NLRP3 mediates osteolysis through inflammation-dependent and -independent mechanisms

Abstract

Activating-mutations in NOD-like receptor (NLR) family, pyrin domain-containing 3 (NLRP3) cause neonatal-onset multisystem inflammatory disease. However, the ontogeny of skeletal anomalies in this disorder is poorly understood. Mice globally expressing the D301N mutation in Nlrp3 (D303N in human) model the human phenotype, including systemic inflammation and skeletal deformities. To gain insights into the skeletal manifestations, we generated mice in which the expression of D301N Nlrp3 (Nlrp3^D301N) is restricted to myeloid cells. These mice exhibit systemic inflammation and severe osteopenia (∼60% lower bone mass) similar to mice globally expressing the knock-in mutation, consistent with the paradigm of innate immune-driven cryopyrinopathies. Because systemic inflammation may indirectly affect bone homeostasis, we engineered mice in which Nlrp3^D301N is expressed specifically in osteoclasts, the cells that resorb bone. These mice also develop ∼50% lower bone mass due to increased osteolysis, but there is no systemic inflammation and no change in osteoclast number. Mechanistically, aside from its role in IL-1β maturation, Nlrp3^D301N expression enhances osteoclast bone resorbing ability through reorganization of actin cytoskeleton while promoting the degradation of poly(ADP-ribose) polymerase 1, an inhibitor of osteoclastogenesis. Thus, NLRP3 inflammasome activation is not restricted to the production of proinflammatory mediators but also leads to cytokine-autonomous responses.—Qu, C., Bonar, S. L., Hickman-Brecks, C. L., Abu-Amer, S., McGeough, M. D., Peña, C. A., Broderick, L., Yang, C., Grimston, S., K., Kading, J., Abu-Amer, Y., Novack, D. V., Hoffman, H. M., Civitelli, R., Mbalaviele, G. NLRP3 mediates osteolysis through inflammation-dependent and -independent mechanisms.

Cell stress promotes NLRP3 association with the scaffold protein, apoptosis-associated speck-like protein containing a caspase activation and recruitment domain (ASC) and caspase-1. This protein complex referred to as the NLRP3 inflammasome is responsible for the maturation of IL-1β and IL-18 (1). This inflammasome is also activated in a constitutive manner by NLRP3 activating-mutations, which cause stimulus-independent assembly of the complex (2). This uncontrolled activation pattern is ultimately harmful to health as it causes autoinflammatory disorders known as cryopyrinopathies, the most severe of which is neonatal-onset multisystem inflammatory disease (NOMID). Besides the common symptoms of cryopyrinopathies, which include excessive serum levels of IL-1β and IL-18, unique and prominent features of NOMID include central nervous system and skeletal anomalies (3). Structural damage to the skeleton includes low bone mass and a risk factor for skeletal fractures, leg deformities, and length discrepancies where effective therapy is limited to invasive surgical interventions. Despite these deleterious clinical outcomes, the pathogenesis of skeletal abnormalities in NOMID patients is poorly understood, but a defect of skeletal remodeling is most likely the culprit.

The function of the NLRP3 inflammasome is more complex than a simple cytokine processing platform. This premise is supported by an accumulating body of evidence, including the observations that skeletal lesions in NOMID patients continue to grow even under treatment with IL-1β biologics such as anakinra and canakinumab, whereas other symptoms related to systemic inflammation resolve (4, 5). These findings suggest alternative, cytokine-autonomous actions of the NLRP3 inflammasome in disease. Indeed, IL-1-independent age-related responses mediated by the NLRP3 inflammasome is linked to chronic diseases in mice, including bone loss (6). Recently, a role for this inflammasome in apoptosis and pyroptosis through caspase-8 and/or caspase-1 has been proposed (7, 8), and NLRP3 inflammasome-dependent pyroptosis may be responsible for the persistent inflammation in mice with impaired IL-1β and IL-18 signaling (9). Interactions of the NLRP3 inflammasome with other networks such as the cytoskeleton and autophagy have also been reported (10). Finally, activation of this inflammasome has been shown to initiate caspase cascades, leading to proteolytic cleavage of poly(ADP-ribose) polymerase 1 (PARP1) also known as ADP-ribosyltransferase Diphtheria toxin-like 1 (ARTD1) (11, 12), a multifunctional protein and presumptive inhibitor of osteoclast (OC) development (13, 14). Thus, inflammatory and noninflammatory pathways operate downstream of the NLRP3 inflammasome. Elucidation of these pathways may have implications not only for bone diseases, but also for other conditions such as diabetes and atherosclerosis, in which low-grade sterile inflammation plays a pathogenic role.

We previously reported that global expression of Nlrp3^D301N in mice recapitulates in great detail the human phenotype of NOMID (15). In this study, we took advantage of cell lineage-specific expression of Nlrp3^D301N to determine whether skeletal manifestations can be uncoupled from systemic inflammation in NOMID. We find that expression of Nlrp3^D301N in OC is sufficient to cause bone loss in the absence of systemic inflammation.

