The PYRIN domain-only protein POP2 inhibits inflammasome priming and activation

Inflammasomes are protein platforms linking recognition of microbe, pathogen-associated and damage-associated molecular patterns by cytosolic sensory proteins to caspase-1 activation. Caspase-1 promotes pyroptotic cell death and the maturation and secretion of interleukin (IL)-1β and IL-18, which trigger inflammatory responses to clear infections and initiate wound-healing; however, excessive responses cause inflammatory disease. Inflammasome assembly requires the PYRIN domain (PYD)-containing adaptor ASC, and depends on PYD–PYD interactions. Here we show that the PYD-only protein POP2 inhibits inflammasome assembly by binding to ASC and interfering with the recruitment of ASC to upstream sensors, which prevents caspase-1 activation and cytokine release. POP2 also impairs macrophage priming by inhibiting the activation of non-canonical IκB kinase ɛ and IκBα, and consequently protects from excessive inflammation and acute shock in vivo. Our findings advance our understanding of the complex regulatory mechanisms that maintain a balanced inflammatory response and highlight important differences between individual POP members.

T o mount an inflammatory response, which is crucial for host defence and tissue repair, against infections and tissue damage, cytosolic pathogen/microbe-associated molecular patterns (PAMPs/MAMPs) and damage-associated molecular patterns (DAMPs or danger signals) are sensed by pattern recognition receptors (PRRs). PRRs belonging to the AIM2-like receptor and Nod-like receptor (NLR) families trigger the assembly of large protein platforms, called inflammasomes 1,2 . The NLRP3 inflammasome has an important function in response to infectious and sterile activators, and its activation depends on a two-step mechanism, referred to as priming and activation 3,4 . Priming of the NLRP3 inflammasome depends on Toll-like receptor (TLR) signalling that induces NF-kBdependent expression and post-translational modification of NLRP3 and interleukin (IL)-1b. Then, NLRP3 is activated by a variety of PAMPs/MAMPs and DAMPs, including the goutcausing uric acid crystals, which ultimately trigger K þ efflux as a shared signal 3,5,6 . Recruitment of the adaptor protein ASC to NLRP3 nucleates ASC polymerization and subsequently leads to caspase-1 activation through an induced-proximity mechanism 7 . Active caspase-1 then executes the proteolytic cleavage, maturation and release of the pro-inflammatory cytokines IL-1b and IL-18 and induces pyroptotic cell death 1,8,9 . Although caspase-1 is activated by the canonical inflammasome pathway, caspases-4 (mouse caspase-11) and -5 are activated by the non-canonical inflammasome pathway. In contrast to caspase-1 activation in the canonical inflammasome complex, the non-canonical pathway does not rely on receptor-and ASC-mediated sensing of molecular patterns; instead, non-canonical inflammasomelinked caspases bind directly to lipopolysaccharide (LPS) through their caspase recruitment domain, thus causing their activation 10 . Although caspase-1 induces IL-1b and IL-18 maturation as well as pyroptotic cell death, caspases-4 and -5 only induce pyroptosis, which causes the release of IL-1a and HMGB1 (refs 3,4). Pyroptosis also releases extracellular oligomeric ASC particles, which can be opsonized and phagocytized by neighbouring cells to propagate and perpetuate inflammasome responses 11,12 . Upon blocking the NLRP3 inflammasome with POP1, the ASC particle release is prevented 13 . The mechanism for canonical inflammasome complex assembly depends on specific protein-protein interactions, which are mediated by homotypic PYRIN domain (PYD) interactions between PRRs and ASC, and by homotypic CARD interactions between ASC and pro-caspase-1 (ref. 14). The PRR-initiated nucleation of the prion-like, self-perpetuating ASC polymerization into filaments or spheres is a crucial step 7,11,[15][16][17] . Hence, the PYD is essential for inflammasome assembly and caspase-1 activation, but we are just beginning to understand the molecular processes that regulate inflammasome activation and resolution. Since excessive inflammasome activation causes inflammatory diseases, a balanced inflammasome response is crucial for the host 18,19 .
We discovered a family of cellular PYD-only proteins (POPs), including POP1 (also known as PYD containing 1 (PYDC1)), POP2 (also known as PYDC2) and POP3 (refs 20-22). These small proteins are composed of only a PYD, and are encoded in the human genome, but are missing from the mouse genome 21,22 . Interestingly, poxviruses encode viral POPs that function as immune-evasion proteins to dampen host immune responses 23,24 . We previously demonstrated that the monocyte/macrophagespecific expression of POP1 prevents ASC-dependent inflammasome responses by binding to ASC and thereby preventing ASC-PRR interactions, which ultimately ameliorates systemic inflammation and auto-inflammatory disease 13 . POP3, which does not bind to ASC, but specifically interacts with the cytosolic DNA sensors AIM2 and IFI16, also blocks inflammasome activation and promotes type I interferon production in vivo 20 . The third member, POP2, is unique, as it not only binds to ASC to inhibit inflammasome responses, but also inhibits NF-kB activation in vitro 25,26 . However, its inflammasome and NF-kB-inhibiting roles have not been investigated in detail, and the function of POP2 in vivo is still elusive.
