NLRP10 Affects the Stability of Abin-1 To Control Inflammatory Responses

This article is featured in In This Issue, p.1

The field of innate immunity is rapidly expanding with the understanding of the molecular details underlying the function of cytosolic pattern recognition receptors known as NOD-like receptors (NLRs). NLR proteins have a tripartite domain organization comprising an N-terminal effector domain, typically a pyrin domain (PYD) or the N-terminal caspase recruitment domain (CARD). A central nucleotide-binding domain, NACHT, mediates protein oligomerization, and a series of leucine-rich repeats (LRRs) at the carboxyl-terminus are involved in ligand sensing. NLRs are evolutionarily conserved between plants and mammals and have been implicated in sensing a wide range of cellular signals that most likely converge into a limited number of cellular pathways (1).

NLR proteins can be assigned to three functional categories: inflammasome-forming NLRs, NF-κB and MAPK (NF-κB/MAPK)-activating NLRs, and transcriptional regulators. NLRP1, NLRP3, and NLRC4 are among the best-characterized NLRs that activate inflammasome assembly by their oligomerization and recruitment of the adaptor protein apoptosis-associated speck-like protein containing a CARD (ASC) via their PYD, resulting in the activation of procaspase-1. Caspase-1 subsequently processes the proinflammatory cytokines pro-ILβ and pro–IL-18, which are involved in a range of cellular processes (2, 3). In contrast to the inflammasome-forming NLRs, the NF-κB/MAPK–signaling NLRs, exemplified by NOD1 and NOD2, contain a CARD N-terminal domain. They each bind different moieties of bacterial peptidoglycan (PGN) and can form oligomers interacting through their CARD domain with the receptor-interacting serine/threonine protein kinase 2 (RIPK2). The E3 ligase X-linked inhibitor of apoptosis protein (XIAP) is then recruited to the NOD-RIPK2 complex where it mediates the polyubiquitylation of RIPK2 (4–8). It further promotes the recruitment of the linear ubiquitin assembly complex (LUBAC), which forms linear ubiquitin chains. This allows for the activation of NF-κB as well as the MAPKs ERK, p38 and JNK, and subsequent production of inflammatory cytokines (9, 10). Many members of the NLR family have been connected to chronic inflammatory diseases, including Crohn disease, fever syndromes, cancer, and diabetes (11–13).

Although the better-described NLRs get a lot of attention, the majority of the 22 human NLR proteins are lacking characterization or a unifying underlying mechanism. One such poorly understood NLR is the NACHT, LRR, and PYD domain-containing protein 10 (NLRP10), which is unique, as it lacks the LRRs present on all other identified NLR proteins and are considered to be involved in ligand binding (14, 15). NLRP10 was first described as a protein with function in ASC and caspase-1 inhibition. It was subsequently shown that human NLRP10 can inhibit ASC-mediated NF-κB activation and IL-1β release in epithelial cells (14). Although Imamura et al. (16) could confirm inhibition of the IL-1β release in macrophages of NLRP10 transgenic mice, the murine Nlrp10 failed to inhibit ASC aggregation in knockin mice overexpressing Nlrp10. The physiological role of NLRP10 in inflammasome formation or activation, however, remains controversial, as Nlrp10 knockout mice have no obvious defect in NLRP3-mediated responses (17). By contrast, inflammasome-independent pro- and anti-inflammatory roles of NLRP10 in Shigella flexneri and Leishmania major infection were also described (18, 19), combined with work suggesting a role of NLRP10 in enhancing T cell responses by increasing IL-12 production by dendritic cells (20). NLRP10 is highly expressed in the epidermis of the skin, where keratinocytes express several pattern recognition receptors important for host defense against bacterial, viral, and fungal pathogens (18, 21–23). Genome-wide association studies suggest a link between NLRP10 and allergic and inflammatory pathways in atopic dermatitis, food allergies, and contact hypersensitivity. Most recently, NLRP10 was found upregulated in psoriasis, which is an immune-mediated genetic disease manifested in the skin and joints and shown to contribute to delayed-type hypersensitivity response in a mouse model of atopic dermatitis (24–27).