MATERIALS AND METHODS

Generation of Nlrp3^D301N/+;LysM mice and Nlrp3^D301N/+;CatK mice

The D301N allele was engineered as previously described (15, 16). Briefly, the targeting construct pPNTlox2PNlrp3D301N was inserted into intron 2 of Nlrp3. 129 SvJ stem cells were electroporated with linearized pPNTlox2PNlrp3D301N, and selected colonies were used to create chimeric mice, which yielded offspring heterozygous for D301N. D301N expression in Nlrp3 was driven by Cre under the control of a lysozyme M (LysM) or cathepsin K (CatK) promoter, which target myeloid cells and OCs, respectively. Nlrp3^fl(D301N)/+ mice were mated with heterozygote LysM-Cre or CatK-Cre mice to obtain Nlrp3^fl(D301N)/+;LysM-Cre (Nlrp3^D301N/+;LysM) mice or Nlrp3^fl(D301N)/+;CatK-Cre (Nlrp3^D301N/+;CatK) mice and wild-type (WT; Nlrp3^+/+;LysM-Cre and Nlrp3^+/+;CatK-Cre) mice, respectively. All mice were on the C57BL6 background, and mouse genotyping was performed with PCR. All procedures were approved by the Institutional Animal Care and Use Committee of the University of California San Diego, La Jolla, and Washington University in St. Louis.

Bone mass and microstructure

Femoral bone structure was analyzed by a microcomputed tomography (µCT) system (μCT 40; Scanco Medical AG, Zurich, Switzerland) as described previously (15). Femora from 2-wk-old male and female Nlrp3^D301N/+;LysM and WT mice or 3-wk-old Nlrp3^D301N/+;CatK and WT mice were stabilized in 2% agarose gel, and µCT scans were taken along the length of the femur as described previously (15). Bones from Nlrp3^D301N/+;LysM and control mice were scanned at 45 kVp (low-energy setting because these bones are undermineralized), whereas those from Nlrp3^D301N/+;CatK mice and littermates were scanned at 55 kVp (higher energy). The volume of interest analyzed was located just proximal to the growth plate, spanning a height of 350 µm each for the metaphyseal region containing all the bone within the cortical shell.

Histology and histomorphometry

Tissue samples were processed as described previously (15). Briefly, samples were fixed in 10% formalin, decalcified in 14% (w/v) EDTA, pH 7.2, for 10–14 d at room temperature, embedded in paraffin, sectioned at 5 μm thickness, and mounted on glass slides. The sections were stained with hematoxylin and eosin or tartrate-resistant acid phosphatase and analyzed as described previously (15).

Peripheral blood analysis and preparation of bone marrow cells

Complete blood counts were performed by the University of California, San Diego American College of Physicians Diagnostic Laboratory as previously described (16). For bone marrow cell preparations, the femora and tibias were harvested into α-MEM containing 10% fetal bovine serum (Invitrogen), 100 μg/ml streptomycin, and 100 IU/ml penicillin G, and bone marrow cells were eluted from the bone marrow cavity by first removing the epiphyses of bones followed by centrifugation at 13,000 rpm for 30 s as previously described (15).

Flow cytometry

Mouse bone marrow cells were prepared as previously described (15). Cells (0.5–1 × 10⁶) were incubated with CD16/32 (BioLegend, San Diego, CA, USA) to block nonspecific Fc binding, stained with PE mAbs anti-Ly6G (enriched on neutrophils; BioLegend), Alexa-Fluor 647- mAb anti-CD11b (enriched on leukocytes; BioLegend), PerCP-Cy 5.5- mAb anti-Ly6C (enriched on leukocytes; BD Pharmingen), PerCP-Cy 5.5- mAb anti-CD117/c-Kit (BioLegend), V450- mAb anti-Gr1 (enriched on myeloid cells; BD Pharmingen, San Jose, CA, USA), or FAM-YVAD-FMK [fluorescent labeled inhibitor of caspases (FLICA); ImmunoChemistry Laboratories] according to the supplier’s instructions. Samples were acquired with a FACSCanto (BD Pharmingen), followed by analysis with FlowJo software (TreeStar, Ashland, OR, USA).

OC formation

Bone marrow macrophages (BMMs) were obtained by culturing bone marrow cells in culture media containing a 1:25 dilution of supernatant from the fibroblastic cell line, conditioned media from CMG 14-12 cells (CMG 14-12), as a source of macrophage colony-stimulating factor (M-CSF) (17), a mitogenic factor for BMMs, for ∼5 d in a 10 cm dish as described previously (15). Nonadherent cells were removed by vigorous washes with PBS, and adherent BMMs were detached with Trypsin-EDTA and plated at 5–10 × 10³ cells per well in a 96-well plate in culture media containing a 1:50 dilution of CMG and 100 ng/ml receptor activator of NF-κB ligand (RANKL), a required cytokine for OC differentiation. Media with supplements were changed every other day and maintained at 37°C in a humidified atmosphere of 5% CO₂ in air.