Here we report that human POP2 simultaneously blocks NLRP3 inflammasome priming and activation in a macrophagespecific transgenic mouse model, and thereby ameliorates pathogen-and damage signal-induced systemic inflammation in vivo. Our study provides insights into the complex molecular mechanisms that regulate inflammasome activation in humans to limit systemic inflammatory disease.

Results
POP2 protects mice from excessive inflammation. Reminiscent of POP1 and POP3, POP2 is also expressed in CD68 þ macrophages in inflamed human lung tissue, as shown by immunohistochemistry using a custom-raised POP2 antibody, which we validated for specificity among POPs 20 (Fig. 1a). In addition, we detected strong POP2 staining in the synovial lining of human rheumatoid arthritis (RA) patient samples, which is rich in CD68 þ macrophages and correlates with disease severity 27,28 (Fig. 1a). Interestingly, similar to POP1 and POP3, POP2 is also lacking in mice, and evolved recently in the mammalian genome of humans and closely related primate species 22,29 . To investigate the physiological function of POP2, we generated transgenic mice expressing POP2 (POP2 TG ) from the human CD68 promoter containing the macrophage-specific IVS-1 enhancer, an approach that we validated earlier for POP1 and POP3 to study macrophage inflammasomes 13,20 . Utilizing the CD68/IVS-1 promoter enables high-level transgene expression in monocytes, tissue-resident macrophages and dendritic cells, but not other cell types 13,20,30 . We observed a comparable restricted expression pattern in our POP2 TG mice and detected expression of POP2 in blood CD11b þ cells utilizing POP2-specific Nanoflares. Specifically, POP2 was expressed in Ly6C þ classical monocytes and, to a much lesser extent, in Ly6C med and Ly6C À monocytes ( Fig. 1b and Supplementary  Fig. 1). POP2 expression was also confirmed in vitro with macrophage colony-stimulating factor (M-CSF)-differentiated peripheral blood-derived macrophages by immunoblot (Fig. 1c). We showed previously that POP2 was capable of inhibiting NLRP3-mediated inflammasome responses using an in vitro reconstitution system 26 . We therefore investigated the in vivo NLRP3 response to monosodium urate (MSU) crystals, which act as a universal DAMP and are responsible for synovial inflammation in gout patients 6,31 . Injection of MSU crystals into the peritoneal cavity of wild type (WT) mice causes neutrophil infiltration and inflammation, which was determined by in vivo image analysis and subsequent quantification of a luminescent myeloperoxidase (MPO) probe, and this response was markedly attenuated in POP2 TG mice (Fig. 1d). MSU crystalinduced neutrophil recruitment depends on the NLRP3inflammasome-mediated production of IL-1b (ref. 32). Accordingly, IL-1b levels in peritoneal lavage fluids of POP2 TG mice were significantly reduced (Fig. 1e). However, unlike in POP1 TG mice, the inflammasome-independent release of IL-6 and tumour necrosis factor (TNF) was also reduced (Fig. 1e) 13 , which is likely caused by its NF-kB-inhibiting activity that has been described in vitro 25,33 . The lipid mediator leukotriene B 4 (LTB 4 ) is produced following MSU crystal-induced tissue damage in response to lysosomal damage and promotes reactive oxygen species production that is required for subsequent NLRP3 activation 34 . In agreement with its function upstream of NLRP3 activation, LTB 4 production in response to MSU crystals was not altered in POP2 TG mice. Hence, POP2 impairs neutrophil infiltration by a mechanism that involves diminished NLRP3inflammasome-mediated IL-1b production.
