mice form significantly fewer granulomas compared to wild type, Nlrp3 2/2 , and Casp-1 2/2 mice (* p,0.05). E-F. 4006 images demonstrate wild type, Pycard 2/2 , and Nlrp3 2/2 mice form similarly sized granulomas while Casp-1 2/2 mice form significantly larger granulomas (**p,0.001). G-H. Images of acid-fast stained lung sections from nongranulomatous regions of Mtb-infected lung tissue were taken at 1006and 4006. Pycard 2/2 mice have bacterial located throughout the lung unlike wild type mice in which bacteria is mainly contained within granulomas (*p,0.01). To view bacteria within a single macrophage, 10006images were cropped, and enlarged. Quantification was performed on sections from 4-8 mouse lungs per genotype from two independent experiments. doi:10.1371/journal.pone.0012320.g006 throughout the lung and have extremely high amounts of bacteria in tissue that is not associated with granulomas ( Figure 6G). The amount of bacteria located outside of the granuloma was quantified by using a scale of 0-4 based on the amount of bacteria present in lung tissue that was not associated with granulomas (see experimental procedures) ( Figure 6H). Pycard 2/2 mice form significantly fewer lung granulomas than wild type mice. The acid-fast staining data demonstrate that Pycard 2/2 animals were unable to contain bacteria within granulomas, shown by massive amounts of Mtb located in non-granulomatous lung tissue compared to wild type mice. Reduced granulomas in the lungs of Pycard 2/2 provide plausible mechanisms to explain the decreased lifespan observed in these animals during chronic Mtb infection.

Discussion
In this report, we focus on the ability of virulent Mtb to stimulate inflammasome activation and the role of the inflammasome in host defense against Mtb. Prior studies have implicated the inflammasome in mycobacterium infection in vitro. Consistent with our data, one report showed that M. marinum, a mycobacterium that naturally infects fish and amphibians, activates IL-1b production in a PYCARD, NLRP3, and caspase-1-dependent manner in vitro, but no in vivo investigation was conducted [33]. Another report showed that M. bovis BCG, the vaccine strain of Mtb, can limit caspase-1 and IL-1b activation due to a putative protease Zmp1. In this latter study, however, there was no analysis of Pycard, Nlrp3 or Nlrc4 to directly implicate the inflammasome components [34]. A third report showed that Mtb infected THP-1 human monocyte cell line secrete IL-1b in a PYCARD and NLRP3-dependent process, but did not investigate the role of the inflammasome in vivo [35].
Our study showed that virulent Mtb H37Rv, along with the attenuated derivative Mtb H37Ra, induced NLRP3, PYCARD, and caspase-1-dependent inflammasome activation in a human macrophage cell line and in primary mouse macrophages. However, these in vitro results do not predict the outcome in mice. Most importantly, our study showed that Pycard is important for host protection during chronic infection with virulent Mtb infection in vivo, demonstrated by decreased survival of Pycard 2/2 mice compared to wild type controls. Pycard did not confer a survival advantage through the production of IL-1b. Rather it functions to prolong host survival through a novel inflammasome-independent role likely through proper granuloma formation during chronic Mtb infection. Surprisingly, neither Nlrp3 nor Casp-1 played a prominent role in host protection during Mtb infection despite its importance in culture. The inconsistencies between the in vitro and in vivo role of Nlrp3 and Casp-1 underscores the absolute necessity for in vivo validation in relating host immune genes to the outcome of microbial infections.
IL-1b cleavage is a hallmark of inflammasome activation, and we observed significant Pycard, Nlrp3, and Casp-1-dependent inflammasome activation during in vitro infection of cultured cells. However, in our in vivo studies, we observed mature IL-1b in the lungs of Mtb infected wild type, Pycard 2/2 , Nlrp3 2/2 , and Casp-1 2/2 mice. This, along with mouse survival data, indicates that Pycard host protection during Mtb infection in vivo is inflammasome independent. The similar survival profiles of Nlrp3 2/2 and Casp-1 2/2 mice compared to wild type mice may be due to compensatory mechanisms of IL-1b processing. It is important to note that the inflammasome is not the only host mechanism for cleaving pro-IL-1b to its active form. Through less well established, caspase-1-independent IL-1b cleavage can be carried out by several other host proteases including granzyme A, chymase, chymotrypsin, and matrix metalloproteinases as well as bacterial enzymes [36,37,38,39]. Meanwhile, Pycard 2/2 mice lacked an additional mechanism to form or maintain granulomas thereby compromising their resistance to Mtb infection.
To investigate the mechanism further, we show that Pycard 2/2 mice have fewer lung granulomas compared to wild type, Nlrp3 2/2 , or Casp-1 2/2 mice, despite having the same amount of inflamed lung tissue. Acid-fast staining for Mtb localization within the lung demonstrated that Pycard 2/2 mice are defective in containing Mtb within granulomas, as shown by large amounts of bacteria found in non-granulomatous lung tissue. In comparison, infected wild type mice have almost no bacteria found outside of granulomas. Our data show that overall the lungs of Pycard 2/2 and wild type mice have the same amount of bacteria. However, the localization of Mtb within the lung is radically different. These data support a working hypothesis that Pycard affects proper cell accumulation during chronic Mtb infection and promotes proper granuloma formation.
Recently, another group implicated an inflammasome-independent role for Pycard in host protection against Mtb in vivo [40]. In contrast to our data, this paper found that Casp-1 deficient mice had decreased survival during Mtb infection. This may be due to differences in the amount of bacteria delivered to the mice or mouse housing conditions. Nonetheless, both their and our papers come to the same conclusion: that Pycard has a role in host protection independent of Nlrp3.
In summary, this work demonstrates that PYCARD promotes survival during the late phase of Mtb infection independent of inflammasome activation. This result also highlights the importance of PYCARD in the process of granuloma formation during chronic Mtb infection. Considering the high rates of chronic Mtb infection worldwide and the increasing number of multi-drug resistant and extensively drug resistant tuberculosis cases, our study suggests that understanding the role of PYCARD during Mtb infection may lead to new and more effective therapies to treat chronic Mtb infection. In contrast, the lack of a role for NLRP3 and caspase-1 in the in vivo containment of Mtb is encouraging for the development of inhibitors for autoinflammatory disorders. These results suggest that anti-NLRP3 and caspase-1 strategies are not likely to cause the inadvertent activation of latent Mtb infection in patients.