To elucidate further functions of NLRP10, we set out to identify new interaction partners of NLRP10. In the current study, we report on a yeast two-hybrid (Y2H) screen that identified the TNFAIP3-interacting protein 1 (Abin-1) as a novel interaction partner of human NLRP10. Abin-1 is a regulatory protein that was first discovered to have an inhibitory role on the transmembrane TNFR1 (28). Since then it was also found to affect the signaling of epidermal growth factor receptor (EGF-R), TLRs, and nuclear receptors such as peroxisome proliferator–activated receptors and the retinoic acid receptor (RAR) (29). Abin-1 is well known to inhibit NF-κB via binding to the ubiquitin-editing and deubiquitylating enzyme TNF-α–induced protein 3 (TNFAIP3 or Zinc finger protein A20) (30) either by physical interaction, linking A20 to ubiquitylated NEMO/IKKγ, or competing with NEMO for binding to polyubiquitinylated signaling proteins (30–32). Although Abin-1 was demonstrated to have effects on the signaling cascades of multiple receptors, interactions that control the cellular levels of Abin-1 itself have not been described until now.

In this paper, we demonstrate that NLRP10 forms a complex with Abin-1, involving ubiquitylation of NLRP10 and reduction of cellular levels of Abin-1 in a dose-dependent manner upon infection with invasive NOD1-activating bacteria. This provides a mechanistic insight into how NLRP10 contributes to the enhanced NOD1-mediated proinflammatory responses reported by us previously (18).

Y2H screening and data analysis were performed by Hybrigenics SA, Paris, France, essentially as described (33). The full NLRP10 was cloned into a Y2H vector optimized by Hybrigenics SA. NLRP10 was screened against a highly complex human placenta library (34).

Mouse embryonic fibroblasts (MEF) were isolated as described (35). Briefly, embryos were dissected on embryonic day 11.5 and sacrificed by decapitation. Samples were genotyped, and the remaining tissue was rinsed, dissected, and incubated in DMEM with 10% FBS plus 1% penicillin/streptomycin (Pen/Strep) at 100 IU/ml and 100 mg/ml, respectively. Nonadherent cells were removed after subsequent incubation. Cells were immortalized by transduction with SV40.

Primary skin fibroblasts were isolated from Nlrp10^−/− knockout and Nlrp10^flox/flox mice (24) essentially as described (36). Briefly, mouse ears were dissected and incubated in RPMI 1640 medium with 30% FBS, 3% Pen/Strep, and 1% amphotericin B with collagenase type 1A (C2674; Sigma-Aldrich). After overnight incubation, cells and tissue were resuspended and medium changed until cells reached confluence.

Bone marrow cells were isolated from the respective animals and cultured in RPMI 1640 with 2 mM l-glutamine, 50 μM β-mercaptoethanol, 1% Pen/Strep, and 10% FBS supplemented with 20 ng/ml GM-CSF (PeproTech). Fresh GM-CSF was added two times (day 3 and 5), and immature bone marrow–derived dendritic cells (BMDC) were harvested after 6 d.

HEK293T and HeLa cells were obtained from the American Type Culture Collection. Cells were cultivated at 37°C with 5% CO₂ in DMEM supplemented with 10% FBS and 1% Pen/Strep.

SDS-PAGE, immunoprecipitation, and immunofluorescence microscopyHEK293T cells were transiently transfected using Lipofectamin 2000 (Thermo Fisher Scientific) according to the manufacturer’s conditions, with the indicated plasmids (3 μg of plasmid per 6-cm dish). Cells were incubated overnight and subsequently lysed in NP-40 buffer (150 mM NaCl, 1% NP-40, 50 mM Tris-HCl, pH 7.5) containing phosphatase inhibitors (20 mM β-glycerophosphate, 5 mM NaF, 100 μM Na₃VO₄) and complete protease inhibitor mixture (Roche). Lysates were cleared for 20 min at 14,000 × g at 4°C. Subsequently, samples were either run on an SDS-PAGE gel or analyzed by immunoprecipitation.