F-actin ring staining and in vitro bone resorption

BMMs were cultured on plastic for ∼3 d in culture media containing a 1:50 dilution of CMG and 100 ng/ml RANKL to generate pre-OCs, which were transferred to bone slices prepared from bovine cortical bones and cultured for 6 d. This protocol minimizes differences in OC activity as a result of differences in OC formation. The bone slices were fixed in 4% paraformaldehyde at room temperature for 10 min. After 3 washes with PBS, the bone slices were incubated in 1:100 Alexa Fluor phalloidin for 15 min at room temperature. The images of actin ring staining were acquired with fluorescence microscopy. Afterward, bone slices were washed with PBS 3 times, incubated with 2 N NaOH for 30 s, washed with PBS, and rubbed with a cotton swab to remove cells. Subsequently, bone slices were incubated with peroxidase conjugated wheat germ agglutinin (50 µg/ml) at 37°C for 30 min away from light, subjected to DAB staining for 20 min at room temperature, vigorously washed with PBS, air dried, and observed under light microscopy. The areas of resorption pits or F-actin rings were quantified using ImageJ (US National Institutes of Health, Bethesda, MD, USA).

Fluorescence microscopy

BMMs and OCs were treated with 100 ng/ml LPS or PBS for 3 h and then with 5 mM ATP or vehicle for 30 min. After 3 gentle washes with PBS, cells were stained with PBS containing a 1:150 dilution of FLICA for 45 min at 37°C and fixed in the fixative solution provided in the kit for 10 min at room temperature. For immunofluorescence, cells were fixed in a buffer containing 3.9% formaldehyde and 0.1% Triton X-100 for 15 min at room temperature and stained with 10 μg/ml NLRP3 antibody or 10 μg/ml control IgG for 1 h at 37°C and then with Alexa anti-mouse 555 for 1 h at 37°C. Cells were then stained with DAPI (1:1000) for 10 min at room temperature and examined under fluorescence microscopy.

TRAP staining

Cytochemical staining for TRAP was used to identify OC as described previously (15). Briefly, cells on a 96-well plate were fixed with 3.7% formaldehyde and 0.1% Triton X-100 for 10 min at room temperature. The cells were rinsed with water and incubated with the TRAP staining solution (Sigma leukocyte acid phosphatase kit) at room temperature for 30 min. Under light microscopy, multinuclear TRAP-positive cells with ≥3 nuclei were scored as OCs.

mRNA expression analysis

Total RNA was harvested from cells using RNeasy Plus Mini Kit (Qiagen, Germantown, MD, USA). cDNA was then synthesized with iScript reverse transcription kit (Bio-Rad, Hercules, CA, USA). Primers were prevalidated and obtained from Bio-Rad. Gene expression was analyzed by SYBR Green gene expression assay (Applied Biosystems, Foster City, CA, USA).

Cloning and retroviral infection

Mouse PARP1 cDNA was obtained from OriGene Technologies (Rockville, MD, USA). The flag-tagged PARP1 was amplified and inserted into the pMx vector. The primers used were ATGAGATCTATGGATTACAAGGATGACGATGACAAGGCGGAGTCTTCG and CTGGCGGCCGCTTACCACAGGGAG. The D214N mutation in PARP1 was generated with the site mutagenesis kit from Agilent Technologies (Santa Clara, CA, USA). Retroviruses were generated by transfection of relevant constructs into Plat-E packaging cells using Fugene 6 (Promega, Madison, WI, USA).

Western blot analysis

Cell extracts were prepared by lysing cells with RIPA buffer (50 mM Tris, 150 mM NaCl, 1 mM EDTA, 0.5% NaDOAc, 0.1% SDS, and 1.0% NP-40) plus phosphatase inhibitors (2 mM NaVO₄, 10 mM NaF, and 1 mM PMSF) and Complete Protease Inhibitor Cocktail (Roche, Basel, Switzerland). Protein concentrations were determined by the Bio-Rad method, and equal amounts of proteins were subjected to SDS-PAGE on 4–12% NuPAGE gels (Invitrogen, Carlsbad, CA, USA). Proteins were transferred onto nitrocellulose membranes and incubated with NLRP3 antibody (1:1000; Adipogen, San Diego, CA, USA), PARP-1 antibody (1:1000; Cell Signaling Technologies, Danvers, MA, USA), heat shock protein 90 antibody (1:2000; Santa Cruz Biotechnology, Santa Cruz, CA, USA), flag antibody (1:5000; Abcam, Cambridge, MA, USA) or β-actin antibody (1:50000; Sigma-Aldrich, St. Louis, MO, USA) for 2 h at room temperature, followed by a 1 h incubation with secondary goat anti-mouse IRDye 800 (Rockland, Limerick, ME, USA) or goat anti-rabbit Alexa-Fluor 680 (Molecular Probes, Grand Island, NY, USA), respectively. The results were visualized using Li-Cor Odyssey Infrared Imaging System (LI-COR Biosciences, Lincoln, NE, USA).

