The Crohn’s disease–related RNF123 prevents NLRP3 inflammasome assembly by catalyzing unanchored K63-linked ubiquitination on NEK7

Significance

Inflammation is a double-edged sword. A moderate inflammatory response is vital for maintaining immune homeostasis, yet excessive inflammation underlies many human diseases. We uncovered RNF123 as a natural brake on the NLRP3 inflammasome-the cellular machinery that triggers intense inflammation. When RNF123 is missing or mutated (e.g., Crohn’s disease–related SNP R854H), this brake fails and inflammation spirals. Mechanistically, RNF123 adds special K63-linked polyubiquitin chains to NEK7, preventing it from switching NLRP3 into an active, inflammation-driving state. Our findings identify RNF123 as a promising drug target for treating the NLRP3-driven inflammatory diseases.

Abstract

The NLRP3 inflammasome is crucial for host defense against pathogen invasion and is implicated in various inflammatory disorders. The pathogenic association and the involved mechanism between the NLRP3 inflammasome and inflammatory diseases have garnered significant attention. Here, we demonstrate that Crohn’s disease–associated SNP RNF123-R854H aggravates colitis in vivo through the NLRP3-dependent pathway. Deficiency of RNF123 also aggravates dextran sodium sulfate-induced colitis, LPS (lipopolysaccharide)-induced endotoxemia, and Alum-induced peritonitis and enhances host defense against bacterial infection via the NLRP3-dependent pathway in vivo and promotes the NLRP3 inflammasome activation in cells. We establish RNF123 as a regulator for the NLRP3 inflammasome, highlighting its implication in the NLRP3 inflammasome-driven inflammatory diseases. Mechanistically, RNF123 catalyzes unanchored K63-linked ubiquitination of NEK7, thereby preventing NEK7-mediated dissociation of the inactive cage-like NLRP3 aggregates and the subsequent NLRP3 inflammasome assembly. Additionally, we prove that K63-linked polyubiquitin chains can be specifically captured by NEK7 in vitro and inhibit the NEK7-licensed NLRP3 inflammasome assembly. We propose that NEK7-captured unanchored K63-polyubiquitin chains serve as a key determinant for the NLRP3 inflammasome activation, acting as a molecular brake to limit the excessive NLRP3 inflammasome activation and preserve immune homeostasis. Our work yields mechanistic insights into the NLRP3 inflammasome regulation and its pathogenic link to inflammatory disease.

The inflammasome is a multiprotein complex that functions as a platform for the activation of Caspase-1, which triggers the processing and secretion of inflammatory cytokines, most prominently interleukin 1β (IL-1β), and induces pyroptosis, a form of lytic cell death (1, 2). The NLRP3 inflammasome has been most extensively studied and regarded as a highly sensitive, yet nonspecific inflammasome, since it can potently respond to the destruction of cellular homeostasis by a broad range of environmental irritants, rather than being activated by a specific microbe-derived molecule (1, 2). Therefore, the NLRP3 inflammasome not only plays a central role in antimicrobial defense, but also is associated with the pathogenesis of several acquired inflammatory diseases, such as septic shock, peritonitis, and inflammatory bowel disease (1, 2).

The NLRP3 protein is composed of three domains: an amino-terminal (N-terminal) pyrin domain (PYD) that recruits apoptosis-associated speck-like protein containing a CARD (ASC), a central nucleotide-binding (NACHT) domain that possesses ATPase activity and mediates self-oligomerization, and a carboxy-terminal (C-terminal) leucine-rich repeat (LRR) domain that participates in maintaining the NLRP3 autoinhibited state to avoid unwanted activation (1, 2). Typically, the activation of the NLRP3 inflammasome is delineated into two phases: priming and activation (2). The priming step is marked by the elevated expression of NLRP3 and pro-IL-1β, driven by the activation of the nuclear factor (NF)-κB signaling pathway (2). Subsequently, in response to a variety of activating stimuli, such as extracellular ATP, nigericin (Nig), and crystalline substances, NLRP3 undergoes oligomerization and recruits pro-Caspase-1 through the adapter ASC (2). This multiprotein complex serves as a signaling platform that triggers the activation of Caspase-1, which then processes pro-IL-1β and pro-IL-18 into their mature, secreted forms, and cleaves gasdermin D (GSDMD) to mediate pyroptosis (2).

Recent advancements have identified the never in mitosis A (NIMA)-related kinase 7 (NEK7), a member of the family of mammalian NIMA-related kinases (NEK) (3), as a component of the NLRP3 inflammasome (4–6). Before stimulation, NLRP3 forms a preassembled and cage-like complex held together by its LRR–LRR domain interactions (7–9). Within this assembly, the PYD of NLRP3 is sequestered to prevent premature activation (7–9). The binding of NEK7 to the concave site of the NLRP3-LRR domain serves as a licensing step for NLRP3 disassembly (7–10). This action exposes the NLRP3-PYD, thereby enabling its interaction with ASC and initiating subsequent activation (7–10). The regulatory mechanism of NEK7 on the NLRP3 inflammasome has emerged as a critical focus in contemporary research (11–14), and an increasing number of small-molecule compounds have been developed to regulate NLRP3 inflammasome activation through the on-target regulation of NEK7 (15). Therefore, NEK7 has been regarded as an attractive therapeutic target for treating inflammatory diseases.

The NLRP3 inflammasome has been established as a crucial regulator of intestinal homeostasis and dysregulation of the NLRP3 inflammasome is closely associated with the development of Crohn’s disease (16–18). However, the exact connections and molecular mechanisms tying the NLRP3 inflammasome to Crohn’s disease pathogenesis warrant further investigation. The previous Genome-wide Association Studies (GWAS) have pinpointed nine genetic regions where specific tag SNPs (Single Nucleotide Polymorphisms) are strongly linked to an elevated risk of Crohn’s disease (19). Among these, a region on chromosome 3 contains three distinct marker SNPs (19). Two of these markers reside within intronic regions of the bassoon gene, which encodes a presynaptic cytomatrix protein, and the third marker is a synonymous SNP within the same gene (19). This region of linkage disequilibrium on chromosome 3, specifically in the 3p21 band, spans an unusually large stretch encompassing 20 genes (19). Penetrance analysis has identified the RNF123-R854H variant as a candidate nonsynonymous SNP at this chromosome 3 locus, associated with an increased risk of Crohn’s disease (19). However, the pathogenic link and underlying mechanism of this RNF123 variant with the pathogenesis of Crohn’s disease remain elusive. RNF123 is an E3 ubiquitin ligase responsible for the ubiquitination of several substrates, but its role in inflammatory diseases, e.g., Crohn’s disease, has not been investigated (20–24).

Here, we show that RNF123-R854H mutation exacerbates intestinal inflammation in the in vivo colitis model via the NLRP3-dependent pathway, validating the pathogenic link of RNF123 with Crohn’s disease. We demonstrate that RNF123 negatively regulates the NLRP3 inflammasome activation in vivo and in cells, establishing RNF123 as a regulator for the NLRP3 inflammasome. Mechanistically, RNF123 catalyzes the unanchored lysine 63(K63)-linked ubiquitination of NEK7, thereby preventing the NEK7-mediated dissociation of NLRP3 aggregates and subsequent inflammasome assembly. Additionally, K63-linked polyubiquitin chains can also be specifically captured by NEK7 in vitro and directly inhibit the NEK7-licensed NLRP3 inflammasome assembly. We posit that NEK7-captured unanchored K63-linked polyubiquitin chains serve as a key determinant for the NLRP3 inflammasome activation, acting as a molecular brake to limit the excessive NLRP3 inflammasome activation and preserve immune homeostasis.

Results

RNF123 Correlates with Crohn’s Disease.