Macrophages release polymerized ASC particles through pyroptosis, an inflammasome-dependent mechanism. ASC particles act as damage signals to further propagate and perpetuate inflammatory responses and are present in the serum  of active Cryopyrin-associated periodic-syndrome patients 11,12 . We recently demonstrated that POP1 expression attenuates the release of ASC particles 13 . Similarly, POP2 TG mice had reduced extracellular ASC particles compared to WT mice in the cleared peritoneal lavage fluid, as determined by immunoblot analysis (Fig. 1f). In another mouse model that better recapitulates a local inflammatory MSU response in a synovium-like subcutaneous 'airpouch' 35 , POP2 TG mice also showed strongly reduced MPO activity (Fig. 1g). Accordingly, we also observed a reduced number of total infiltrating leukocytes and, in particular, fewer neutrophils and Ly6C þ inflammatory monocytes ( Fig. 1h and Supplementary Fig. 2). To further investigate whether the effect of POP2 was limited to localized MSU crystal responses and DAMPs, we analysed the systemic response to intraperitoneal (i.p.) injection of bacterial LPS, which requires canonical NLRP3 and non-canonical inflammasomes 36,37 . LPS induced a potent neutrophil influx into the peritoneal cavity in WT mice, but this response was strongly reduced in POP2 TG mice (Fig. 1i). Concomitantly, IL-1b, IL-6 and TNF as well as LPS-induced cell death measured by lactate dehydrogenase (LDH) activity in cleared peritoneal lavage fluids were all reduced in POP2 TG mice (Fig. 1j). This reduced cytokine response was not limited to the peritoneal cavity, but was also notable on a systemic level, since serum IL-1b, IL-6 and TNF were also reduced in POP2 TG mice (Fig. 1k). Although the canonical NLRP3 inflammasome contributes to LPS-induced shock 37 , it is primarily mediated by the non-canonical inflammasome 36 . To better determine the contribution of POP2 during acute shock, we developed an NLRP3-dependent model. We primed mice with a low-dose LPS before inducing acute shock by i.p. injection of nigericin, which is a microbial toxin derived from Streptomyces hygroscopicus. This shock model absolutely depends on NLRP3, since all Nlrp3 À / À mice survived, but only 15% of WT mice survived past 15 h. POP2 TG mice were significantly protected since 78% of mice were still alive and healthy after 72 h, which was the termination point of this study (Fig. 1l). Taken together, our experiments clearly demonstrate that POP2 expression in monocytes and macrophages significantly ameliorates localized and systemic NLRP3 inflammasome responses to DAMPs, MAMPs and toxins in vivo. Interestingly, and in contrast to similar experiments performed with POP1 and POP3 transgenic mice, monocyte/ macrophage-specific expression of POP2 also reduces local and systemic expression of TNF and IL-6. In addition, POP2 potently prevented death during NLRP3-mediated acute shock.
Expression of POP2 in mouse macrophages inhibits caspase-1. POP2 binds to the PYD of ASC in human cells (Fig. 2a) 26 , and as expected from the high degree of homology between human and mouse ASC and the apparent phenotype that we observed upon POP2 expression in mice, human POP2 also co-purified murine ASC in immortalized bone marrow-derived macrophage (iBMDM; Fig. 2a). ASC polymerization is nucleated by active NLRP3 through PYD-PYD-mediated interactions and is a prerequisite for the subsequent proximity-induced caspase-1 activation. Experimentally, the ASC polymerization step can be captured by non-reversible crosslinking of total cell lysate (TCL) and the detection of monomeric and oligomeric ASC by immunoblotting 38 . We utilized the potent and highly specific NLRP3 activator nigericin in LPS-primed BMDM 7,37 to induce ASC polymerization and found that polymerized ASC, but not soluble ASC, was markedly reduced in LPS-primed POP2 BMDM compared to WT BMDM (Fig. 2b). This indicated that POP2 was capable of impairing NLRP3 inflammasome function, which was corroborated by the fact that nigericin-induced caspase-1 activation was significantly blunted in POP2 BMDM compared to WT BMDM, as determined by a caspase-1 fluorochromeinhibitor of caspases (FLICA)-based flow cytometry assay ( Fig. 2c and Supplementary Fig. 3). POP2 expression reduced nigericin-mediated caspase-1 activation from 19.6 to 5.4% of cells, which is only slightly less potent than the 1.4% of cells that we observed in Asc À / À BMDM. We obtained a comparable result for NLRP3 activation by extracellular ATP (Fig. 2d and Supplementary Fig. 3). In addition, the extracellular release of the active p10 subunit of caspase-1 and IL-1b (refs 39,40) in response to LPS priming and MSU stimulation was also reduced in POP2 BMDM compared to WT BMDM (Fig. 2e). Interestingly, under these conditions and at this time point we did not observe any significant reduction of IL-1b expression in TCLs of POP2 BMDM compared to WT BMDM, which indicates that the effect of POP2 on inflammasome-mediated processing of IL-1b is more pronounced than its NF-kB-inhibiting activity at these early time points (Fig. 2e). In addition, POP2 expression ameliorated the IL-1b release from BMDM in response to NLRP3 and AIM2 inflammasome activation with ATP or nigericin and poly(dA:dT), respectively ( Fig. 2f,g). Asc À / À BMDM showed no detectable IL-1b release in response to ATP (Fig. 2f). The POP2-mediated inhibition of IL-1b release was not restricted to soluble activators, but was also observed in response to the crystalline NLRP3 activators MSU, calcium pyrophosphate dehydrate (CPPD) and silica (SiO 2 ; Fig. 2h). This effect was observed in two independent POP2 TG lines underscoring its specificity and excluding its nonspecific effects from transgene integration-derived gene disruption (Fig. 2h). POP2 not only inhibited IL-1b release through the NLRP3 and AIM2 inflammasome, but also other ASC-dependent inflammasomes, including the Pyrin inflammasome activated with the Rho-glucosylating toxin TcdB from Clostridium difficile, the NLRC4 inflammasome activated with flagellin, and the NLRP1b inflammasome activated with lethal toxin from Bacillus anthracis (Fig. 2i) [41][42][43][44][45] . Secretion of IL-18 in response to nigericin, poly(dA:dT) and MSU, which is also dependent on caspase-1, was similarly reduced in POP2 BMDM (Fig. 2j). Caspase-1 activation not only promotes cytokine release, but also induces pyroptotic cell death, which results, among others, in the release of LDH and IL-1a (ref. 46). We observed reduced LDH release in POP2 BMDM and Asc À / À BMDM compared to WT BMDM upon nigericin treatment and DNA 10 (Fig. 2l). Overall, our results show that POP2 inhibits ASC oligomerization, caspase-1 activation, cytokine release and pyroptosis, which together can explain the ameliorated inflammatory responses observed in POP2 TG mice in vivo.