Immunoassays

BMMs were plated at 5 × 10⁴ cells per well on a 96-well plate and maintained for 24 h in culture media containing a 1:10 dilution of CMG, treated with 100 ng/ml LPS or PBS without changing media for 3 h, with 5 mM ATP for 30 min, and conditioned media were collected. Serum cytokine levels were analyzed by Luminex assay (Bio-Rad) according to the manufacturer’s instructions. The levels of serum C-telopeptide of type I collagen (CTX-1), a biomarker of bone resorption, were quantified using a Rat LAPS EIA kit from Immunodiagnostics Systems. Bone marrow IL-1β levels were quantified using the eBioscience (San Diego, CA, USA) ELISA kit.

Statistical analysis

Statistical significance was assessed by Student’s t test for independent samples, unless otherwise stated.

RESULTS

OC lineage expresses functional NLRP3 inflammasome

WT BMMs express NLRP3 (Fig. 1A, B) as reported widely. Here, we find that NLRP3 expression decreases during OC differentiation of WT BMMs treated with RANKL and M-CSF, and it remains at a lower abundance during the differentiation process (Fig. 1B). Immunofluorescence carried out with control isotype IgG (Fig. 1C, D) or anti-NLRP3 antibody (Fig. 1E) reveals punctate patterns of NLRP3 expression in mononucleated cells and OCs stained with anti-NLRP3 antibody (Fig. 1E, arrowhead), but not isotype IgG, suggesting that the NLRP3 inflammasome is functional. Active inflammasome components can also be detected by fluorescence microscopy as cellular foci on incubation with the caspase-1 FLICA FAM-YVAD-FMK probe (18). Indeed, the NLRP3 inflammasome is seen as individual foci of fluorescence in mononucleated cells, whereas OCs (47 ± 16% per field) exhibit multiple foci of inflammasome complexes (Fig. 1F, arrowhead). Thus, despite a decrease in expression levels, NLRP3 assembles an active inflammasome complex during OC formation in the absence of priming inflammatory signals.

Figure 1.

OC lineage expresses functional NLRP3 inflammasome. Quantitative PCR (A), Western blot (B), immunofluorescence (C–E), or FLICA (F) analysis of NLRP3 expression in WT BMMs or BMMs treated with RANKL and M-CSF for 2, 3, or 4 d to induce OC formation (referred to as OC 2, 3, 4 d). For quantitative PCR, OC data were normalized to BMM values set up as 1. For immunofluorescence, WT OC 4 d were incubated with IgG control (C, D) or anti-NLRP3 antibody (E). (D) Phase contrast image of C. No fluorescent signal was detected in FLICA controls incubated without the FAM-YVAD-FMK probe (data not shown). Arrowhead points out active inflammasome foci, and circular dashed line depicts OC. Scale bar, 100 (C–E) or 200 µm (F). Data are representative of ≥2 independent experiments and expressed as means ± sd.

Expression of Nlrp3^D301N in myeloid cells causes constitutive activation of the NLRP3 inflammasome

We recently reported that mice globally expressing Nlrp3^D301N produce exceedingly high levels of IL-1β, develop systemic inflammation, and show growth retardation (15). They also display diminished bone mass as a consequence of accelerated OC formation and bone resorption, thus modeling human NOMID. These mice usually die within 2–3 wk of age, thus precluding studies in adult animals. In an attempt to overcome this early lethality and to identify the cells responsible for the bone phenotype, we mated Nlrp3^fl(D301N)/+ mice with lysosyme M-Cre (LysM-Cre) mice to generate mice with myeloid activation of the single Nlrp3^D301N allele (Nlrp3^D301N/+;LysM mice) and Nlrp3^+/+;LysM-Cre (WT) mice. Likewise, the mating of Nlrp3^fl(D301N)/+ mice with cathepsin K-Cre (CatK-Cre) generates mice with OC-restricted activation of Nlrp3^D301N (Nlrp3^D301N/+;CatK mice) and Nlrp3^+/+;CatK-Cre (WT) mice.

To determine NLRP3 inflammasome activation in vivo, we used the FLICA FAM-YVAD-FMK probe that binds to active caspase-1 (18). Flow cytometry analysis shows that nearly all Nlrp3^D301N/+;LysM CD11b⁺ bone marrow cells are FLICA⁺ (Fig. 2A, lower), and the percentage of CD11b⁺/FLICA⁺ cells is significantly higher in mutant mice compared with WT mice (Fig. 2B). Interestingly, ∼10% of WT CD11b⁺ cells and 20% of CD11b⁻ cells are also FLICA⁺ (Fig. 2A). We also used fluorescence microscopy (18) to determine inflammasome activation in vitro. Although the NLRP3 inflammasome is activated in LPS-primed WT BMMs only in the presence of ATP (Fig. 2C, upper, D), LPS-primed Nlrp3^D301N/+;LysM BMMs display inflammasome activation even in the absence of ATP (Fig. 2C, lower, D). Thus, the D301N mutation causes constitutive activation of the NLRP3 inflammasome.

Figure 2.