Given that RNF123-R854H has been linked as a potential high-risk pathogenic variant for Crohn’s disease (19) (Fig. 1A), we constructed mice harboring the RNF123-R854H mutation (Rnf123^R854H) and mice deficient in RNF123 (Rnf123^−/−), and verified the function of RNF123 and its R854H mutation in a mouse model of experimental colitis. We confirmed the R854H mutation in Rnf123^R854H mice by sequencing and validated the RNF123 expression in Rnf123^R854H and Rnf123^−/− mice by immunoblot (SI Appendix, Fig. S1A). We then induced colitis in these mice by treating them with dextran sodium sulfate (DSS). Compared to the wild-type littermates (Rnf123^WT), Rnf123^R854H and Rnf123^−/− mice were highly susceptible to the DSS-induced colitis and showed a reduced survival rate, more dramatic weight loss, shortened colon length, and increased clinical severity (Fig. 1 B–E and SI Appendix, Fig. S1B). Hematoxylin and eosin staining revealed that Rnf123^R854H and Rnf123^−/− mice exhibited higher histological scores, characterized by increased inflammatory cell infiltration, mucosal subepithelial edema, epithelial detachment, and the presence of widespread ulcers (Fig. 1F). Additionally, the concentrations of the proinflammatory cytokines IL-1β and IL-18 released from the Rnf123^R854H and Rnf123^−/− mouse colons were significantly higher than those in the Rnf123^WT mice, whereas the TNF-α and IL-6 levels were comparable among these mice (Fig. 1G and SI Appendix, Fig. S1C).

Fig. 1.

RNF123 correlates with Crohn’s disease. (A) Information of SNP RNF123-R854H. (B–E). The survival rates (B), the changes of body weight (C), colon length (D), and clinical score (E) of Rnf123^WT, Rnf123^R854H, and Rnf123^−/− mice in DSS-induced colitis model (mice: n = 10 for each group). (F) Hematoxylin and eosin–stained images of the colons from Rnf123^WT, Rnf123^R854H, and Rnf123^−/− mice in DSS-induced colitis model. The histological scores are shown in the Right panel (mice: n = 5 for each group). (Scale bar, 100 µm.) (G). ELISA analysis of the IL-1β and IL-18 released into the colonic culture supernatants from Rnf123^WT, Rnf123^R854H, and Rnf123^−/− mice in DSS-induced colitis model (mice: n = 10 for each group). (H). Clinical scores of Rnf123^WT, Rnf123^R854H, and Rnf123^−/− mice with MCC950 treatment or Nlrp3-knockout (Nlrp3^−/−) in DSS-induced colitis model (mice: n = 10 for each group). (I). ELISA analysis of the IL-1β released into the colonic culture supernatants from mice in (H). (J) Immunoblot analysis of the IL-1β and Caspase-1 cleavage in the colons from mice in (H). (K) SDD-AGE analysis of the ASC aggregation in the colons from mice in (H). Data are represented as mean ± SD. In (B), the log-rank (Mantel-Cox) test was performed for survival curve (DSS group). In (C–I), the two-way ANOVA was performed. *P < 0.05, **P < 0.01, ***P < 0.001. β-actin was used as a loading control. DSS, dextran sodium sulfate; WT, wild-type.

The NLRP3 inflammasome is closely associated with the development of Crohn’s disease (16–18). Thus, we wondered whether the effect of RNF123 deficiency and the RNF123-R854H mutation on intestinal inflammation relies on the NLRP3 inflammasome. We employed NLRP3 inhibitor MCC950 (25) in the DSS-induced colitis model and found that MCC950 alleviated the clinical severity of colitis induced by DSS and reduced the production of IL-1β in mouse colons (Fig. 1 H and I and SI Appendix, Fig. S1D). The R854H mutation and deficiency of RNF123 could not aggravate colitis and facilitate the IL-1β production upon MCC950 treatment (Fig. 1 H and I and SI Appendix, Fig. S1D). Further, we bred Rnf123^WT, Rnf123^R854H, and Rnf123^−/− mice with Nlrp3-knockout mice (Nlrp3^−/−) to generate corresponding NLRP3-deficient mice. We found that the Rnf123^R854H, Rnf123^−/−, and Rnf123^WT mice with NLRP3 deficiency exhibited similar severity of colitis and IL-1β production (Fig. 1 H and I and SI Appendix, Fig. S1D). Additionally, we assessed key indicators of inflammasome formation in colon tissues from mice in the DSS-induced colitis model. We detected the cleavage of Caspase-1 and IL-1β by immunoblot, and ASC aggregation by semidenaturing detergent agarose gel electrophoresis (SDD-AGE) (26, 27). We found that both the RNF123-R854H mutation and RNF123 deficiency promoted the cleavage of Caspase-1 and IL-1β, as well as ASC aggregation, indicating an elevated level of NLRP3 inflammasome activation (Fig. 1 J and K). Collectively, the above results indicate that RNF123 deficiency or its R854H mutation leads to more severe intestinal inflammation in the in vivo colitis model, via the NLRP3-dependent pathway. We confirm that the RNF123-R854H variant, a nonsynonymous SNP at the Chromosome 3 locus, contributes to an increased risk of Crohn’s disease and establish RNF123 as a Crohn’s disease–associated gene.

Deficiency of RNF123 Exacerbates LPS-Induced Endotoxemia and Alum-Induced Peritonitis In Vivo.

The aforementioned results suggest that RNF123 may be associated with NLRP3 inflammasome-driven inflammatory diseases. Therefore, we further investigated the function of RNF123 in LPS-induced endotoxemia, a widely used model to study sepsis and systemic inflammation. The lethality induced by LPS, especially at a sublethal dose, is mediated through the proinflammatory cytokine IL-1β, which involves the activation of the NLRP3 inflammasome (28, 29). We first examined lethality in mice treated with different concentrations of LPS. Our dose-ranging study defined a lethality window for LPS in our model. Intraperitoneal (i.p.) administration of 10 mg/kg LPS reproducibly induced partial mortality within 24 h, a response bracketed by the absence of death at 5 mg/kg and the rapid lethality observed with doses above 20 mg/kg (SI Appendix, Fig. S2A). We then challenged wild-type (WT), Rnf123^−/−, Nlrp3^−/−, and Rnf123^−/−Nlrp3^−/− (double knockout) mice with LPS via i.p. injection at concentrations of 5 mg/kg or 10 mg/kg. We observed that Rnf123^−/− mice exhibited an earlier death onset and a higher mortality rate than wild-type mice (Fig. 2A). Meanwhile, the Nlrp3^−/− and Rnf123^−/−Nlrp3^−/− mice, which were deficient in expressing NLRP3, all survived beyond 24 h after LPS injection (Fig. 2A). High levels of IL-1β could be detected in the serum of mice after i.p. injection of 10 mg/kg LPS after 6 to 12 h (SI Appendix, Fig. S2B). Then we tested secretion of proinflammatory cytokines in the serum of the aforementioned mice in these conditions. As depicted in Fig. 2B, Rnf123^−/− mice showed a significantly higher level of IL-1β compared to the wild-type mice. Meanwhile, knockout of Nlrp3 significantly decreased IL-1β secretion, and no obvious differences were observed between Nlrp3^−/− and Rnf123^−/−Nlrp3^−/− mice (Fig. 2B). Additionally, there was no difference in the serum levels of IL-6 and TNF-α among all the groups of mice (SI Appendix, Fig. S2C). Furthermore, we collected cells from the peritoneal lavage fluid of mice 6 h after i.p. injection of 10 mg/kg LPS, and detected key indicators of inflammasome activation. We observed that deficiency of RNF123 facilitated the cleavage of Caspase-1 and IL-1β and the aggregation of ASC, confirming an elevated level of NLRP3 inflammasome activation (Fig. 2 C and D). The above results indicate that the deficiency of RNF123 exacerbates LPS-induced endotoxemia via the NLRP3-dependent pathway in vivo.

Fig. 2.