POP2 inhibits inflammasome priming in mouse macrophages. NLRP3 activation requires TLR-mediated priming of macrophages, and since POP2 has been implicated in inhibiting NF-kB activation in vitro 25 , we assessed the contribution of POP2-mediated modulation of macrophage priming on NLRP3 inflammasome activation. LPS priming of BMDM caused inflammasome-independent release of the pro-inflammatory cytokine TNF, and this response was therefore not affected in Asc À / À BMDM. However, POP2 BMDM showed a modest, but reproducible and significant reduction in the release of TNF (Fig. 3a) and IL-6 ( Fig. 3b) in response to LPS. This response was not limited to TLR4 activation by LPS, but was also observed in response to the TLR2 agonist Pam 3 CSK 4 (Fig. 3c). As TNF and IL-6 production are regulated by NF-kB through the phosphorylation and degradation of IkBa, we investigated LPS-mediated IkBa phosphorylation. Concurrent with the reduced cytokine release and the previously observed reduced p65 nuclear translocation 25 , we observed slightly reduced IkBa phosphorylation in POP2 BMDM compared to WT BMDM after LPS treatment (Fig. 3d). However, other key MAPK-induced signalling pathways, that is, p38, JNK and ERK were not significantly affected (Fig. 3d). Similar to POP2 BMDM, we found reduced IkBa phosphorylation in iBMDM stably expressing POP2 compared to WT iBMDM, establishing that the effect we observed is not due to transgene integrationmediated gene disruption (Fig. 3e). Next, we investigated the consequence of POP2 expression on the transcription of inflammasome components and cytokines by quantitative PCR (qPCR). In agreement with its NF-kB-inhibitory effect, LPSinduced transcription of Il1b was reduced in POP2 BMDM compared to WT BMDM, but transcription of Il-18 and Casp1 was not affected (Fig. 3f). Although Nlrp3 expression is known to be upregulated in response to LPS in an NF-kB-dependent manner 47 , we did not observe reduced Nlrp3 expression in POP2 BMDM, nor even slightly enhanced Nlrp3 mRNA expression levels. However, NLRP3 protein expression was not significantly altered in POP2 BMDM compared to WT BMDM, nor were ASC or caspase-1. In agreement with our mRNA results, pro-IL-1b protein levels were also reduced (Fig. 3g). Overall, our data demonstrate that POP2 inhibits inflammasome priming in addition to inflammasome complex formation, which likely contributes to the inhibition of inflammasome responses in vivo. Hence, POP2 has a unique and fundamentally different function in inflammasome regulation compared to POP1 and POP3, which do not affect NF-kB signalling 13,20 .   POP2 has a comparable function in human macrophages. To ensure that the responses and pathways that we observed in transgenic mice were relevant to human innate immune responses, we investigated inflammasome regulation in human macrophages and the human THP-1 monocytic cell line. We generated cells stably expressing POP2 fused to a 6xHis tag and analysed caspase-1 activation in response to NLRP3 activation with nigericin in LPS-primed cells by FLICA assay. As in BMDM, POP2 THP-1 cells displayed significantly reduced caspase-1 activity compared to control THP-1 cells, which did not increase over time (Fig. 4a,b and Supplementary Fig. 4). Reminiscent of BMDM, stable GFP (green fluorescent protein)-POP2 expression in THP-1 cells also significantly reduced caspase-1-mediated pyroptosis, as determined by kinetic analysis of propidium iodide (PI) uptake by live cells (Fig. 4c). Conversely, stable short hairpin RNA (shRNA)-mediated silencing of POP2 in THP-1 cells using two independent targeting sequences (shPOP2#1 and shPOP2#2), which both reduced POP2 expression by B75% (Fig. 4d), increased caspase-1 activation, as determined by FLICA assay (Fig. 4e and Supplementary Fig. 4). We next investigated cytokine release from LPS-primed GFP-POP2-expressing THP-1 cells, which showed markedly reduced IL-1b release in response to MSU crystals and nigericin compared to control cells (Fig. 4f). This response was independent of the fusion tag, as cells stably expressing His-POP2 showed a comparable response (Fig. 4g). In addition to IL-1b, IL-18 secretion was also impaired by POP2 expression (Fig. 4h). Conversely, LPS-primed shPOP2#1 and shPOP2#2 THP-1 cells displayed elevated IL-1b (Fig. 4i) and IL-18 (Fig. 4j) release in response to the NLRP3 agonists MSU crystals and nigericin, the NLRP7 agonist-diacylated lipopeptide FSL-1 and the AIM2 agonists dsDNA analogue poly(dA:dT). To further validate this response, we also silenced POP2 in primary human macrophages by transiently transfecting two independent POP2 short interfering RNAs (siRNAs), which also increased IL-1b release in response to MSU crystals and ATP in LPS-primed cells compared to scrambled control siRNA-transfected cells (Fig. 4k). To further study the mechanism of POP2mediated NLRP3 inflammasome inhibition, we investigated if POP2 affects the recruitment of ASC to activated NLRP3. Significantly, immunoprecipitation of ASC from LPS-primed and nigericin-treated THP-1 cells co-purified NLRP3 from WT and control THP-1 cells, but not from POP2-expressing cells (Fig. 4l). Control IgG was used as a specificity control and did not purify any NLRP3. Hence, the interaction between POP2 and ASC prevents the recruitment of ASC to activated NLRP3, which ultimately prevents ASC oligomerization, caspase-1 activation and cytokine release in human and mouse macrophages and provides a mechanism for the anti-inflammatory activity of POP2 on ASC-containing inflammasomes.