Expression of Nlrp3^D301N in myeloid cells causes constitutive activation of the NLRP3 inflammasome. A, B) Flow cytometry analysis of bone marrow cells labeled with Alexa-Fluor 647-anti-CD11b antibody and FLICA. C, D) FLICA analysis of WT and Nlrp3^D301N/+;LysM BMMs treated with 100 ng/ml LPS for 4 h and then exposed or not to 5 mM ATP for 30 min. Arrowhead points out active inflammasome foci. Scale bar, 100 µm. Data are representative of ≥3 independent experiments and expressed as means ± sd. *P < 0.05 vs. WT and **P < 0.001 vs. LPS for 4 h without ATP.

Expression of Nlrp3^D301N at various stages of OC development induces osteopenia

µCT analysis of the femora reveals significantly lower bone mass (BV/TV), reduced thickness (Tb.Th), number of the trabeculae (Tb.N), bone mineral density, and increased trabecular space (Tb.Sp) in Nlrp3^D301N/+;LysM mice (Fig. 3A, B) and Nlrp3^D301N/+;CatK mice (Fig. 3C, D) relative to corresponding WT counterparts. Bones from Nlrp3^D301N/+;LysM mice and littermates were scanned at 45 kVp (low-energy setting) because these mice were young (2 wk old), whereas 3-wk-old Nlrp3^D301N/+;CatK and control bones were scanned at 55 kVp. This results in higher baseline BV/TV in WT mice of the former cohort. OC surface (Oc.S/BS) is increased in Nlrp3^D301N/+;LysM compared with WT mice (Fig. 3B), consistent with intense OCs stained in red on trabecular bone surfaces of mutant mice (Supplemental Fig. S1A). In contrast, OC surface (Fig. 3D and Supplemental Fig. S1B) is comparable between Nlrp3^D301N/+;CatK and control mice. These results are consistent with the hypothesis that the low bone mass of Nlrp3^D301N/+;LysM mice, but not Nlrp3^D301N/+;CatK mice, is due at least in part to increased osteoclastogenesis. To test this hypothesis, we exposed BMMs to RANKL and M-CSF to induce OC differentiation in vitro. OC formation is increased in Nlrp3^D301N/+;LysM compared with WT cultures (Fig. 4A), consistent with accelerated expression of OC differentiation markers in mutant cells (Supplemental Fig. S1C), although some variability in osteoclastogenesis was noticed. In contrast, OC differentiation is comparable between Nlrp3^D301N/+;CatK and WT controls (Fig. 4B). Although the number of CD11b⁺/Gr1⁺ cells is notably higher in the bone marrow of Nlrp3^D301N/+;LysM (Fig. 4C and Supplemental Fig. S1D) relative to WT mice, it is comparable between Nlrp3^D301N/+;CatK and control mice (Fig. 4D and Supplemental Fig. S1E). Moreover, although the number of CD11b^high/Gr1^high/CD117⁺ cells is similar in bone marrow of mutant and WT mice (Fig. 4Ei, Fi), the percentage of CD11b^low/Gr1^low/CD117⁺ cells (cells with higher OC differentiation potential) is higher in marrow of Nlrp3^D301N/+;LysM (Fig. 4Eii) mice, but not Nlrp3^D301N/+;CatK (Fig. 4Fii) mice, relative to WT littermates. These findings suggest that the OC precursor pools are expanded in Nlrp3^D301N/+;LysM mice, but not Nlrp3^D301N/+;CatK mice.

Figure 3.

Nlrp3^D301N/+;LysM mice and Nlrp3^D301N/+;CatK mice develop osteopenia. Longitudinal and cross sections of 3D μCT reconstruction of distal femoral metaphyses from 2-wk-old WT and Nlrp3^D301N/+;LysM mice (A) or 3-wk-old WT and Nlrp3^D301N/+;CatK mice (C). Scale bar, 500 µm. μCT analysis of trabecular bone mass (BV/TV), trabecular thickness (Tb.Th), trabecular number (Tb.N), bone mineral density (BMD), trabecular space (Tb.Sp), and histomorphometric analysis of OC surface/bone surface (Oc.S/BS) from WT and Nlrp3^D301N/+;LysM mice (B) or WT and Nlrp3^D301N/+;CatK mice (D). Data are from 5 to 6 mice (A, B) or 4 to 5 mice (C, D) per genotype and are expressed as the mean ± sem (μCT data) or the mean ± sd (Oc.S/BS data). *P < 0.05.

Figure 4.

Expression of Nlrp3^D301N in myeloid cells, but not OC, causes expansion of OC precursors and increases OC formation. A, B) OC 4 d. Scale bar, 500 µm. C, D) Flow cytometry analysis of bone marrow cells stained with antibodies against CD11b, Gr1, and CD117. E, F) Expression of CD117 was analyzed by gating cells expressing (i) CD11b^high/Gr1^high or (ii) CD11b^low/Gr1^low. Data are from 1 experiment with the same trend observed in 2 other independent experiments (A) or representative of ≥2 independent experiments (B–F). *P < 0.05.