Deficiency of RNF123 facilitates NLRP3 inflammation activation in vivo. (A) Survival of WT, Rnf123^−/−, Nlrp3^−/−, and Rnf123^−/−Nlrp3^−/− mice after i.p. injection with a sublethal dose of LPS (Left panel: 5 mg/kg; Right panel: 10 mg/kg) (mice: n = 8 for each group). (B) ELISA analysis of the IL-1β secretion in the serum from WT, Rnf123^−/−, Nlrp3^−/−, and Rnf123^−/− Nlrp3^−/− mice after i.p. injection of LPS (10 mg/kg) for 6 or 12 h (mice: n = 5 for each group). (C and D). Immunoblot analysis of the IL-1β and Caspase-1 cleavage (C), and SDD-AGE analysis of the ASC aggregation (D) in the cells from the peritoneal lavage fluid of mice in (B). (E) ELISA analysis of the IL-1β secretion from the peritoneal lavage fluid of Rnf123^WT and Rnf123^−/− mice after i.p. injection of Alum for 12 h (mice: n = 5 for each group). (F) Absolute numbers of PECs and percentages of neutrophils and Ly6C⁺ monocytes from Rnf123^WT and Rnf123^−/− mice after i.p. injection of Alum for 12 h (mice: n = 5 for each group). (G and H). Immunoblot analysis of the IL-1β and Caspase-1 cleavage (G), and SDD-AGE analysis of the ASC aggregation (H) in the cells from the peritoneal lavage fluid of mice in (E). (I) Body weight changes of Rnf123^WT and Rnf123^−/− mice after intranasal instillation infection with A. baumannii (5.0 × 10⁸ CFU per mouse) (mice: n = 5 for each group). (J) Bacterial burden in the lungs on Day 2 of Rnf123^WT and Rnf123^−/− mice after infection with A. baumannii (mice: n = 5 for each group). (K) ELISA analysis of the IL-1β secretion in serum from Rnf123^WT and Rnf123^−/− mice after infection with A. baumannii for 48 h (mice: n = 5 for each group). (L) Hematoxylin and eosin–stained images of lung sections of Rnf123^WT and Rnf123^−/− mice after infection with A. baumannii for 48 h. (Scale bar, 100 µm.) (M and N) Immunoblot analysis of the IL-1β and Caspase-1 cleavage (M), and SDD-AGE analysis of the ASC aggregation (N) in the lungs from mice in (L). The bacterial infection of (J–N) were same as (I). Data are represented as mean ± SD. In (A), the log-rank (Mantel-Cox) test was performed for survival curve (WT & Rnf123^−/−). In (B and I), two-way ANOVA was performed. In (E, F, J, and K), the two-tailed Student’s t test was performed. *P < 0.05, **P < 0.01, ***P < 0.001. β-actin was used as a loading control. WT, wild-type; Alum, aluminum.

Next, we carried out a model of peritonitis specifically induced by the NLRP3 inflammasome activator Alum, a widely used adjuvant that promotes leukocyte recruitment via the induction of NLRP3 inflammasome-dependent IL-1β secretion (27, 30, 31). As shown in Fig. 2E and SI Appendix, Fig. S2D, the deficiency of RNF123 in Rnf123^−/− mice enhanced the secretion of IL-1β compared to their wild-type littermates the Rnf123^WT mice upon Alum treatment, while the production of TNF-α and IL-6 was not affected. We also found that the number of Alum-induced peritoneal exudate cells (PECs) was higher in Rnf123^−/− mice compared to Rnf123^WT mice (Fig. 2F). An analysis of PECs revealed that Alum treatment led to an increase in the percentages of Ly6G-positive neutrophils and Ly6C-positive inflammatory monocytes (Fig. 2F and SI Appendix, Fig. S2E). Notably, Rnf123^−/− mice exhibited a more pronounced expansion of these populations compared to their wild-type littermates (Fig. 2F and SI Appendix, Fig. S2E). Additionally, the cells from the peritoneal lavage fluid in Rnf123^−/− mice after Alum treatment showed enhanced Caspase-1 and IL-1β cleavage and ASC aggregation compared to the Rnf123^WT mice, indicating an increased level of NLRP3 inflammasome activation (Fig. 2 G and H). These results indicate that the deficiency of RNF123 exacerbates Alum-induced peritonitis in vivo.

Deficiency of RNF123 Enhances Host Defense against Bacterial Infection.

The NLRP3 inflammasome contributes to the host defense against Acinetobacter baumannii and other gram-negative bacterial infections, while the AIM2 inflammasome is instrumental in defending against Francisella novicida infection (32–34). We intended to explore whether RNF123 participates in the host defense against bacterial infection. The Rnf123^−/− mice and their wild-type littermates Rnf123^WT mice were intranasally infected with A. baumannii and F. novicida, and body weight loss was monitored over time. As shown in Fig. 2I, Rnf123^WT mice lost more body weight than Rnf123^−/− mice after A. baumannii infection. In contrast, there was no difference in body weight between Rnf123^WT and Rnf123^−/− mice after F. novicida infection (SI Appendix, Fig. S2F). We counted bacteria in the lungs of Rnf123^WT and Rnf123^−/− mice after 2 d of infection with A. baumannii and F. novicida. The bacterial burdens in the lungs of Rnf123^−/− mice were significantly lower than those in Rnf123^WT mice with A. baumannii infection, but showed no difference upon F. novicida infection (Fig. 2J and SI Appendix, Fig. S2G). The results prove that the deficiency of RNF123 enhances the host defense against A. baumannii, rather than F. novicida infection. Next, we tested the cytokines in the serum of Rnf123^WT and Rnf123^−/− mice with A. baumannii infection. IL-1β confers protection in multiple models of bacterial infection (35). As shown in Fig. 2K, the level of IL-1β in the serum of Rnf123^−/− mice was significantly higher than that in the serum of Rnf123^WT mice upon A. baumannii infection. Instead, the production of TNF-α and IL-6 after bacterial infection was comparable between Rnf123^WT and Rnf123^−/− mice (SI Appendix, Fig. S2H). Additionally, hematoxylin and eosin staining revealed greater infiltration of immune cells and injury in the lungs of Rnf123^WT mice, relative to that in the lungs of Rnf123^−/− mice, after infection with A. baumannii (Fig. 2L and SI Appendix, Fig. S2I). RNF123 deficiency potentiated Caspase-1 and IL-1β cleavage and promoted ASC aggregation in the lungs of A. baumannii–infected mice, indicating the enhanced inflammasome activation (Fig. 2 M and N). Collectively, these results demonstrate that the deficiency of RNF123 increases host defense against bacterial infection in an NLRP3-dependent manner.

RNF123 Negatively Regulates the NLRP3 Inflammasome Activation in Cells.

Then we investigated the function and mechanism of RNF123 in regulating the NLRP3 inflammasome in cells. We prepared primary peritoneal macrophages (PMs) from Rnf123^WT and Rnf123^−/− mice, and employed a widely used two-signal model of inflammasome activation in PMs, including priming cells with LPS followed by stimulation with the NLRP3 inflammasome activators (ATP, Nig, and Alum). As shown in Fig. 3A and SI Appendix, Fig. S3A, immunoblot analysis revealed that Rnf123^−/− PMs exhibited enhanced LPS- and ATP-triggered Caspase-1 (p20) and IL-1β (p17) cleavage compared to Rnf123^WT PMs, whereas the expression of pro-Caspase-1 and pro-IL-1β was comparable between the two groups. Meanwhile, we performed SDD-AGE to detect the ASC aggregation in Rnf123^WT and Rnf123^−/− PMs with LPS and ATP treatment, and found that the deficiency of RNF123 enhanced ASC aggregation (Fig. 3B). We also performed immunofluorescence staining of ASC specks in PMs, and found that Rnf123^−/− PMs showed more ASC specks compared to Rnf123^WT PMs upon LPS and ATP treatment (Fig. 3C and SI Appendix, Fig. S3B). Moreover, we detected the expression and secretion of inflammatory cytokines by qPCR and ELISA. PMs from Rnf123^WT and Rnf123^−/− mice were primed with LPS for indicated times and then activated with ATP. We observed that the deficiency of RNF123 in the Rnf123^−/− PMs significantly promoted IL-1β secretion, whereas the secretion of TNF-α and IL-6 remained unaffected (Fig. 3D and SI Appendix, Fig. S3C). The mRNA levels of Il1b, Tnfa, and Il6 were not changed by Rnf123 knockout (Fig. 3D and SI Appendix, Fig. S3C). The pyroptotic cell death induced by inflammasome activation was also detected. PMs were primed with LPS and activated with the NLRP3 inflammasome activator ATP, Nig, or Alum, followed by detection of lactate dehydrogenase (LDH) release. As shown in Fig. 3E, LPS and NLRP3 inflammasome activators triggered cell death, while the deficiency of RNF123 in Rnf123^−/− PMs exacerbated the cell death compared to Rnf123^WT PMs. To examine the function of RNF123 in other inflammasomes, we treated PMs with double-stranded DNA (poly (dA: dT)), a trigger of AIM2 (absent in melanoma 2) inflammasome and Flagellin, a trigger of NLRC4 (NLR family CARD domain containing 4) inflammasome. As shown in Fig. 3F, compared to Rnf123^WT PMs, Rnf123^−/− PMs exhibited increased IL-1β secretion in response to all three NLRP3 inflammasome activators, while the IL-1β triggered by poly (dA: dT) and Flagellin showed no significant difference between Rnf123^WT and Rnf123^−/− PMs. The mRNA levels of inflammatory cytokines were comparable between Rnf123^WT and Rnf123^−/− PMs upon exposure to various stimuli (Fig. 3F). The above results collectively prove that the deficiency of RNF123 promotes activation of the NLRP3 inflammasome.