Immunoprecipitation was carried out for 4 h at 4°C by adding anti-FLAG beads (M2 gel; Sigma-Aldrich), GFP-Trap_MA agarose beads (Chromotek), anti-HA probe (Y11, sc-805; Santa Cruz Biotechnology) bound to protein G Dynabeads or anti-Myc beads (9E10 agarose; Santa Cruz Biotechnology) to the cell extracts. The beads were precipitated by centrifugation steps and washed three times in NP-40 buffer before SDS loading buffer was added. Proteins were separated by Laemmli SDS-PAGE and transferred to a nitrocellulose membrane (Bio-Rad). Proteins were detected by incubation of the membrane consecutively with primary and secondary Abs and finally with SuperSignal West Femto Maximum Sensitivity Substrate (Pierce). Signals were recorded on an electronic camera system (Vilbert Fusion FX). Primary Abs were mouse anti-FLAG M2 (F3165; Sigma-Aldrich), rabbit anti-Myc (A-14; Santa Cruz Biotechnology), rabbit anti–Abin-1 (4664; Cell Signaling Technology), mouse anti–Abin-1 (1A11E3; Thermo Fisher Scientific), anti-GFP (Roche), anti-HA (Y11, sc-805; Santa Cruz Biotechnology), anti-GAPDH (FL-335; Santa Cruz Biotechnology), anti-Actin (A2066; Sigma-Aldrich), and anti-NLRP10 monoclonal rat Ab 8H2 as published in (18). Secondary Abs were HRP-conjugated goat anti-mouse IgG (170-6616; Bio-Rad) and HRP-conjugated goat anti-rabbit IgG (170-6515; Bio-Rad).

For indirect immunofluorescence microscopy, cells were seeded on coverslips, fixed in 4% paraformaldehyde in PBS, and permeabilized with 0.5% Triton X-100 for 5 min. Cells were incubated in 3% BSA in PBS. Staining was done by consecutive incubation with primary and secondary Abs in 3% BSA. The secondary Abs were Alexa 546–conjugated goat anti-mouse IgG or Alexa 546–conjugated goat anti-rabbit IgG (Molecular Probes). DNA was stained with Hoechst 33258 (Carl Roth; Sigma-Aldrich). The images were acquired on a Leica DMi8 microscope using high numerical aperture 40× and 63× objectives and processed using ImageJ (National Institutes of Health, Bethesda, MD) and the LasX software (Leica).

Bacterial infection of cells was performed using the S. flexneri strains BS176 afaE and M90T afaE as described previously (37). M90T afaE is a wild-type (WT) invasive strain of S. flexneri serotype 5a harboring the plasmid pIL22, which encodes the afimbrial adhesion afaE from uropathogenic Escherichia coli, whereas the BS176 is a virulence plasmid–cured, noninvasive isogenic strain of M90T (38). Briefly, logarithmically growing bacteria were added to the cells, which were transferred to antibiotic-free medium and incubated for 15 min at room temperature prior to transfer to 37°C. S. flexneri M90T afaE was added at a multiplicity of infection of 10 to cells, and the medium was replaced with DMEM containing 100 μg/ml gentamicin 30 min after incubation at 37°C.

Cells were stimulated with TNF (10 or 20 ng/ml; InvivoGen) or E. coli K12 LPS (10 ng/ml; InvivoGen) or MDP (10 μg/ml; InvivoGen).

Knockdown of Abin-1 and NLRP10 in HeLa cells was performed for 72 h using the Viromer Blue transfection reagent (Lipocalyx). For each reaction, 25 nM of the following small interfering RNAs (siRNAs) were used: Abin-1 5′-CAAGAAGTTGTTGATGAGCAA-3′ (SI00748727; Qiagen); NLRP10 (NLRP10A): 5′-CAGCTCCTATTTCACGGATGA-3′ (SI04328254; Qiagen); NLRP10 (NLRP10B): 5′-AAGGAGGGCAAAGATAATATA-3′ (SI00654423; Qiagen); and a nontargeting control siRNA (AllStars Negative Control; Qiagen) as described (39).