We also analyzed the impact of Nlrp3^D301N on OC bone-resorbing activity in vitro in the presence of RANKL and M-CSF. BMMs were induced to differentiate into pre-OCs for 3 d on culture dishes and then lifted and cultured on bone slices. Nlrp3^D301N/+;LysM OCs form more resorption pits (Fig. 5A) with overall larger area (Fig. 5B) than WT counterparts, despite comparable OC number between genotypes (Fig. 5C). Accordingly, F-actin rings, surrogate markers of OC activity, are frequent (Fig. 5D) and overall are larger (Fig. 5E) in mutant relative to WT OCs. Nlrp3^D301N/+;CatK OCs also form a larger number of resorption lacunae, which are more intensely stained compared with WT OCs (Fig. 5F), and larger F-actin ring area (Fig. 5G), despite a comparable number of OCs on bone slices (Fig. 5H), results corroborated by the higher release of CTX-1 by mutant cells (Fig. 5I). Thus, a hyperactive NLRP3 inflammasome promotes the expansion of OC precursor pools, affects the reorganization of actin cytoskeleton, and up-regulates OC activity.

Figure 5.

Expression of Nlrp3^D301N stimulates OC activity. Pit staining (A), pit area/field (B), OC number on bone slices (C), percentage of OC displaying F-actin rings (D), and ring area/OC (E) in cultures from WT and Nlrp3^D301N/+;LysM cells. Scale bar, 200 µm. Pit staining (F), F-actin ring area (G), OC number on bone slices (H), and CTX-1 levels in the supernatants from OC cultures on bone slices for 6 d (I) from WT and Nlrp3^D301N/+;CatK cells. Scale bar, 200 µm. Data are representative of ≥3 independent experiments and expressed as the mean ± sd. *P < 0.05.

Nlrp3^D301N/+;LysM mice, but not Nlrp3^D301N/+;CatK mice, develop systemic inflammation

Nlrp3^D301N/+;LysM mice exhibit peripheral neutrophilic leukocytosis, lymphopenia, thrombocytosis, and anemia (Supplemental Fig. S2A) and overproduction of IL-18 (Supplemental Fig. S2B), IL-6 (Supplemental Fig. S2C), and granulocyte-colony stimulating factor (G-CSF) (Fig. 6A) in serum. IL-1β levels are also higher in Nlrp3^D301N/+;LysM mouse bone marrow compared with control littermates (Fig. 6B). Although LPS-primed WT BMMs produce IL-1β only on exposure to ATP in vitro, LPS-primed Nlrp3^D301N/+;LysM BMMs secrete high levels of IL-1β at baseline and a further modest induction in the presence of ATP (Fig. 6C). Nlrp3^D301N/+;LysM mice also exhibit stunted growth and smaller body size (Supplemental Fig. S2D), early lethality (Supplemental Fig. S2E), and granulocytic infiltrates in the synovium (Supplemental Fig. S3F) and bone marrow (Supplemental Fig. S1D). In contrast to Nlrp3^D301N/+;LysM mice, ∼90% of Nlrp3^D301N/+;CatK mice survive for >8 wk and exhibit slightly decreased body weight (Supplemental Fig. S3A), and ∼20% of mutant mice develop inflammation of the skin (Supplemental Fig. S3B) where cathepsin K is also expressed (19). CD11b⁺/Ly6C⁺ (Supplemental Fig. S2C, D), serum G-CSF levels (Fig. 6D), and IL-1β levels in bone marrow (Fig. 6E) are similar between WT and Nlrp3^D301N/+;CatK mice. Likewise, LPS-primed WT and Nlrp3^D301N/+;CatK BMMs produce comparable levels of IL-1β, whereas the cytokine is undetectable in OC cultures (Fig. 6F). These findings are consistent with the mutant Nlrp3 allele being transcriptionally silent (due to lack of neoR excision) and the fact that CatK-Cre is not expressed in BMMs at early stage of differentiation. Thus, expression of Nlrp3^D301N at early stages, but not late stages, of OC development causes systemic inflammation.

Figure 6.

Expression of Nlrp3^D301N in myeloid cells, but not OC, causes cytokine over-production. G-CSF serum levels from 2-wk-old (A, 3–7 mice/genotype) or 3-wk-old mice (D, 4–5 mice/genotype). IL-1β levels in the supernatants from centrifuged bone marrow from WT and Nlrp3^D301N/+;LysM mice (B) or WT and Nlrp3^D301N/+;CatK mice (E). IL-1β levels in CM from WT and Nlrp3^D301N/+;LysM BMMs (C), WT, and Nlrp3^D301N/+;CatK BMMs or OC 4 d (F) treated with 100 ng/ml LPS for 3 h and then with 5 mM ATP for 30 min. Data are representative of ≥3 independent experiments and expressed as means ± sem (A, D–F) or means ± sd (B, C). *P < 0.05; **P < 0.001.