Fig. 3.

RNF123 negatively regulates the NLRP3 inflammasome activation in cells. (A) Immunoblot analysis of the IL-1β and Caspase-1 cleavage and secretion in the supernatants or cell lysates of Rnf123^WT and Rnf123^−/− primary PMs primed with LPS for 8 h, followed by the stimulation of ATP for 30 min. (B) SDD-AGE analysis of the ASC aggregation in Rnf123^WT and Rnf123^−/− PMs treated as (A). (C) Statistics of the ASC specks in Rnf123^WT and Rnf123^−/− PMs treated as (A). The ASC specks were counted and normalized with the group of Rnf123^WT PMs treated with LPS plus ATP. (D) ELISA and qPCR analysis of the IL-1β secretion and expression from Rnf123^WT and Rnf123^−/− PMs primed with LPS for the indicated times, followed by the stimulation of ATP for 30 min. (E) LDH release detection of Rnf123^WT and Rnf123^−/− PMs primed with LPS for the indicated times, followed by the stimulation of ATP (30 min), Nig (45 min), or Alum (6 h). Data represent as the mean percentage of LDH release relative to the total cell lysate. (F) ELISA and qPCR analysis of the IL-1β secretion and expression from Rnf123^WT and Rnf123^−/− PMs primed with LPS for 8 h, followed by the stimulation of ATP (30 min), Nig (45 min), Alum (6 h), poly (dA: dT) (1 h) or Flagellin (1 h). (G) Immunoblot analysis of the IL-1β and Caspase-1 cleavage and secretion in the supernatants or cell lysates of si-Ctrl or si-RNF123 treated THP-1 cells primed with LPS for 8 h, followed by the stimulation of ATP for 1 h. (H) qPCR analysis of the RNF123 expression and ELISA analysis of the IL-1β secretion in THP-1 cells treated as (G). (I) ELISA and qPCR analysis of the IL-1β secretion and expression in RAW-ASC cells with or without the transfection of RNF123, followed by the treatment of LPS (8 h) and ATP (30 min) or Nig (45 min). Data are represented as mean ± SD. Two-tailed Student’s t test was performed. *P < 0.05; **P < 0.01; ***P < 0.001. β-actin was used as a loading control. Ctrl, control; EV, empty vector; WT, wild-type; Nig, nigericin; Alum, aluminum.

Next, we investigated the role of RNF123 in the NLRP3 inflammasome activation in THP-1 cells. We knocked down NEK7 by siRNA in THP-1 cells, followed by LPS and ATP treatment. We observed that NEK7-knockdown markedly attenuated IL-1β secretion, suggesting that the activation of the NLRP3 inflammasome in THP-1 cells relies on NEK7 under our experimental conditions (SI Appendix, Fig. S3D). Then we knocked down RNF123 in THP-1 cells, followed by LPS and ATP treatment. We found that RNF123-knockdown in the THP-1 cells enhanced the cleavage and secretion of Caspase-1 and IL-1β, whereas it showed little effect on IL-6 and TNF-α, confirming that the deficiency of RNF123 contributes to NLRP3 inflammasome activation in human cells (Fig. 3 G and H and SI Appendix, Fig. S3 E and F). We also constructed Rnf123-knockout J774A.1 cells, and found that the deficiency of RNF123 in the J774A.1 cells promoted the cleavage and secretion of Caspase-1 and IL-1β (SI Appendix, Fig. S3 G and H), consistent with the phenomenon in PMs and THP-1 cells. Furthermore, we constructed RAW264.7 macrophage cell line ectopically expressing ASC [RAW-ASC cells (36)], and exogenously expressed RNF123 (SI Appendix, Fig. S3I). Upon treatment with LPS and ATP or Nig, the overexpression of RNF123 substantially inhibited the secretion, rather than the expression, of IL-1β (Fig. 3I). The cryopyrin-associated periodic syndrome (CAPS)-associated NLRP3 mutant R258W (mutation of arginine to tryptophan at position 258) mediates the autoactivation of the NLRP3 inflammasome (4, 37). We constructed Nlrp3-knockout RAW-ASC cells, and reconstituted them with wild-type and R258W mutant NLRP3 (SI Appendix, Fig. S3J). Then we transfected RNF123 into these cells and observed that overexpression of RNF123 inhibited the IL-1β secretion mediated by both wild-type and R258W mutant NLRP3 (SI Appendix, Fig. S3K).

In addition, we tested the function of the RNF123-R854H mutation in regulating the NLRP3 inflammasome activation in PMs. We extracted PMs from Rnf123^WT and Rnf123^R854H mice, treated them with LPS and ATP, and measured the key indicators of NLRP3 inflammasome activation. The immunoblot and SDD-AGE analysis revealed that the R854H mutation of RNF123 promoted the cleavage of Caspase-1 and IL-1β, and facilitated the aggregation of ASC (SI Appendix, Fig. S4 A and B). The qPCR and ELISA analysis showed that the RNF123-R854H mutation enhanced the secretion, but not the expression of IL-1β, while it exhibited little effect on TNF-α and IL-6 (SI Appendix, Fig. S4C). The LDH release assay showed that the RNF123-R854H mutation exacerbated the cell death triggered by LPS and ATP treatment (SI Appendix, Fig. S4D). The above results demonstrate that the RNF123-R854H mutation promotes the NLRP3 inflammasome activation.

RNF123 Catalyzes Unanchored K63-Linked Ubiquitination of NEK7.

To elucidate the mechanism by which RNF123 regulates the NLRP3 inflammasome, the interaction between RNF123 and each of the NLRP3 inflammasome components was examined. We purified Flag-tagged RNF123, NEK7, NLRP3, ASC, and Caspase-1 to near homogeneity (SI Appendix, Fig. S5), incubated the RNF123 with these proteins in vitro, and performed coimmunoprecipitation. As shown in SI Appendix, Fig. S6A, RNF123 pulled down NEK7, but not NLRP3, ASC, or Caspase-1, with anti-RNF123 antibody. Consistently, NEK7 also pulled down RNF123 with anti-NEK7 antibody. Then, we detected the interaction between RNF123 and NEK7 under physiological conditions. In J774A.1 cells and PMs, NEK7 pulled down comparable amounts of RNF123 in the resting state or upon LPS and ATP treatment (SI Appendix, Fig. S6B). Furthermore, we detected the colocalization between endogenous RNF123 and NEK7 in PMs. As shown in SI Appendix, Fig. S6C, RNF123 colocalized with NEK7 in the resting state (PBS group) or in the presence of LPS and ATP (LPS and LPS + ATP groups). RNF123 did not colocalize with NLRP3 with LPS treatment alone (SI Appendix, Fig. S6C). However, with the subsequent addition of ATP, RNF123 colocalized with both NLRP3 and ASC in inflammasome puncta, which may be attributed to the association between RNF123 and NEK7 (SI Appendix, Fig. S6C). The above results indicate that RNF123 specifically interacts and colocalizes with NEK7, suggesting that NEK7 may be a direct target for RNF123 to regulate NLRP3 inflammasome activation.

Next, we tested the capacity of RNF123 to catalyze the ubiquitination of NEK7. We prepared PMs from Rnf123^WT and Rnf123^−/− mice, and treated them with LPS and ATP. We observed that LPS and ATP could trigger the ubiquitination of NEK7, and the deficiency of RNF123 in Rnf123^−/− PMs led to a decrease in the signal of both wild-type and K63-linked ubiquitination, but not the K48-linked ubiquitination, of NEK7 (Fig. 4A). In addition, we knocked out Rnf123 in J774A.1 cells and observed reduced wild-type and K63-linked ubiquitination signal of NEK7 (SI Appendix, Fig. S7A). Consistently, siRNA-mediated knockdown of RNF123 in THP-1 cells also resulted in the reduction of the NEK7 K63-linked ubiquitination (SI Appendix, Fig. S7B). The above results collectively demonstrate that RNF123 regulates the endogenous K63-linked ubiquitination of NEK7 under physiological conditions with the treatment of inflammatory stimuli.

Fig. 4.