Gene reporter assays were performed as described (40). Briefly, HEK293T cells transfected with a β-galactosidase–encoding plasmid and the NF-κB luciferase reporter plasmid and increasing amounts of an Abin-1 expression plasmid normalized with cDNA using FuGENE6 (Roche). Cells were induced with 10 ng/ml TNF and incubated 16–24 h at 37°C. The next day, cells were lysed in 100 μl of NF-κB luciferase lysis buffer, and the luciferase activity was measured with a luminometer. The normalized luciferase activity was calculated by dividing luciferase activity through β-galactosidase activity.

pcDNA3.1 (Invitrogen), pcDNA-NLRP10-myc, pCMV-NLRP10-FLAG, pcDNA-NLRP10ΔPYD-myc, and pcDNA-NLRP10Δ-FLAG are described (18). pcDNA-NOD1 (41) pCMV–Abin-1–FLAG was generated by PCR-based cloning in pCMV-Tag2B. The Abin-1 E476AA mutation was generated using the quick-site mutagenesis protocol. pCMV-NLRP10, pCMV-NLRP10ΔPYD, pCMV-NLRP10K179A-GFP, pCMV-NLRP10PYD-GFP were generated by PCR-based cloning in a pCMVTag2B vector where the FLAG was replaced by eGFP. Mouse HA–Abin-1 plasmid was a gift from H. Häcker (St. Jude Children’s Research Hospital) and mouse FLAG-NLRP10 was obtained by cloning Mm.37991 cDNA clone IRCLp5011A1123D (Source Bioscience) into pCMV-Tag2B (Agilent Technologies). All cloned inserts were sequenced.

The levels of human IL-8 and msCXCL1 (keratinocyte-derived chemokines) were determined by ELISA using the DuoSet kits DY208 and DY453 (R&D, Biotechne) according to the manufacturer’s instructions.

Western blots were quantified by Vilber Imaging software and normalized to a loading control. Subsequent analysis was performed on GraphPad Prism 7 (GraphPad Software). Student two-sided t test was used where appropriate.

To find new interaction partners for human NLRP10, we conducted a Y2H screen using full-length human NLRP10 as bait. The screen identified Abin-1 (TNIP1) (Fig. 1A) as a promising interaction candidate. The identified prey encoded the ubiquitin-binding domain (UBD) of Abin-1, which also interacts with the poly-UBD of NEMO (30) (Fig. 1A), a previously published interaction partner of NLRP10 (18).

The physical interaction of Abin-1 and NLRP10 was subsequently confirmed in human cells by coimmunoprecipitations from transiently transfected HEK293T cells (Fig. 1B). We also found that the mouse homologs of Abin-1 and NLRP10 showed interaction when overexpressed in human HEK293T cells (Supplemental Fig. 1A). To establish the protein domains in NLRP10 responsible for the interaction with Abin-1, coimmunoprecipitations with myc–Abin-1 and various deletion constructs of GFP-NLRP10 were performed (Fig. 1B) using transiently transfected HEK293T cells. Myc–Abin-1 not only bound to the full-length GFP-NLRP10 but also to the Walker-A mutant and PYD deletion construct. By contrast, binding to the PYD domain could not be demonstrated. The NLRP10 interaction site on Abin-1 was also reported to mediate binding of NEMO (30), which we showed before to bind also NLRP10 (18). We subsequently mapped the interaction domain of NEMO and NLRP10 and show that NEMO binds to the PYD deletion construct and to full-length NLRP10, suggesting that NEMO might be able to compete for binding of Abin-1 to NLRP10 (Fig. 1C).

Recently, we showed that NLRP10 enhances proinflammatory responses toward infection with the invasive Gram-negative bacterial pathogen S. flexneri. Next, we asked if infection with S. flexneri affects the interaction between Abin-1 and NLRP10. Coimmunoprecipitation experiments in transiently transfected HEK293T cells showed that interaction of the two proteins increased with time of infection (Fig. 2A). This was also confirmed in a reverse pulldown (Supplemental Fig. 1B). During the coimmunoprecipitation experiments described above, we observed that the levels of ectopically expressed Abin-1 protein increased over time upon S. flexneri infection and that NLRP10 overexpression counteracted this effect and led to overall lower levels of Abin-1 (Fig. 2A). To exclude that this was due to a particularity of the expression constructs, we confirmed this effect using alternatively tagged expression vectors (Supplemental Fig. 1C). Abin-1 protein levels inversely correlated with the amount of NLRP10 expressed during S. flexneri M90T infection in both HEK293T and HeLa cells; however, noninvasive S. flexneri BS176, which are not able to trigger NOD1 signaling, did not induce reduction of Abin-1 levels (Fig. 2B).