Ethics statement
All studies were conducted in accordance with the National Institutes of Heath Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee (IACUC) guidelines of the University of North Carolina at Chapel Hill (protocol #07-170 and 09-195).
shRNA knockdown THP-1 cell lines THP-1 cells were obtained from ATCC. Generation of shRNA knockdown and control vectors for PYCARD and NLRP3 have been described previously [13,32]. All shRNA hairpins were confirmed by sequencing (Table S1). Lentiviral packaging and transduction of THP1 cells has been previously described [41]. Verification by RT-PCR indicated a significant reduction in PYCARD and NLRP3 expression ( Figure S1 A-B). Functionally, shPYCARD and shNLRP3 had reduced IL-1b production in response to LPS ( Figure S1C).

Bacterial strains
Mtb H37Rv and H37Ra were obtained from the laboratory of William R. Jacobs, Jr. [42] and ATCC, respectively. Bacteria were grown to log phase in Middlebrook 7H9 broth (Difco) with 0.2% glycerol, 16 albumin dextrose saline, and 0.05% Tween 80. Inoculum was assessed by plating infection media on Middlebrook 7H10 agar plates supplemented with glycerol and ADS as described above. Colony forming units (cfu) were counted after 21 days of incubation.

Mtb infection of cultured cells
Cell lines were cultured in RPMI (Gibco) with 10% FBS (Hyclone). Bone marrow derived macrophages were harvested from 6-to 8-week-old mouse femurs and cultured for 6 days in DMEM supplemented with L-glutamine, non-essential amino acids, 10% fetal bovine serum, and 20% L929 conditioned media. Thioglycolate elicited macrophages were obtained by peritoneal lavage 4 days after intraperitoneal injection with 3% thioglycolate and cultured in DMEM supplemented as above. All cells were infected under BSL3 conditions with Mtb H37Rv or H37Ra at an MOI of 10 and incubated for 8 hours at 37uC with 5% CO 2 . Cellfree supernatants were harvested, double filtered with 0.2 m filters, and assayed for cytokines by ELISA and western blot.

Cytokine determination
Cytokines were measured from infected cell supernatants with human or mouse BD OptEIA IL-1b, TNFa, and IL-6 ELISA Sets (BD Biosciences) and IL-18 ELISA (MBL International). In vivo cytokines were measured from lung homogenate extracts by centrifuging homogenized lung tissue to create a tissue-free supernatant. Amounts of pro-and cleaved IL-1b in lung homogenate extracts were determined by western blot. Immunoblots were probed with goat anti-mouse IL-1b primary antibody (R&D Biosystems). Bands were visualized by Super Signal Chemiluminescence (Pierce).

Histopathology
Lungs were fixed in 10% buffered formalin and stained with H&E to evaluate airway inflammation and identify granulomas. The percent of inflamed lung tissue was determined by dividing the inflamed areas by the total lung area. Granuloma frequency was determined by counting the number of granulomas present in the total lung section. Granuloma size was measured by defining the granuloma borders for 26 lesions using Image J software and determining the average. Ziehl-Neelsen (ZN) staining was used to determine bacterial localization. The entire Mtb infected mouse lung was scored on a scale from 0-4: 0 = no bacteria found outside of granulomas, 1 = 1-5 bacteria found outside of granulomas, 2 = 6-20 bacteria found outside of granulomas, 3 = approximately half of each field contains 20-50 bacteria outside of granulomas, and 4 = greater than half of each field contains greater than 50 bacteria outside of granulomas. H&E lung sections were blindly scored by E.M., I.C.A., M.S., and P.H. ZN stained lung sections were blindly scored by M.S. and P.H. For all histology quantification 4-8 mouse lungs per genotype were analyzed.

Statistics
Data are presented as the means +/2 standard deviation (SD) unless otherwise noted. Analysis Of Variance (ANOVA) followed by Tukey-Kramer HSD for multiple comparisons was performed on complex data sets. Statistical significance for single data points was assessed by the Student's two-tailed t-test. Survival curves were generated utilizing the product limit method of Kaplan and Meier and comparisons were made using the log rank test. In all cases, a p-value of less than 0.05 was considered statistically significant.

Supporting Information
Table S1 Target sequences for shRNA THP-1 cell lines. Target sequences for PYCARD and NLRP3 shRNA hairpins used to construct stable knockdown THP-1 cell lines.