NEK7 is modified with unanchored K63-linked polyubiquitin chains. (A) Immunoprecipitation analysis of the ubiquitination of endogenous NEK7 in wild-type (Rnf123^WT) or Rnf123-knockout (Rnf123^−/−) primary PMs with LPS (8 h) and ATP (30 min) treatment. (B) Immunoprecipitation analysis of the ubiquitination of exogenously expressed NEK7 from HEK293T cells cotransfected with WT-, K63-, K48-, or K63R-Ub in the presence or absence of RNF123. (C) Immunoprecipitation analysis of the ubiquitination of exogenously expressed NEK7 and NLRP3 from HEK293T cells cotransfected with WT- or K63-Ub in the presence or absence of RNF123. (D) Immunoprecipitation analysis of the ubiquitination of exogenously expressed NEK7 from HEK293T cells cotransfected with K63-Ub in the presence or absence of RNF123 or its mutants. (E) Immunoprecipitation analysis of the ubiquitination of exogenously expressed NEK7 from HEK293T cells cotransfected with K63-Ub and RNF123, followed by the treatment of IsoT or vOTU. (F) Immunoprecipitation analysis of the ubiquitination of exogenously expressed Myc-tagged NEK7 from HEK293T cells in the in vitro ubiquitination assay. After transfection, NEK7 was immunoprecipitated from cell lysate and reacted with purified E1, Ube2N(2V1), K63-Ub, and RNF123, followed by the IsoT or vOTU treatment. (G) Putative interaction between NEK7 (PDB 6S76) and K63-Ub2 (PDB 3H7P) using Haddock and the Rosetta software suite. Ubiquitin is shown in magenta and NEK7 is shown in blue. (H and I). Immunoprecipitation analysis of the interaction between purified NEK7, NEK7 mutants and K63-Ub4 (H) or K63-polyUb (I). β-actin was used as a loading control. WT, wild-type.

To further investigate the mechanism of RNF123-mediated NEK7 ubiquitination, we transfected RNF123-Flag with NEK7-Myc and HA-tagged ubiquitin mutants K48-Ub, K63-Ub, and K63R-Ub into HEK293T cells, followed by immunoprecipitation and immunoblot analysis. The K48 and K63 mutants (K48/K63-Ub) are characterized by the replacement of all lysine residues, except for the lysine at position 48 or position 63, with arginine residues. The mutant K63R-Ub harbors a single lysine-to-arginine substitution at position 63. The results showed that RNF123 elevated the ubiquitination signal of NEK7 with the transfection of K63-Ub rather than K48-Ub or K63R-Ub, demonstrating that K63-linked ubiquitination is the major type of ubiquitin modification targeted by RNF123 on NEK7 (Fig. 4B). Further, we compared the catalytic activity of RNF123 on NEK7 and NLRP3 in the overexpression system. As shown in Fig. 4C, RNF123 augmented both the wild-type and K63-linked ubiquitination of NEK7, but not of NLRP3, indicating that NEK7 is a specific substrate for RNF123. Next, we examined whether the impact of RNF123 on NEK7 ubiquitination relies on its catalytic activity. RNF123 contains a RING-type domain (amino acids 1254-1292), which is crucial for its catalytic activity (21). Additionally, we observed two typical Zn²⁺ binding regions in the RING-type domain of RNF123 (PDB 2MA6), which are surrounded by two groups of cysteines, respectively (SI Appendix, Fig. S7C). We constructed a RNF123 truncation ΔRING, in which the RING-type domain of RNF123 was deleted, and constructed two mutants 3CA (C1269A C1288A C1291A) and 4CA (C1254A C1257A C1274A C1277A), in which the cysteines surrounding the Zn²⁺ were mutated to alanine. Then we transfected NEK7-Myc and K63-Ub-HA, together with Flag-tagged wild-type RNF123, truncation RNF123-ΔRING, mutant 3CA and 4CA, into HEK293T cells, followed by the immunoprecipitation of NEK7. The results showed that RNF123 potently enhanced the signal of K63-linked ubiquitination of NEK7, whereas ΔRING, 3CA, and 4CA exhibited a marginal effect, proving that the catalytic activity of RNF123 is essential for its ability to ubiquitinate NEK7 (Fig. 4D). To test the linkage of RNF123-mediated K63-linked ubiquitination on NEK7, we immunoprecipitated exogenously expressed NEK7-Myc from HEK293T cells cotransfected with RNF123-Flag and K63-Ub-HA, followed by the treatment with IsoT and vOTU. The vOTU destroys polyubiquitin chains covalently linked to a target protein as well as unanchored polyubiquitin chains (38, 39). IsoT only cleaves the unanchored polyubiquitin chains (39, 40). As shown in Fig. 4E, both IsoT and vOTU reduced the ubiquitination signal of NEK7 catalyzed by RNF123, indicating that RNF123 catalyzes the unanchored K63-linked ubiquitination of NEK7. To confirm the function of RNF123 on NEK7 ubiquitination, we carried out an in vitro ubiquitination assay with purified E1, E2, RNF123-Flag, and recombinant Myc-tagged NEK7 immunoprecipitated from HEK293T cells. Ube2N has been reported as a crucial E2 ubiquitin-conjugating enzyme for the formation of unanchored K63-linked polyubiquitin chains (39, 41). The knockout of UBE2N expression significantly abolished the signal of the K63-linked ubiquitination of NEK7 in HEK293T cells, suggesting that Ube2N is the major E2 enzyme responsible for K63-linked ubiquitination of NEK7 (SI Appendix, Fig. S7D). Therefore, we applied Ube2N as the E2 enzyme to cooperate with purified RNF123-Flag in the in vitro ubiquitination assay. As shown in Fig. 4F, NEK7 was modified with K63-linked polyubiquitin chains catalyzed by Ube2N-RNF123, which could be completely removed by IsoT and vOTU, indicating that these polyubiquitin chains noncovalently bind to NEK7. RNF123 is also known as KPC1, which is demonstrated to consist of the Kip1 ubiquitination-promoting complex (KPC) heterodimer with KPC2 (UBAC1) (20, 21). We carried out the in vitro ubiquitination assay of Ube2N-RNF123 with the additional inclusion of purified KPC2 (SI Appendix, Fig. S5). As shown in SI Appendix, Fig. S7E, the addition of KPC2 exhibited no effect on the signal of the NEK7 ubiquitination, indicating that KPC2 is not directly involved in the RNF123-mediated NEK7 ubiquitination. Collectively, the above results indicate that RNF123 catalyzes the unanchored K63-linked ubiquitination of NEK7.

NEK7 Directly Captures K63-Linked Polyubiquitin Chains In Vitro.