NOD1 overexpression, which is known to lead to autoactivation of NOD1 and to a lesser extent TNF activation, also increased protein levels of Abin-1 (Fig. 2C). This effect of NOD1 was counteracted by NLRP10 expression (Fig. 2C). In line with the results obtained with bacterial infection, an increase in NLRP10/Abin-1 complex formation was also observed upon NOD1 overexpression (Fig. 2D), suggesting that NOD1 activation, rather than effects of the bacterial infection are mediating this response.

As Abin-1 was expressed from the highly active CMV promotor, we assumed that changes in Abin-1 protein levels were due to differential regulation of protein stability. Accordingly, we did not observe any effect of overexpression of NLRP10 on the mRNA levels of neither endogenous nor ectopically expressed Abin-1 (Supplemental Fig. 2A, 2B). We thus tested if NLRP10 contributed to proteasomal degradation of Abin-1. To this end, HEK293T cells expressing Abin-1 were treated with cycloheximide (CHX) to block protein neosynthesis, and protein levels were analyzed over time. First, although NLRP10 was readily degraded in a time-dependent manner (Fig. 2E), the Abin-1 protein appeared to be more stable (Fig. 2E). Overexpression of NLRP10 did not have an effect on the rate of Abin-1 degradation (the decrease in protein over 4 h of CHX treatment), although it had an effect on the total Abin-1 protein levels (for quantification see Supplemental Fig. 2C). In contrast to nontreated or TNF-treated cells, overexpression of NOD1 led to a more pronounced stabilization of NLRP10 and degradation of Abin-1 (Fig. 2E, lower panel), suggesting a contribution of NOD1 and/or NOD1 signaling.

To visualize the levels of endogenous Abin-1, we conducted indirect immunofluorescence studies in HeLa cells transiently transfected with GFP-NLRP10. Consistent with previous reports, the Abin-1 Ab detected dot-like structures (42), whereas NLRP10 was homogenously distributed throughout the cell (18). In the course of infection, cells overexpressing GFP-NLRP10 showed lower signals for endogenous Abin-1 compared with neighboring cells without NLRP10 overexpression (Fig. 2F, white arrows; Fig. 2G for quantification). In some cells overexpressing NLRP10, endogenous Abin-1 formed more pronounced foci, which colocalized with NLRP10 over the time of infection (Fig. 2F, 4 h time point). Data generated with primary cells from mice corroborated these findings. We observed higher cytosolic signals upon Shigella infection with an anti–Abin-1 Ab when staining MEF derived from Nlrp10^−/− mice but not in MEF derived from Nlrp10^fl/fl mice by indirect immunofluorescence (Supplemental Fig. 3A, 3B). Although we could not obtain robust detection of endogenous Abin-1 in immunoblot to validate the immunofluorescence data, similar results were obtained with isolated primary skin fibroblasts from Nlrp10^fl/fl and Nlrp10^−/− mice, corroborating these results (Supplemental Fig. 3C).

To functionally link Abin-1 to Shigella-induced proinflammatory signaling, we conducted siRNA-mediated knockdown of NLRP10 and Abin-1 in HeLa cells, which were subsequently infected with S. flexneri. Bacterial-induced IL-8 levels, which are mainly dependent on NOD1 (43), as well as basal IL-8 secretion were significantly increased upon Abin-1 knockdown (Fig. 3A). In line with our earlier observation (18), reduction of NLRP10 led to decreased IL-8 release from the cells (Fig. 3A). Accordingly, NLRP10 depletion resulted in enhanced Abin-1 protein levels independent of stimulation (Fig. 3B). When combining knockdown of Abin-1 and NLRP10, we observed a reduction of both, Shigella- and TNF-induced IL-8 responses. However, the TNF-induced IL-8 release was only marginally affected by NLRP10 knockdown compared with the Shigella-mediated response (Fig. 3A). siRNA treatment thereby significantly reduced expression of the target proteins (Fig. 3B). Notably, we observed that Abin-1 depletion led to enhanced NLRP10 levels (Fig. 3B). This showed that changes in Abin-1 levels correlate with proinflammatory responses induced by S. flexneri and that NLRP10 can counteract Abin-1 function. In addition, we measured IL-1β but did not detect any release of IL-1β from these cells in this experiment (data not shown).