The NF-kB-inhibitory effect of POP2 observed in BMDM was also recapitulated in human cells since GFP-POP2 THP-1 cells (Fig. 5a) and His-POP2 THP-1 cells (Fig. 5b) secreted less IL-6 in response to LPS, MSU crystals and nigericin than control THP-1 cells, and displayed reduced IkBa phosphorylation in response to LPS (Fig. 5c). Conversely, shPOP2#1 and shPOP2#2 THP-1 cells showed elevated IL-6 levels in response to MSU crystals and nigericin (Fig. 5d) and displayed increased phosphorylation of IkBa in response to LPS (Fig. 5e). This confirms that POP2 functions as an inhibitor of inflammasome priming and activation in mouse and human macrophages. It is unclear how POP2 affects TLR-mediated NFkB activation, but we find that the kinase activity of the upstream canonical IKK complex was not altered in stable POP2-expressing cells after LPS treatment, which was demonstrated by the unchanged or even increased phosphorylation of IKKa/IKKb and TAK1 (Fig. 5f). However, while LPS-induced phosphorylation of the related non-canonical IKKe was impaired to a similar level as IkBa, phosphorylation of TBK1, an IKKe-binding kinase, was not altered (Fig. 5f). We observed a similar result in POP2 BMDM compared to WT BMDM (Fig. 5g). Conversely, we observed slightly accelerated and prolonged phosphorylation of IKKe in shPOP2#1 THP-1 cells (Fig. 5h). The role of IKKe in NF-kB signalling is not well established and, while it is sufficient to directly phosphorylate IkBa on S 36 (refs 48-52), it is not necessary for all NF-kB responses, and consequently only a select set of NF-kB-regulated genes is altered in Ikbke À / À mice, which interestingly includes Il1b (ref. 53). The increased NF-kB activity in POP2-silenced THP-1 cells directly translated into increased protein expression of pro-IL-1b, but did not have an impact on caspase-1 and NLRP3 expression (Fig. 5i). However, mRNA levels of IL1B and TNF, but not NLRP3, were significantly increased (Fig. 5j). Our findings indicate that POP2 interferes with the non-canonical NF-kB activation through IKKe, which contributes to its anti-inflammatory activity.
POP2 is a late-response regulator of inflammatory responses. The permanent cellular presence of an inflammasome inhibitor and thereby the constant blocking of the inflammasome would have negative implications on homeostasis 19 . Therefore, we investigated the expression pattern of POP2 in human macrophages and THP-1 cells. POP2 was potently induced in response to LPS in a time-dependent manner, peaking at 20 h in human macrophages (Fig. 6a) and 8 h in THP-1 cells (Fig. 6b), which is reminiscent of POP1 (ref. 13). The relatively late expression of POP2 suggests that it may be involved in the resolution of inflammasome responses. The inducible expression of POP2 was not restricted to LPS-mediated TLR4 activation, but also occurred in response to pro-inflammatory cytokines and interferons (Fig. 6c). Importantly, POP2 also responded to the anti-inflammatory cytokines IL-10/TGFb and IL-4/IL-13 with elevated expression, which polarize macrophages towards an alternative anti-inflammatory M2c and M2a type, respectively 54 , suggesting that POP2 was not only induced by pro-inflammatory stimuli to support resolution of inflammatory responses, but also by anti-inflammatory macrophages to eventually maintain an anti-inflammatory milieu. Although POP1 responds very similar to these stimuli, POP2 was generally more inducible and was also weakly induced by interferon (IFN)-b, which failed to upregulate POP1 (Fig. 6c). In addition, POP2 transcripts were increased even further in the presence of an NF-kB inhibitor during TLR2 or TLR4 activation with Pam 3 CSK 4 or LPS in primary macrophages and in THP-1 cells (Fig. 6d,e), while TNF transcripts were reduced as expected (Fig. 6e). Therefore, the ability of POP2 to inhibit NF-kB should elevate its own transcripts, which could ensure that sufficient POP2 levels are maintained during the resolution of inflammatory responses. Furthermore, LPSinducible POP2 expression was also recapitulated in GFP-POP2 THP-1 cells (Fig. 5c) and in BMDM-expressing POP2 from the hCD68/IVS-1 promoter (Fig. 6f), which supported our model system for in vivo studies of POP2. We also found that protein expression levels of POP2 in peripheral blood-derived monocytes (PBMCs) of POP2 TG mice ( Supplementary Fig. 5) inversely correlated to the levels of TNF and IL-1b induced by LPS and MSU, respectively, which substantiated the importance of regulating POP2 expression levels in vivo (Fig. 6g). The correlation between POP2 expression and IL-1b release (R 2 ¼ 0.4093) was stronger than between POP2 expression and TNF