Previous studies indicated that unanchored polyubiquitin chains can directly bind to target proteins and exert their function in vitro (42, 43). Therefore, we wondered whether NEK7 owns the capacity to directly capture unanchored K63-linked polyubiquitin chains. We measured the interaction between polyubiquitin chains and each of the inflammasome components. We incubated purified Flag-tagged NLRP3, NEK7, and ASC with K48-linked polyubiquitin chains (K48-polyUb), K63-linked polyubiquitin chains (K63-polyUb), and monomeric ubiquitin (Mono-Ub), respectively, and pulled down the inflammasome components. The coimmunoprecipitation analysis showed that NEK7 bound more K63-polyUb, but not K48-polyUb or Mono-Ub, than NLRP3 and ASC (SI Appendix, Fig. S7F). Then we performed the coimmunoprecipitation of the purified NEK7-Flag and NLRP3-Flag with K63-linked polyubiquitin chains containing 4 ubiquitin moieties (K63-Ub4). As shown in SI Appendix, Fig. S7G, NEK7 pulled down a significant amount of K63-Ub4, and the addition of NLRP3 could not weaken the association between NEK7 and K63-Ub4. In contrast, although NLRP3 could also pull down K63-Ub4, the addition of NEK7 competed with NLRP3 for binding to K63-Ub4 and reduced the amount of K63-Ub4 attached to NLRP3. The results demonstrate that NEK7 has a more potent capacity for binding K63-linked polyubiquitin chains than NLRP3. To determine the structural association between NEK7 and K63-linked polyubiquitin chains, we performed molecular docking analysis between NEK7 (PDB 6S76) and K63-Ub2 (PDB 3H7P) to infer a plausible location of the NEK7/K63-linked polyubiquitin chain. The docking result showed that four regions on NEK7 mediated its interaction with K63-Ub2 (Fig. 4G and SI Appendix, Fig. S7H). Among these, Region 1 and 2 are located on the N-terminal lobe of NEK7, and Region 3 and 4 are located on the C-terminal lobe of NEK7 (Fig. 4G). Especially, Region 2, the α-helix from D72-L89 (aspartic acid, D; leucine, L), also mediated the interaction between NEK7 and Mono-Ub (SI Appendix, Fig. S7H). We mutated the amino acids of NEK7 on Region 1-4 to alanine (referred to as mutR1-4, see SI Appendix, Fig. S7H), purified the Flag-tagged NEK7 mutants (SI Appendix, Fig. S5), and detected the interaction between NEK7-mutR1-R4 and K63-linked polyubiquitin chains through coimmunoprecipitation. Notably, NEK7-mutR2 and -mutR3 pulled down a lesser amount of K63-Ub4 and K63-polyUb compared to wild-type NEK7, NEK7-mutR1, and -mutR4 (Fig. 4 H and I). The results indicate that amino acids on NEK7 Region 2 and 3 are important for mediating the association between NEK7 and K63-linked polyubiquitin chains. Additionally, the NEK7 mutation K63M-K64M (mutation of lysine to methionine at position 63 and 64), which abolishes the catalytic activity of NEK7 (4), did not impair the interaction between NEK7 and K63-Ub4 or K63-polyUb, suggesting that NEK7 binds K63-linked polyubiquitin chains independent of its catalytic function (Fig. 4 H and I). We also conducted Surface Plasmon Resonance (SPR) assay to measure the equilibrium dissociation constant (K_D) of NEK7 and its mutants with K63-Ub4. The results showed that NEK7 bound to K63-Ub4 with a K_D value of 0.5766 nM (SI Appendix, Fig. S7I). Consistent with the coimmunoprecipitation data, both NEK7-mutR2 and -mutR3 showed higher K_D values with K63-Ub4 compared to wild-type NEK7, corroborating their reduced affinity with K63-linked polyubiquitin chains (SI Appendix, Fig. S7I). NLRP3 also showed relatively weaker affinity with K63-Ub4 compared to NEK7 (SI Appendix, Fig. S7I). Collectively, the above results demonstrate that K63-linked polyubiquitin chains can be specifically captured by NEK7.

RNF123-Mediated Unanchored K63-Linked Ubiquitination of NEK7 Inhibits NEK7-Licensed NLRP3 Inflammasome Assembly and Activation.

Then we investigated the effect of RNF123-mediated NEK7 unanchored K63-linked ubiquitination on NLRP3 inflammasome assembly and activation. We performed the in vitro ubiquitination assay with purified E1, Ube2N, Flag-tagged RNF123 and its mutant 3CA (SI Appendix, Fig. S5), recombinant Myc-tagged NEK7 and mutR2 immunoprecipitated from HEK293T cells. As shown in Fig. 5A, NEK7 was obviously modified with K63-linked polyubiquitin chains catalyzed by Ube2N-RNF123. The RNF123 catalytic-dead mutant 3CA could not mediate the ubiquitination of NEK7 in this in vitro ubiquitination assay, validating that RNF123 mediates the NEK7 ubiquitination depending on its catalytic activity (Fig. 5A). RNF123 catalyzed the ubiquitination of NEK7-mutR2 only to a minimal extent, confirming that the amino acids on NEK7 Region 2 are also critical for RNF123-mediated NEK7 ubiquitination (Fig. 5A). Further, based on the above in vitro ubiquitination assay, we tested the effect of RNF123 on NEK7-mediated NLRP3 inflammasome assembly and activation. The coimmunoprecipitation analysis revealed that the above NEK7 ubiquitinated by RNF123 showed weaker association with purified NLRP3-Flag (Fig. 5B). NEK7 dissociates NLRP3 cage-like aggregates and relieves the shielding of NLRP3-PYD (7–10). The dissociation of NLRP3 aggregates by NEK7 leads to the appearance of smaller NLRP3 species, which can be detected by centrifugation (7). We purified recombinant His-tagged NEK7 and NEK7-mutR2, and employed them into the in vitro ubiquitination assay (SI Appendix, Fig. S5). The ubiquitinated NEK7-His was then eluted, incubated with purified NLRP3-Flag, and analyzed by sucrose gradient ultracentrifugation. The experiment showed that the addition of NEK7 led to a decrease in the peak for the NLRP3 aggregates and induced the appearance of smaller NLRP3 species, whereas the ubiquitination of NEK7 mediated by RNF123 impeded the NEK7-mediated dissociation of NLRP3 aggregates (Fig. 5C). Further, we incubated the recombinant proteins Flag-tagged NLRP3, ASC with His-tagged NEK7 or NEK7-mutR2 after the in vitro ubiquitination assay, and measured the ASC aggregation by SDD-AGE. The results showed that the ubiquitination of NEK7 mediated by RNF123 reduced the NEK7-mediated ASC aggregation (Fig. 5D). The above results demonstrate that the RNF123-catalyzed ubiquitination of NEK7 prevents the NEK7-mediated NLRP3 inflammasome assembly and activation. However, compared to the wild-type NEK7, RNF123 showed a much weaker inhibitory effect on the NEK7-mutR2-mediated NLRP3 inflammasome assembly and activation (Fig. 5 B–D). Specifically, RNF123 was less effective at inhibiting the interaction between NEK7-mutR2 and NLRP3, the NEK7-mutR2-mediated dissociation of NLRP3 aggregates, and the aggregation of ASC (Fig. 5 B–D). Meanwhile, RNF123 mutant 3CA almost abolished the ability to inhibit the NLRP3 inflammasome assembly and activation (Fig. 5 B–D). These results confirm that the effect of RNF123 on NEK7-mediated NLRP3 inflammasome assembly and activation is dependent on its function for catalyzing NEK7 ubiquitination. To verify the on-target regulation of RNF123 on NEK7, we established an NLRP3 inflammasome activation system in HEK293T cells. We knocked out NEK7 in HEK293T cells and reconstituted them with Flag-tagged wild-type or mutR2 NEK7 (SI Appendix, Fig. S8A). Both types of NEK7 could mediate Caspase-1 cleavage after Nig treatment with the coexpression of NLRP3, ASC, and Caspase-1 (SI Appendix, Fig. S8B). Then we transfected RNF123-Flag and K63-Ub-HA into this system, and found that overexpression of RNF123 inhibited NEK7-mediated Caspase-1 cleavage, but showed little effect on NEK7-mutR2-mediated Caspase-1 cleavage (SI Appendix, Fig. S8C). The results demonstrate that RNF123 regulates NLRP3 inflammasome activation by specifically targeting NEK7. Collectively, the above results indicate that RNF123-mediated unanchored K63-linked ubiquitination of NEK7 inhibits NEK7-licensed NLRP3 inflammasome assembly and activation.

Fig. 5.

RNF123-mediated unanchored K63-linked ubiquitination of NEK7 inhibits NEK7-licensed NLRP3 inflammasome assembly and activation. (A) Immunoprecipitation analysis of the ubiquitination of exogenously expressed Myc-tagged NEK7 and NEK7-mutR2 from HEK293T cells in the in vitro ubiquitination assay mediated by purified RNF123 or RNF123-3CA. (B) Immunoprecipitation analysis of the interaction between purified NLRP3 and exogenously expressed Myc-tagged NEK7 and NEK7-mutR2 from HEK293T cells after the in vitro ubiquitination assay as (A). (C) Sucrose gradient ultracentrifugation of NLRP3 aggregates with the addition of purified NEK7 or NEK7-mutR2 after the in vitro ubiquitination assay as (A). Aliquots of the fractions were immunoblotted with anti-NLRP3 antibody, and corresponding quantification was performed and shown as Right panel. (D) SDD-AGE analysis of the ASC aggregation induced by purified NLRP3 and NEK7 or NEK7-mutR2 after the in vitro ubiquitination assay as (A).