Taken together, our data showed that activation of NOD1 increased the complex formation between NLRP10 and Abin-1. NLRP10 expression led to degradation of Abin-1, which was enhanced by NOD1. Combined with previously reported data, this suggests that NLRP10 supports proinflammatory signaling by decreasing Abin-1 levels.

To expand on our in vitro data, we next analyzed NLRP10 and Abin-1 expression in primary cells derived from Nlrp10 knockout animals. We observed that keratinocytes showed the highest expression of Nlrp10, albeit low expression of Abin-1. However, immune cells such as bone marrow–derived macrophages (BMDM) and BMDC showed good expression of both Nlrp10 and Abin-1 (Fig. 4A). Infection of GM-CSF–differentiated Nlrp10^−/− BMDC with S. flexneri led to a trend toward lower CXCL1 and IL-6 release upon 6 h of infection compared with WT cells, which was significant for CXCL1 after 6 h of Shigella infection (Fig. 4B). The proinflammatory cytokine release of BMDC upon MDP and LPS stimulation was not significantly different between the genotypes (Fig. 4B). In BMDM, no significant effect of NLRP10 on cytokine release was observed (data not shown). Lack of a pronounced phenotype of NLRP10 depletion in mouse myeloid cells might be explained by a differential regulation of Abin-1 levels in these cells upon infection or the differences in Abin-1 expression in these cells (Fig. 4A). Indeed, we observed that in contrast to our data on human epithelial cells and murine MEF and keratinocytes, endogenous Abin-1 levels were highly reduced upon infection in dendritic cells and were not affected by NLRP10 expression, suggesting differences in Abin-1 regulation in these cell types (Supplemental Fig. 3D).

To conclude, depletion of endogenous Nlrp10 led to slight reduction of CXCL1 and IL-6 responses toward Shigella infection in BMDC. However, mouse myeloid cells showed a differential regulation of Abin-1 levels, suggesting that Abin-1 might not be a target of NLRP10 in these cells.

Abin-1 is a ubiquitin sensor, and modulation of ubiquitylation is an important strategy of Shigella pathogenesis (30). To examine if ubiquitin binding of Abin-1 is involved in its regulation by NLRP10, we tested if a UBD in ABIN proteins and NEMO (UBAN)-domain mutant of Abin-1 (E476AA) showed the same kinetics and effects upon infection and NLRP10 overexpression. The E376AA mutation lies in the highly conserved DFxxER motif of the ABIN homology domain 2 (AHD2) in the UBAN of Abin-1 that confers binding of Abin-1 to ubiquitin. The introduction of this mutation has been shown to render Abin-1 unable to downregulate NF-κB (31) (Supplemental Fig. 1D). Side-by-side comparison of WT Abin-1 and Abin-1 E476AA showed that upon infection with S. flexneri M90T, WT Abin-1 levels decreased more in response to NLRP10 overexpression (4 h postinfection [p.i.] ∼17-fold reduced) than the levels of Abin-1 E476AA (4 h p.i. ∼1.5-fold reduced) (Fig. 5A). In general, Abin-1 E476AA levels were higher compared with the WT protein and less regulated upon infection (Fig. 5A). Accordingly, we observed a reduced binding of Abin-1 E476AA to NLRP10 in the presence of NOD1, suggesting that ubiquitylation of NLRP10 might be involved (Fig. 5B). We then asked if Abin-1 affects the ubiquitylation status of NLRP10. We could detect that NLRP10 is ubiquitylated under steady-state conditions and that this ubiquitylation is reduced upon the overexpression of Abin-1, whereas Abin-1 E476AA was slightly less efficient to drive deubiquitylation of NLRP10 (Fig. 5C). We also analyzed the effect of NLRP10 and NOD1 on Abin-1 ubiquitylation. This revealed increased ubiquitylation of WT but not of Abin-1 E476AA, whereas NLRP10 overexpression reduced Abin-1 levels and accordingly the ubiquitin signal. NOD1 expression increased Abin-1 ubiquitylation and led to higher molecular complexes (Fig. 5D).

These data highlight the importance of the ubiquitin-binding property of Abin-1 in downstream regulation of NF-κB upon NOD1 stimulation. The UBAN mutant of Abin-1 did not show the same binding capacity to NLRP10, and the levels of this Abin-1 mutant were not affected by NLRP10 overexpression. As Abin-1 E476AA is unable to downregulate NF-κB signaling, this supports the functional role of the NLRP10/Abin-1 complex in regulating inflammatory pathways.