release (R 2 ¼ 0.2781), which might be due to POP2 affecting inflammasome priming and activation for IL-1b release, but only NF-kB-priming activity for TNF release. On the basis of the observed potent inflammasome-inhibiting activity of POP2, we tested whether exogenous POP2 could blunt inflammasome responses, and potentially function as a novel class of NLRP3 inflammasome-targeting drug that simultaneously prevents inflammasome priming and activation. Cell-penetrating peptides are frequently used for the intracellular delivery of proteins 55 , and we recently demonstrated that TAT-GFP is efficiently taken up by macrophages 13 . We therefore generated recombinant, highly purified cell-permeable POP2 by fusing it with the cell-penetrating HIV TAT peptide (TAT-POP2) and used TAT-GFP as a control. Addition of TAT-GFP to the culture medium did not significantly affect IL-1b release by nigericin in LPS-primed BMDM, while addition of TAT-POP2 significantly reduced IL-1b release (Fig. 6h). The results also showed that our purification approach excluded endotoxin contamination, as neither TAT-GFP-nor TAT-POP2primed macrophages caused any IL-1b release, similar to nigericin treatment alone in the absence of priming. Comparable results were obtained from iBMDM (Fig. 6i), demonstrating that intracellularly delivered POP2 can block NLRP3 inflammasome activity.

Discussion
ASC-containing inflammasomes, and particularly the NLRP3 inflammasome, are crucial for facilitating host defence, wound healing and tissue repair, and for maintaining metabolic health 3,56 . However, deregulated, excessive inflammasome responses also cause acute inflammation and auto-inflammatory ARTICLE diseases, and therefore a balanced NLRP3 response is essential for maintaining homeostasis 18,19 . Several regulatory factors are known to affect NLRP3 responses. Upstream regulatory events include lysosomal stability, mitochondrial integrity and membrane permeability; direct regulation on NLRP3 include kinases and deubiquitinases; and downstream signalling events include ASC and caspase-1 modification and activation. However, the key step linking activated NLRP3 to caspase-1 activation is the induced nucleation of ASC polymerization in human and mice, which ultimately drives the induced-proximity activation of  (j) Proposed function of POP2 as a dual regulator that simultaneously inhibits NF-kB-mediated inflammasome priming and nucleation in macrophages in response to MSU crystals and other activators. POP2 is positioned in a negative feedback loop for inflammasome regulation since it is upregulated by IL-1b, which is blocked upon POP2 expression. For NF-kB signalling POP2 is positioned in a feed-forward regulatory loop since its inhibition of NF-kB leads to POP2 upregulation. Together, POP2 efficiently prevents cytokine release and pyroptosis. Three independent repeats were performed for each experiment in a-i. Significance was calculated by a standard two-tailed unpaired t-test, and error bars represent s.e.m., *Po0.05. caspase-1 (refs 7,57). Since this step is mediated by homotypic PYD-PYD interactions, it is not surprising that several proteins regulating specifically this key step belong to the POP family 16,22 . We recently reported that POP1 and POP3 block crucial inflammasome interactions by either binding to ASC or AIM2-like receptors, respectively 13,20 . Here we show that a third POP family member, POP2, also regulates PYD-PYD interactions. While in vitro experiments indicated that POP2 regulates PYD-PYD interactions and NF-kB, it had been unknown whether POP2 also regulates these pathways in vivo, which is particularly challenging as all POPs are lacking from mice and evolved in higher primates and humans 22,29 . Therefore, we generated a transgenic mouse that expresses POP2 in the monocyte/macrophage lineage, which allowed us to demonstrate the anti-inflammatory function of POP2 in vivo, and also its protective function in a newly developed NLRP3-mediated acute shock model using a bacterial toxin. Importantly, we showed that human POP2 functions similarly in primary human macrophages and transgenic mouse macrophages, likely dependent on highly conserved PYDs. In contrast to other POPs, which solely interfere with inflammasome assembly and activation, POP2 also impaired the essential inflammasome-priming step by reducing TLR-induced NF-kB activation. Hence, POP2 has a dual inhibitory activity, which ultimately prevents inflammasome activation, pyroptosis, cytokine and ASC particle release. It is therefore feasible that POP2 has a unique position among NLRP3 regulators, since both activities are combined in a single protein.