We then investigated the effect of RNF123-R854H mutation on NEK7 ubiquitination and NEK7-licensed NLRP3 inflammasome assembly. We transfected HEK293T cells with NEK7-Myc, K63-Ub-HA, and Flag-tagged wild-type or mutant RNF123, including the catalytic-dead mutants ΔRING and 3CA, and several nonsynonymous SNP mutants P51R, R387Q, K596E, and R854H. Then we immunoprecipitated NEK7 and detected its ubiquitination through immunoblot analysis. As shown in SI Appendix, Fig. S9A, wild-type RNF123 and SNP mutants P51R, R387Q, K596E enhanced the K63-linked ubiquitination of NEK7, whereas RNF123-R854H and the catalytic-dead mutants exhibited a marginal effect on the NEK7 ubiquitination. Consistently, the in vitro ubiquitination assay with purified Flag-tagged wild-type RNF123 and its mutants showed that RNF123-R854H could not strengthen the K63-linked ubiquitination of NEK7 (SI Appendix, Figs. S5 and S9B). The above results indicate that the R854H mutation reduces the capacity of RNF123 for catalyzing the ubiquitination of NEK7. To elucidate the mechanism by which R854H mutation diminishes RNF123’s catalytic activity for NEK7, we first performed an in vitro ubiquitination assay with purified E1, Ube2N, RNF123-Flag, and RNF123-R854H-Flag in the absence of the substrate NEK7. We observed that RNF123 accelerated the formation of polyubiquitin chains, and the R854H mutation had no significant effect on this process (SI Appendix, Fig. S9C). This result indicates that the R854H mutation does not affect the catalytic function of RNF123. We next examined the interactions of both wild-type RNF123 and its R854H mutant with NEK7 as well as with Ube2N. We expressed Flag-tagged RNF123 or its mutant R854H with NEK7-Myc in HEK293T cells, and performed coimmunoprecipitation with anti-Flag antibody. We found that the R854H mutation reduced the amount of NEK7 immunoprecipitated by RNF123 (SI Appendix, Fig. S9D). However, the R854H mutation did not affect the binding between RNF123 and Ube2N in this overexpression system (SI Appendix, Fig. S9E). Additionally, we prepared PMs from Rnf123^WT and Rnf123^R854H mice, and tested the interaction of wild-type or R854H mutated RNF123 with NEK7 and Ube2N under physiological conditions. We found that RNF123-R854H pulled down less NEK7 compared to the wild-type RNF123, regardless of LPS and ATP treatment (SI Appendix, Fig. S9F). In contrast, this mutation had little effect on the interaction between RNF123 and Ube2N (SI Appendix, Fig. S9F). The above results indicate that the R854H mutation on RNF123 impairs the interaction between RNF123 and its substrate NEK7, thereby reducing its catalytic activity for NEK7. Next, we tested the effect of RNF123-R854H on NEK7-mediated NLRP3 inflammasome assembly and activation based on the above in vitro ubiquitination assay with purified proteins as Fig. 5 B–D. As shown in SI Appendix, Fig. S9 G–I, RNF123-mediated ubiquitination of NEK7 inhibited the association between NEK7 and NLRP3, impeded the NEK7-mediated dissociation of NLRP3 aggregates, and reduced the NEK7-mediated ASC aggregation. In contrast, the RNF123-R854H mutant exhibited a marginal effect on NEK7-mediated NLRP3 inflammasome assembly in vitro (SI Appendix, Fig. S9 G–I). These results prove that the R854H mutation diminishes the function of RNF123 to inhibit NEK7-licensed NLRP3 inflammasome assembly and activation, which accounts for the phenomenon that the R854H mutation of RNF123 enhances the NLRP3 inflammasome activation and exacerbates experimental colitis.

K63-Linked Polyubiquitin Chains Directly Prevent the NEK7-Licensed NLRP3 Inflammasome Assembly and Activation In Vitro.

Then we investigated the direct effect of K63-linked polyubiquitin chains on the NLRP3 inflammasome assembly and activation in vitro. We performed coimmunoprecipitation analysis with purified Flag-tagged NEK7, NEK7-mutR2, and NLRP3 in the presence or absence of K63-polyUb or K63-Ub4. As shown in SI Appendix, S10 A and B, K63-Ub4 and K63-polyUb impaired the association between NEK7 and NLRP3, but had a minor effect on the binding between NEK7-mutR2 and NLRP3. The results demonstrate that K63-linked polyubiquitin chains captured by NEK7, rather than NLRP3, impede the interaction between NEK7 and NLRP3. Then we explored the mechanism by which K63-linked polyubiquitin chains on NEK7 impede the interaction between NEK7 and NLRP3. Although Region 2 of NEK7 is not directly involved in mediating NLRP3–NEK7 binding (10), we speculated that polyubiquitin chains binding to this region may create steric hindrance for the NLRP3–NEK7 interaction. We performed structural alignment of the NEK7-K63-Ub2 docking model (from Fig. 4G) with the NEK7–NLRP3 complex (PDB 6NPY), and observed that polyubiquitin chains bound to NEK7 indeed sterically clash with the NLRP3-LRR domain (SI Appendix, Fig. S10C). We removed the segment (Leu850-Trp1036) in the NLRP3-LRR domain, which caused steric hindrance with polyubiquitin chains, to obtain the NLRP3-truncated LRR variant. Using purified Flag-tagged NEK7, NLRP3, and NLRP3-truncated LRR (SI Appendix, Fig. S5), we performed coimmunoprecipitation assays in the presence or absence of K63-polyUb or K63-Ub4. The results showed that both K63-polyUb and K63-Ub4 inhibited the interaction between NEK7 and full-length NLRP3, but had a minimal effect on NEK7 binding to the NLRP3-truncated LRR (SI Appendix, Fig. S10 D and E). These results demonstrate that polyubiquitin chains binding to Region 2 of NEK7 sterically hinder the NLRP3–NEK7 interaction.

Further, we tested the dissociation of NLRP3 aggregates by NEK7 and NEK7-mutR2 with or without K63-linked polyubiquitin chains. We incubated purified Flag-tagged NLRP3 with NEK7 or NEK7-mutR2, and simultaneously added K63-polyUb or K63-Ub4. Then the sucrose gradient ultracentrifugation was performed and showed that K63-Ub4 and K63-polyUb prevented the NLRP3 aggregate dissociation mediated by NEK7, but not by NEK7-mutR2 (SI Appendix, Fig. S10F). Additionally, we also measured the ASC aggregation by SDD-AGE in this in vitro assay and observed that K63-Ub4 and K63-polyUb inhibited the aggregation of ASC mediated by NEK7, but had a minimal impact on NEK7-mutR2-mediated ASC aggregation (SI Appendix, Fig. S10G). Negative-stain electron microscopy images of purified NLRP3-Flag showed particles as a circular and cage-like shape with a diameter of 20 to 30 nm (SI Appendix, Fig. S10 H and I). Incubation of NLRP3 with excess purified NEK7-Flag destroyed the cage-like form of NLRP3 and resulted in an irregular shape (SI Appendix, Fig. S10 H and I). Addition of K63-polyUb, rather than K48-polyUb or Mono-Ub, to the mixture of NLRP3 and NEK7 restored the regular shape of NLRP3 (SI Appendix, Figs. S10 H and I). In summary, the above results prove that NEK7-captured K63-linked polyubiquitin chains can directly inhibit NLRP3 inflammasome assembly and activation.