Our current understanding of NLRP10 is limited, and its function in immune signaling pathways require further clarification. NLRP10 has been proposed to have both anti-inflammatory and proinflammatory properties. Some groups suggested an inhibitory role for NLRP10 in ASC-mediated NF-κB activation as well as caspase-1–mediated IL-1β secretion (14, 16). However, Krishnaswamy et al. (44) did not find any effect of NLRP10 on the inflammasome activation in macrophages and dendritic cells of NLRP10 knockout mice. Recent studies showed a role of NLRP10 in the resolution of local inflammatory responses during L. major infection (19) and a positive effect of NLRP10 on the IL-12 release by dendritic cells (20). The pathways underlying these diverse functions of NLRP10 remain to be uncovered.

In this study, we identified the A20-binding inhibitor of NF-κB (Abin-1) as a novel interaction partner of human NLRP10. This interaction was verified in coimmunoprecipitation experiments with human cells. Domain mapping identified the PYD to be dispensable for binding. Although a somewhat better binding of Abin-1 to a Walker A box mutation in NLRP10 was seen, this might be nonphysiological, as mutations in the ATP binding site can cause conformational instability in NLRs. In HeLa cells, the two proteins partially colocalized during infection with the invasive Gram-negative bacterial pathogen S. flexneri, whereas the interaction between the two proteins increased with time of infection.

In previous work, we showed that NLRP10 contributes to skin contact hypersensitivity responses (24) and positively regulates innate immune responses mediated by NOD1 upon S. flexneri infection in both epithelial cells and dermal fibroblasts via modulation of p38 and NF-κB signaling pathways (18). NOD1 participates in sensing PGN moieties and has been previously implicated in Shigella infection (43). NOD1 signals via downstream interactions with CARD-containing kinase RIPK2. Of note, a co-occurrence of NOD1, RIPK2, NLRP10, and Abin-1 is found across genomes (Supplemental Fig. 4), suggesting a conserved function of the complex described in this paper. Of note, ties of Abin-1/NEMO to S. flexneri response have been reported, whereby IpaH9.8, a Shigella effector possessing E3 ligase activity, interacts with Abin-1 to promote NEMO polyubiquitylation and subsequent degradation (45). Our work reveals a novel aspect of Abin-1 function, as it shows an upregulation of Abin-1 protein levels, which is in some cells dependent on NLRP10 and NOD1 activation. Why Abin-1 levels are not regulated in the same manner in different cell types and why there is only partial dependency on NOD1 signaling needs to be addressed in future research.

Abin-1 is known to modulate different cellular signaling pathways, and its list of interacting partners includes a range of cellular and nuclear receptors (reviewed in Ref. 29). It is involved in a number of cell-wide signaling events, including reducing activation of transcription factors Elk2 and C/EBPβ, coexpressing peroxisome proliferator–activated receptor and RAR, binding the HIV protein Nef or matrix. Its specific interaction partners include NEMO, A20, RIP-1, TRAF-1, FADD, p105, Nef, and ERK2 (reviewed in Ref. 46). It acts downstream of the TNF-α receptor via an interaction with NEMO (IKKγ), reducing IKK complex activation and decreasing NF-κB–dependent gene expression, thereby inhibiting NF-κB, JNK, ERK, and p38 MAPK signaling and acting as an important regulator of homeostasis (47, 48). This study expands on these previous observations and identifies a novel Abin-1/NLRP10 complex formation during bacterial S. flexneri infection. A change in the stoichiometry of the NLRP10/Abin-1 complex was observed after activation of NOD1, as the complex formation increased. This effect was also observed by immunofluorescence whereupon Abin-1 seemed to be differently distributed in cells that overexpress NLRP10 and both Abin-1 and NLRP10 showed a more pronounced colocalization p.i. The interaction between Abin-1 and NLRP10 was mediated by the NACHT domain of NLRP10, which also binds NEMO, whereas the binding of NEMO was negatively affected by the presence of the PYD domain. This might suggest that proteins occupying the PYD domain might regulate NEMO binding and raises the possibility that NEMO and Abin-1 might compete for binding to NLRP10. We assume that this competition could serve as a switch mechanism to direct NLRP10-mediated inflammatory response to various triggers.