Future studies will need to address the specific mechanism by which POP2 affects IkBa activation, which is still elusive, as its binding partner ASC does not affect activation of NF-kB in mice 58 . Nevertheless, there is some evidence that suggests that ASC may affect IkBa activation in human cells 59,60 . A POP2 effect on p65 transactivation has been suggested based on expression in HEK293T cells 33 , but this would be downstream of IkBa phosphorylation. However, we find that POP2 attenuates IkBa phosphorylation in macrophages. POP2 does not seem to affect the canonical NF-kB pathway, since activation of IKKa/b proceeded normally, but it affects the activation of the non-canonical IKKe. Interestingly, IKKe is sufficient to directly phosphorylate IkBa on S 36 (refs 48-52) and is required for transcription of a select set of NF-kB-regulated genes, including IL1B 53 . IKKe is not necessary for all canonical NF-kB responses 53 , but its precise role is still incompletely understood.
Although POP2 expression in our transgenic (TG) mice was below the detection level in naive mice, we could detect POP2 transcripts in monocytes by fluorescence-activated cell sorting (FACS) using Smartflares and POP2 protein in differentiated macrophages by immunoblot. In spite of this low basal expression level, POP2 significantly ameliorated NLRP3-mediated inflammation in vivo. We observed a close inverse correlation of POP2 expression levels and cytokine secretion, which further emphasized the importance of regulating POP2 expression levels. The inverse correlation between POP2 and IL-1b was stronger than the one between POP2 and TNF, in accordance with the dual inhibitory role of POP2, which blocks Il1b transcription by inhibiting NF-kB signalling through reduced IKKe and IkBa phosphorylation as well as the inflammasome-mediated caspase-1 activation responsible for IL-1b maturation and release, while TNF secretion is only blocked on a transcriptional level. In vivo we also observed that POP2 expression mediates a more pronounced reduction in IL-1b than in IL-6 and TNF in response to LPS. However, NF-kB-mediated priming and inflammasome activation are intimately intertwined and difficult to separate. Interestingly, POP2-mediated inhibition of NF-kB signalling occurs late during inflammatory responses but not on the immediate early NF-kB response. In contrast, POP2 affected caspase-1 activation even after short priming, which suggests a preferential effect on inflammasome activation over transcriptional priming, or that higher POP2 expression levels, after inducible upregulation, are necessary to reduce NF-kB activation. While these POP2-regulated signalling pathways are beneficial for host defence and metabolic and intestinal homeostasis, prolonged activation or defects in timely resolution can result in systemic hyperinflammation and sepsis-like syndromes deleterious for the host. We tested bacterial PAMPs and toxins, and POP2 ameliorated these responses; we, therefore, expect that POP2 would also ameliorate inflammatory responses triggered by bacterial infection. However, a POP2-mediated disruption of early host defence responses could potentially result in an unfavourable elevated bacterial burden. Therefore, the low basal expression of POP2 in conjunction with a late inducible expression of POP2 in response to pro-inflammatory stimuli may reflect a successful mechanism to preferentially dampen the excessive host responses, without impairing early host defence. Significantly, POP2 was also upregulated by anti-inflammatory cytokines, which polarize towards anti-inflammatory M2 macrophages. Hence, POP2 may not only contribute to the resolution and termination of inflammatory responses, but may also support the maintenance of an anti-inflammatory milieu when expressed in alternatively activated M2 macrophages. Furthermore, we provide evidence that POP2 transcripts are suppressed by NF-kB, as blocking NF-kB increased POP2 transcripts. Hence, POP2 is positioned in another regulatory loop, where it inhibits NF-kB, which in turn promotes its own expression. This feed-forward mechanism would ensure that sufficient POP2 levels are maintained during the resolution phase. NF-kB activation usually promotes expression of pro-inflammatory genes and suppresses anti-inflammatory genes. However, a fundamental mechanism of LPS or TLR tolerance during the resolution phase is characterized by impaired NF-kB activation and altered NF-kB dimer composition 61 . Hence, gene expression of the anti-inflammatory POP2 may be regulated by a comparable mechanism during the resolution phase or during NF-kB p65/p50 and p65/p65 inhibition. Since excessive NLRP3 inflammasome activation is linked to numerous inflammatory diseases, strategies that interfere with this pathway are excellent candidates for ameliorating a wide spectrum of inflammasome-linked disorders. We therefore evaluated a cell-permeable POP2 for its inflammasomeinhibitory activity, which could give rise to novel therapies targeting simultaneously inflammasome priming and activation.
Here we substantiate that cell-permeable POP2 mimetics can reduce NLRP3-mediated cellular responses.
In the same issue, Periasamy et al. 62 used a similar approach to determine the in vivo role of POP2 by generating POP2 TG mice under the human POP2 promoter and came to a comparable conclusion that inducible POP2 expression prevented ASC-containing inflammasome-dependent-and inflammasomeindependent pro-inflammatory responses. Surprisingly, despite reduced pro-inflammatory cytokines, including IL-18, POP2 TG mice displayed increased IFN-g levels, which protected from Francisella tularensis and Streptococcus pneumoniae infection 62 .