PARP1 is targeted for proteolytic processing by the NLRP3 inflammasome in myeloid cells and during OC differentiation

NLRP3 and NLR family, caspase activation, and recruitment domain-containing 4 (NLRC4) inflammasomes promote the cleavage of PARP1 (11, 12), a presumptive negative regulator of OC formation (13, 14), suggesting that these pathways may intersect during OC development. Indeed, although PARP1 is detected in bone marrow cells freshly isolated from WT, it is reduced or absent in Nlrp3^D301N/+;LysM cells (Fig. 7A). Accordingly, PARP1 abundance is higher in expanded WT BMMs relative to Nlrp3^D301N/+;LysM BMMs despite comparable NLRP3 levels (Fig. 7B). Although PARP1 mRNA expression is not altered during osteoclastogenesis (Fig. 7C), PARP1 protein is not detected in WT or Nlrp3^D301N/+;LysM OC total lysates (OC 4 d; Fig. 7B), suggesting that PARP1 degradation occurs during OC differentiation. The inflammasome activator, ATP, causes PARP1 cleavage (Fig. 7D) as previously reported (11). Importantly, the caspase-1 inhibitor, YVAD-FMK (YVAD), which inhibits ATP induced-PARP1 cleavage in LPS-stimulated WT BMMs (Fig. 7D), also attenuates OC differentiation from WT and mutant cells (Fig. 7E) and prevents PARP1 degradation during this process (Fig. 7F). Accordingly, retroviral-mediated forced expression in BMMs of noncleavable PARP1, which harbors the D214N mutation (PARP1^D214N) (12, 24), but not WT PARP1, blunts OC differentiation (Fig. 7G, H and Supplemental Fig. S4A, B). Thus, although PARP1 is cleaved in a site-specific manner following NLRP3 inflammasome activation in BMMs, it undergoes extensive proteolysis for osteoclastogenesis to proceed.

Figure 7.

PARP1 proteolysis is induced on activation of the NLRP3 inflammasome, and inhibition of PARP1 processing attenuates OC differentiation. A) Western blot analysis of bone marrow cells freshly isolated from 4 WT and 2 Nlrp3^D301N/+;LysM mice. Western blot analysis (B) or quantitative PCR analysis (C) of samples from BMMs treated with RANKL and M-CSF for 2 or 4 d (referred to as OC 2, 4 d). D) WT BMMs treated with vehicle or 100 ng/ml LPS for 3 h and then with 10 μM YVAD for 30 min before exposure to 5 mM ATP for 30 min (cPARP1, cleaved PARP1). E) OC 4 d (BMMs were treated with RANKL and M-CSF for 4 d in the absence or presence of 10 μM YVAD. Scale bar, 500 µm. F) Western blot analysis of lysates from OC cultures for 2, 3, or 4 d in the absence or presence of 10 μM YVAD. HSP90 was used as loading control. The intensity of each band was quantified by densitometry. The numbers below the blots represent the ratio of PARP1 over heat shock protein 90 intensity. G, H) OC formation from WT BMMs transduced with green fluorescent protein, WT PARP1, or PARP1^D214N and then treated with RANKL and M-CSF for 4 d. Data are representative of ≥2 independent experiments (A–F) or from 1 experiment with the same trend observed in another experiment (G, H) and expressed as the mean ± sd. *P < 0.05; **P < 0.001 vs. green fluorescent protein control cultures.

Discussion

Our work reveals 2 distinct, non–mutually exclusive mechanisms through which Nlrp3^D301N alters bone homeostasis in mice (Fig. 8): inflammation-dependent and -independent regulation of OC differentiation and function. In support of the former mechanism, mice expressing the active Nlrp3^D301N allele in myeloid cells exhibit an inflammatory phenotype similar to that of mice globally expressing the mutation (15), consistent with the preponderant role of the NLRP3 inflammasome in the myeloid lineage (16). Mice expressing Nlrp3^D301N in myeloid cells do not exhibit growth plate abnormalities but display the severe osteopenia seen in global mutants. Indeed, the severe systemic inflammation and massive osteolysis in both mouse strains consistently correlate with a higher number of OC in vivo, but with moderate OC development in vitro, implying that the exuberant osteoclastogenesis in NOMID mice may be secondary to the expansion of myeloid cells, including OC precursors. It may be that NOMID cells lose OC formation potential to some extent in vitro. Alternatively, because the NLRP3 inflammasome is activated during OC differentiation in WT cells, the active Nlrp3^D301N mutant may not offer a competitive differentiation advantage in an in vitro microenvironment. In contrast, within the bone marrow milieu of NOMID mice, overexpressed proinflammatory cytokines such as IL-1β and IL-6 may act on stromal cells and osteoblasts to stimulate the expression of osteoclast-inducing factors such as M-CSF and RANKL, which in turn promote the differentiation of expanded OC precursors into OCs and ultimately accelerate bone resorption.

Figure 8.

Mechanisms through which the Nlrp3^D301N inflammasome regulates osteolysis in NOMID. The Nlrp3^D301N inflammasome in myeloid cells, but not differentiated OCs causes systemic inflammation leading to increased expression of osteoclastogenic mediators and degradation of PARP1, ultimately expansion of OC precursors and OC formation. This inflammasome also promotes the reorganization of actin-based cytoskeleton, through an unknown (?) but shared mechanism of inflammasome-mediated OC activity in inflammatory and noninflammatory osteolysis.