We also demonstrate that NLRP10 is able to control cytosolic levels of Abin-1. This is pathologically relevant as GWAS and array studies show a correlation of Abin-1 levels with multiple pathologies such as systemic lupus erythematosus, systemic sclerosis, rheumatoid arthritis, and leukemia/lymphoma (29, 49, 50). Germline knockout of Abin-1 results in fatality and liver apoptosis with anemia (28), whereas knockin of an Abin-1 form which lacks the ability to interact with NEMO leads to a lupus-like autoimmunity in the spleen and lymph nodes (47). Levels of Abin-1 are also critical in psoriasis development as blocking of Abin-1 increased proliferation of keratinocytes in psoriasis, whereas overexpression decreased keratinocyte proliferation and had a protective effect. This is especially interesting as NLRP10 was recently shown to be upregulated in psoriasis (27). Thus, although transcription of TNIP1 is regulated in part by all-trans retinoic acid (ATRA) and RAR (51), we now show that cellular protein levels of Abin-1 are posttranslationally regulated by NLRP10. The binding of NLRP10 that regulates available Abin-1 levels could be a contributing aspect of disease progression. A notable point is the comorbidity of psoriasis with Crohn disease (a NOD1/2–linked condition) (reviewed in Ref. 52), further strengthening the importance of the NOD/NLRP10/Abin-1 pathway.

Exactly how NLRP10 decreased the stability of Abin-1 at the molecular detail and why there is only a certain degree of specificity for NOD1 activation remains elusive. Of note, in HeLa cells, Abin-1 reduction also led to higher endogenous NLRP10 protein levels, suggesting a more complex interplay. Polyubiquitylation of NLRP10 might play a role here, as we could show that Abin-1 affects the ubiquitination status of NLRP10. The UBD (UBAN) of Abin-1 is known to be specifically responsible for binding to polyubiquitin and polyubiquitinylated proteins (28), and disruption of polyubiquitin binding in Abin-1 promotes NF-κB and proinflammatory signaling, which in vivo is associated with glomerulonephritis and lupus-like disease (31, 53, 54). We consequently observed that the interaction of the UBAN mutant Abin-1E476AA has a weaker binding affinity for NLRP10 upon NOD1 overexpression and that the levels of this mutant were not affected by NLRP10 overexpression upon S. flexneri infection. It is plausible that other E3 ligases acting downstream of NOD1, such as XIAP (4, 55), might be involved in this process. Given the involvement of Abin-1 in proinflammatory responses, these data could be a stepping stone in clarifying the role of NLRP10 in inflammatory reactions in vivo. Of note, mouse myeloid cells exhibited a different regulation of Abin-1 upon infection compared with fibroblasts and human epithelial cell lines. Whereas upon infection with S. flexneri Abin-1 protein levels decreased in dendritic cells, Abin-1 protein levels were enhanced in fibroblasts, human HeLa, and HEK293T cells. Furthermore, depletion of NLRP10 had only mild effects on cytokine responses in myeloid murine cells. Compared to the clear effects in human and mouse epithelial cells, this suggests that NLRP10 has a primary function in the epithelium. It remains to be established if there might be different regulatory networks triggered by NLRP10 in human and mouse cells. A predominant role of NLRP10 in epithelial versus myeloid cells is also in line with the rather subtle phenotypes observed in NLRP10 knockout mice upon infection, as these responses are mainly governed by myeloid immune cells.

In summary, we provide mechanistic insight into the signaling cascade of NLRP10 and how it impacts NOD1-mediated pathways. We revealed a novel interaction between NLRP10 and Abin-1 and a subsequent regulation of Abin-1 protein levels by NLRP10, which can explain not only the regulation of Abin-1 but also the duality of the pathways currently described for NLRP10. We propose that the interaction between NLRP10 and Abin-1 serves as one node dictating further downstream inflammatory effects and believe this information sheds light on the role of NLRP10 in innate immune signaling.

We thank Yvonne Postma for excellent technical assistance and Kornelia Ellwanger for help with the mouse studies and for organization of animal care.

The authors have no financial conflicts of interest.