To investigate the physiological role of endogenous NEK7-captured unanchored K63-linked polyubiquitin chains, we immunoprecipitated NEK7 from PMs treated with inflammatory stimuli, and detected its ubiquitination by immunoblot analysis. As shown in SI Appendix, Fig. S11A, NEK7 underwent K63-linked ubiquitination upon stimulation with LPS. NEK7 remained significantly ubiquitinated following the addition of ATP, but the ubiquitination signal was slightly reduced compared to LPS treatment alone (SI Appendix, Fig. S11A). Then we applied reimmunoprecipitation (Re-IP) and deubiquitinating enzyme digestion experiments to investigate the linkage type of K63-linked ubiquitination on endogenous NEK7. NEK7 was immunoprecipitated from PMs after LPS and ATP treatment, and then subjected to heat denaturation to disrupt noncovalent interactions. Immunoblot analysis showed that the K63-linked ubiquitination signal on NEK7 was reduced after heat denaturation (SI Appendix, Fig. S11B). The deubiquitinating enzyme AMSH displays strong preference for cleaving K63-linked polyubiquitin chains (44, 45). We found that the K63-linked polyubiquitin chains on endogenous NEK7 formed upon LPS and ATP treatment could be removed by IsoT and AMSH treatment (SI Appendix, Fig. S11C). These results indicate that endogenous NEK7 is modified with unanchored K63-linked polyubiquitin chains after treatment with inflammatory stimuli. We also validated the phenomenon in J774A.1 cells (SI Appendix, Fig. S11 D–F). Then, we attempted to release the NEK7-captured unanchored K63-linked polyubiquitin chains triggered by inflammatory stimuli, and examined whether these polyubiquitin chains can directly inhibit NEK7-licensed NLRP3 inflammasome activation. We transfected RAW-ASC cells with Flag-tagged NEK7. Upon LPS and ATP treatment, we immunoprecipitated the NEK7 from the cell lysate and found that NEK7 was modified by K63-linked polyubiquitin chains (SI Appendix, Fig. S11G). It has been reported that polyubiquitin chains remain relatively stable at high temperatures (39). Then we prepared unanchored K63-linked polyubiquitin chains from NEK7 as the protocol described in SI Appendix, Fig. S11H, under conditions that denature NEK7 but preserve the structure and function of polyubiquitin chains. Immunoblot analysis of the supernatant with the anti-ubiquitin antibody clearly showed that unanchored K63-linked polyubiquitin chains could be separated from NEK7 after LPS and ATP treatment (supernatant 1-3, S1-3), and that the molecular weight of the main bands was greater than 25 kDa, indicating that most of these unanchored polyubiquitin chains contain more than 4 ubiquitin moieties (SI Appendix, Fig. S11I). The ubiquitin bands could be diminished by pretreatment with IsoT (supernatant 3, S3), further proving that these polyubiquitin chains were linked in a noncovalent manner (SI Appendix, Fig. S11I). Then we applied the separated supernatant to the in vitro assay to test its impact on NEK7-mediated NLRP3 inflammasome activation with purified proteins. The supernatant 2 (S2), which contained a large amount of unanchored K63-linked polyubiquitin chains released from NEK7, impeded the binding between NEK7 and NLRP3, NEK7-mediated dissociation of NLRP3 aggregates and ASC aggregation (SI Appendix, Fig. S11 J–L). The IsoT-treated supernatant 3 (S3), which contained a negligible amount of K63-linked polyubiquitin chains, did not show inhibitory effect on the phenomenon of NLRP3 inflammasome activation (SI Appendix, Fig. S11 J–L). The results provide additional evidence that the unanchored K63-linked polyubiquitin chains attach to NEK7 under physiological conditions upon treatment with inflammatory stimuli, and directly prevent the NEK7-licensed NLRP3 inflammasome assembly and activation.

Discussion

In this study, we investigated the role of RNF123 and unanchored K63-linked ubiquitination of NEK7 in governing the NLRP3 inflammasome and inflammatory diseases. First, we observed that mice deficient in RNF123 or harboring the RNF123-R854H mutation display more severe intestinal inflammation in the in vivo colitis model than their wild-type littermates, via an NLRP3-dependent pathway. This provides experimental evidence that the SNP RNF123-R854H variant contributes to an increased risk of human Crohn’s disease. Second, we found that the deficiency of RNF123 in vivo exacerbates LPS-induced endotoxemia and Alum-induced peritonitis, and enhances host defense against bacterial infection in an NLRP3-dependent pathway. Also, RNF123 is proved to negatively regulate the NLRP3 inflammasome in cells. These findings unveil the regulatory function of RNF123 in the NLRP3 inflammasome and confirm its pathogenic association with NLRP3 inflammasome-driven inflammatory diseases. Third, we elucidate the underlying mechanism by which RNF123 catalyzes the unanchored K63-linked ubiquitination of NEK7, thereby preventing the NEK7–NLRP3 interaction and hindering the subsequent NEK7-mediated NLRP3 inflammasome assembly and activation. We uncover a noncanonical posttranslational modification that regulates the NLRP3 inflammasome. Finally, utilizing an in vitro experimental assay, we demonstrate that K63-linked polyubiquitin chains can be specifically and directly captured by NEK7 in vitro, and inhibit the NEK7-licensed NLRP3 inflammasome assembly and activation. This points to a yet-unknown mechanism that NEK7-captured unanchored K63-linked polyubiquitin chains serve as a molecular brake, curbing excessive NLRP3 inflammasome activation and maintaining immune homeostasis (SI Appendix, Fig. S12).

RNF123, also known as KPC1, forms the KPC with its partner KPC2 and regulates multiple processes in cells (20–24). To date, no studies have reported the involvement of RNF123 in the assembly and activation of the NLRP3 inflammasome. Here, we found that RNF123 catalyzes the unanchored K63-linked ubiquitination on NEK7 and negatively regulates NEK7-licensed NLRP3 inflammasome assembly and activation in vitro and in vivo. We observed that RNF123 consistently interacts with NEK7 throughout the process of inflammasome activation (SI Appendix, Fig. S6B). However, the K63-linked ubiquitination of NEK7 is induced by LPS treatment and slightly diminishes to some extent upon the extra addition of ATP in PMs and J774A.1 cells (SI Appendix, Fig. S11 A and D). This dynamic change suggests that some deubiquitinating enzyme or other regulator, besides RNF123, may also be involved in the regulation of NEK7 ubiquitination. Previous studies have reported that NEK7 can also directly bind to human NLRP3 (8). Additionally, in human cells, NLRP3 activation remains dependent on NEK7 after the short-term stimulation with LPS and Nig (for example, 1 h), while the long-term stimulation results in functional redundancy between NEK7 and IKKβ (46). Crucially, our experimental conditions (ATP for 1 h) precisely fall within this established stimulation window where NLRP3 activation is strictly NEK7-dependent (SI Appendix, Fig. S3D). Therefore, our demonstration of RNF123’s efficacy in human cell lines under these conditions confidently positions RNF123 as a key regulator of the canonical NEK7-dependent NLRP3 inflammasome pathway in human cells. The exploration of RNF123’s function in the context of the noncanonical, IKKβ-dependent signaling emerges as a logical and important frontier for future research. RNF123-R854H has been identified as a candidate nonsynonymous SNP at the chromosome 3 locus associated with an increased risk of Crohn’s disease (19, 47). Our findings provide experimental evidence and mechanistic explanation for the association between the RNF123-R854H variant and Crohn’s disease (SI Appendix, Fig. S4 and S9), and put forward that the SNP RNF123-R854H can be regarded as a risk factor for the occurrence and exacerbation of Crohn’s disease. Collectively, our findings identify RNF123 as an E3 ligase for NEK7 and a regulator for the NLRP3 inflammasome, highlighting its implication in inflammatory diseases.

Ubiquitination is a well-recognized regulator of innate immunity (48), yet the direct modulation of the NEK7–NLRP3 interaction by NEK7 ubiquitination remains unknown. Here, we found that unanchored K63-linked polyubiquitin chains specifically modify NEK7 both in vitro and under physiological conditions, and directly inhibit the association between NEK7 and NLRP3, ultimately leading to the reduction of NLRP3 inflammasome activation. The binding affinity between NEK7 and K63-linked polyubiquitin chains is strong in vitro (SI Appendix, Fig. S7I). However, the capture of unanchored K63-linked polyubiquitin chains by NEK7 in cells still relies on RNF123 (Fig. 4 A–E). We analyzed that the main reasons are as follows: First, high affinity in vitro does not necessarily translate to efficient binding in cells, as in vitro experiments only provide the maximal affinity under optimal conditions. Intracellular conditions may not be conducive to binding due to factors such as pH and the presence of other proteins. Second, we speculated that RNF123 not only directs polyubiquitin chains to NEK7, but also activates Ube2N to promote the production of unanchored polyubiquitin chains (SI Appendix, Fig. 9C), both of which collectively lead to the increased unanchored ubiquitination of NEK7. Additionally, we observed that there exists a small portion of anchored K63-linked polyubiquitin chains resistant to IsoT cleavage (SI Appendix, Fig. 11 C and E). The functional implications of this residual covalent K63-linked ubiquitination on NEK7 are intriguing and warrant further investigation. Nevertheless, our findings uncover an additional type of posttranslational modification in the NLRP3 inflammasome and establish NEK7-loaded unanchored K63-linked polyubiquitin chains as a key determinant for the association between NEK7 and NLRP3.

Materials and Methods

All materials used in this study are listed in SI Appendix. The detailed experimental methods, encompassing biochemical and cellular assays, in vivo animal models, structural analyses, and imaging, are fully described therein. Finally, the procedures for quantification and statistical analysis are provided in SI Appendix.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Study data are included in the article and/or SI Appendix.