We also find that the NLRP3 inflammasome regulates the fate of PARP1 in the OC cell lineage, consistent with previous reports showing that this inflammasome activates caspase-7, which translocates to the nucleus to cleave PARP1 (12). Beside its role in DNA repair, PARP1 also regulates cell differentiation by regulating gene expression directly through poly(APD-ribosyl)ation (PARylation) of transcription factors such as Sox2 (20) or indirectly through PARylation of histones, an event that causes chromatin decondensation, thereby enabling transcription factors access to specific binding sites (21). Our results suggest a negative role of PARP1 at least in the late steps of OC differentiation as NLRP3 inflammasome activation inversely correlates with PARP1 abundance, and expression of noncleavable PARP1 impairs full OC development. This view is consistent with the accumulating evidence on nonapoptotic functions of PARP family members in cell differentiation. Indeed, evidence indicates that tankyrases 1 and 2 (aka PARP5 and PARP6, respectively) are negative regulators of osteoclastogenesis as mutations in 3BP2, which prevent tankyrase-mediated 3BP2 destruction cause massive bone destruction in cherubism (22). PARP10 interferes with NF-κB essential modulator poly-ubiquitination, thereby inhibiting NF-κB signaling, a critical event in osteoclastogenesis (23). Intriguingly, PARP1 cleaved fragments function as coactivators for transcription factors such as NF-κB (12, 24), whereas noncleavable PARP1 represses NF-κB-dependent gene expression and attenuates endotoxemia-induced inflammatory responses in mice (25). Thus, the actions of PARP1 are complex, and further studies are needed to elucidate the underlying mechanisms of PARP1 in the regulation of OC formation in physiologic and pathologic conditions.

Conversely, restricted activation of the NLRP3 inflammasome to the OC lineage is sufficient to cause osteopenia in the absence of systemic inflammation or changes in OC number. Lack of IL-1β overproduction in this mouse strain is not surprising because OCs do not express this cytokine. Increased bone resorption without change in osteoclastogenesis in mice with OC-restricted activation of the NLRP3 inflammasome implies that enhanced OC activity drives bone resorption in this mouse model. We find that NLRP3 assembles a functional inflammasome in WT and mutant pre-OCs and OCs in the absence of proinflammatory LPS-priming signals and exogenously added ATP. Although inflammasome-activating signals may be dispensable for mutant cells, RANKL-induced calcium fluxes, reactive oxygen species production, or mitochondrial alterations during OC differentiation are potential NLRP3 inflammasome activators in WT cells (26, 27). In addition, ATP and cellular debris from dying cells, including OCs, which have a short lifespan in cultures, may also activate this inflammasome. Furthermore, bone extracellular matrix degradation products from bone may further activate the inflammasome, thereby providing a positive feedback loop that amplifies bone resorption. Consistent with the role of this inflammasome in OC activity, bone resorption is increased in vitro in OCs expressing the active inflammasome mutant. Remarkably, the number of OCs displaying F-actin rings is higher in mutant cells, which also form larger F-actin rings than WT cells.

Our findings indicate that the NLRP3 inflammasome plays a role in cytoskeletal reorganization in OCs. Despite the gaps in knowledge regarding the underpinning molecular mechanisms, this response is worth noting as it occurs in the absence of inflammatory cues and is consistent with the emerging evidence on cytokine-autonomous actions of the NLRP3 inflammasome complex in various disorders (6) and the concept of numerous substrates and effectors of the inflammasomes. Interestingly, caspase-1 has been shown to inactivate several enzymes involved in glycolysis (28), although inflammasome-independent functions of caspase-1 have been reported. It is also noteworthy that our results contrast with a publication showing that ASC, but not NLRP3, is important for actin polymerization in leukocytes (29). The reasons for this discrepancy are unclear, but may be related to the fact that we studied a gain-of-function Nlrp3 mutation, whereas the earlier published work used Nlrp3-deficient cells (29), or may reflect a cell context-specific action of the inflammasome.

In conclusion, our study provides insights into the mechanisms through which dysregulation of the NLRP3 inflammasome causes bone resorption in NOMID. Specifically, we show that Nlrp3^D301N induces osteolysis through the control of PARP1 metabolism, the maturation of inflammatory mediators such as IL-1β, and up-regulation of OC activity.

Glossary

ASC

apoptosis-associated speck-like protein containing a caspase activation and recruitment domain

BMMs

bone marrow macrophages

CMG 14-12

conditioned media from CMG 14-12 cells

CTX-1

C-telopeptide of type I collagen

FLICA

fluorescent labeled inhibitor of caspases

G-CSF

granulocyte-colony stimulating factor

M-CSF

macrophage colony-stimulating factor

NLRP3

NOD-like receptor (NLR) family, pyrin domain-containing 3

NOMID

neonatal-onset multisystem inflammatory disease

OC

osteoclast

PARP1

poly(ADP-ribose) polymerase 1

RANKL

receptor activator of NF-κB ligand

WT

wild-type

μCT

microcomputed tomography