It is remarkable that humans/primates have evolved inflammasome-inhibitory proteins, including the POPs 16 that are lacking from mice. It is possible that differentially spliced inflammasome components perform this function in mice, such as ASC splice forms 63 , or other, as of yet unidentified POPs are present in the mouse genome that function as homologous or orthologous proteins. Although we have only recently started to appreciate the existence of these proteins in human innate immune cells, we already understand their important role in keeping inflammasome pathways tightly controlled. Our present study demonstrates the unique position of POP2 within the POP family to restrict inflammasome-mediated inflammatory disease as a dual activity inhibitor in vivo (Fig. 6j).

Methods
Mice. pCD68/IVS-1-POP2 was generated by replacing CAT in pCAT-Basic containing the human CD68 promoter and the MF-specific IVS-1 enhancer 13,20,30 with GFP-POP2 and flanking the cassette with AatII restriction sites. The AatII fragment was excised, purified and B6.TgN(CD68-POP2) TG mice (POP2 TG ) were generated by pronuclear injection into C57BL/6 embryos by the Northwestern University Transgenic and Targeted Mutagenesis Facility. C57BL/6J WT (#000664) and Casp1/11 À / À mice backcrossed for 410 generations onto C57BL/6J (#016621) were obtained from the Jackson Laboratories and C57BL/6J Nlrp3 À / À and Asc À / À mice from Vishva M. Dixit (Genentech) and described earlier 37,58 . Mice were housed in a specific pathogen-free animal facility, and all experiments were performed on age-and gender-matched, randomly assigned 8-14-week-old mice conducted according to procedures approved by the Northwestern University Committee on Use and Care of Animals. The investigators were not blinded to the genotype of mice.
MSU crystal preparation. MSU crystals were prepared by crystallization of a supersaturated uric acid solution as previously described 6,67 . Briefly, uric acid was dissolved in 0.01 M NaOH at 70°C, adjusted for pH 8 and filtered to 0.22 mM. After 7 days at room temperature, MSU crystals were rinsed twice with ethanol, air-dried and ultraviolet-sterilized.
LPS and MSU-induced peritonitis. Eight-to twelve-week-old male WT and CD68-POP2 mice had their abdomen shaved under anaesthesia, and were randomly selected for i.p. injection with PBS, LPS (2.5 mg kg À g , E. coli 0111:B4, #LPS25, Sigma-Aldrich) or MSU crystals (3 mg). After 4 h, mice were i.p. injected with XenoLight Rediject Inflammation probe (200 mg kg À 1 , #760536, PerkinElmer) or luminol (200 mg kg À 1 ) from a 50 mg ml À 1 stock solution of Luminol sodium salt (#A4685, Sigma), dissolved in sterile PBS and stored at À 20°C (ref. 68) and in vivo bioluminescence was captured by imaging (IVIS Spectrum, PerkinElmer) 10 min post injection with a 5-min exposure on anaesthetized mice 13,20 . Images were quantified with the Living Image software (PerkinElmer). Peritoneal lavages or blood (mandibular bleed) were collected at the indicated time points after PBS, LPS or MSU injection, and cytokine levels were quantified by ELISA.
MSU airpouch. Subcutaneous air pouches were generated as previously described 35 . Briefly, anaesthetized 8-14-week-old WT or POP2 TG mice were injected with 3 and 2 ml of sterile air into the subcutaneous tissue of the back on days 0 and 3. MSU crystals (3 mg) in 1 ml of sterile PBS or 1 ml PBS were injected into the pouch 7 days after the first air injection, and IVIS imaging for MPO activity was performed as described above 7 h later. Pouch lavage was performed 8 h after MSU injection on killed mice by injecting 2.5 ml of PBS containing 5 mM EDTA for analysis of the cellular infiltrate by flow cytometry.
For phospho-protein analysis, cells were starved for 1 h in a serum-free medium before lysis in 95°C-preheated Laemmli buffer at different time points, with or without LPS treatment, and proteins were separated by SDS-PAGE, transferred to polyvinylidene difluoride membranes and analysed by immunoblot analysis with the appropriate primary antibodies and horseradish peroxidase-conjugated secondary antibodies (whole-donkey antibody to rabbit IgG (#NA934V, GE Healthcare) and whole-sheep antibody to mouse IgG (#NXA931, GE Healthcare), SuperSignal West Femto Maximum Sensitivity Substrate (#34095, ThermoFisher) and image acquisition (Ultralum).
For GST pull down, POP2 was expressed from pGEX-4T1 and affinity-purified as a GST fusion protein from E. coli BL21 (ref. 26). Protein lysates were prepared from LPS-treated (4 h) iBMDM or THP-1 cells by lysis (50 mM HEPES pH 7.4, 120 mM NaCl, 10% glycerol, 2 mM EDTA, 0.5% Triton X-100, supplemented with protease inhibitors), and cleared lysates were incubated with immobilized GST-POP2 or GST control for 16 h at 4°C, followed by extensive washing with lysis buffer and analysis by immunoblot.