Introduction

Paget’s disease of the bone (PDB) is an age-dependent bone disease, wherein patients exhibit symptoms of focal areas of bone fragility, bone deformity, pathological fracture, and nerve root compression. PDB pathology has been attributed to osteoclast hyperactivation, leading to excessive bone resorption and irregular bone remodeling [1]. Compared with normal osteoclasts, Pagetic osteoclasts are increased in number, nuclei, activity, and function [2, 3]. Furthermore, Pagetic osteoclast precursors display heightened osteoclastogenic potential, as they have an increased responsiveness to RANKL, a critical cytokine for osteoclast differentiation [4]. The molecular mechanisms that underlie hyperactivation of Pagetic osteoclasts, however, remain unknown.

While environmental factors, such as viral infections and toxins, can contribute to the etiology of PDB, recent evidence indicates that genetic factors play a predominant role in PDB pathogenesis [5, 6]. Mutations in Sqstm1, which encodes the autophagy receptor p62, have been repeatedly reported among familial PDB patients [7], and p62 dysfunction results in increased generation and activity of osteoclasts in vitro [8]. Recent genome-wide association studies (GWAS) have identified another autophagy receptor, Optineurin (OPTN), that is genetically linked to PDB [9], with the PDB-associated OPTN variant resulting in a decreased gene expression [10]. OPTN is ubiquitously expressed and mediates the delivery of ubiquitinated cargos, such as damaged organelles, protein aggregates, and intracellular pathogens, to autophagosomes during selective autophagy [11]. OPTN also functions in multiple non-autophagic processes, including vesicle trafficking, Golgi organization, and modulation of the NF-κB pathway [12,13,14,15,16]. In addition to PDB, genetic mutations in OPTN underlie primary open-angle glaucoma (POAG) and amyotrophic lateral sclerosis (ALS). While the impacts of OPTN in POAG and ALS have been intensively studied [17,18,19], the specific role of OPTN in PDB remains to be established.

A recent study demonstrated that while siRNA-mediated knockdown of Optn increased osteoclastogenesis in vitro, only ~10% of 15- to 18-month-old mice harboring a loss-of-function mutation in the ubiquitin-binding domain (OptnD477N/D477N) developed PDB lesions [10]. However, the role of OPTN in molecular pathways beyond ubiquitin binding has not be examined in relation to PDB pathogenesis. Therefore, we generated Optn global knockout (Optn–/–) mice and performed detailed bone phenotyping analysis. We discovered that Optn–/– mice spontaneously develop late-onset polyostotic osteolytic lesions that are reminiscent of clinical findings in PDB patients. Similarly, we observed an increased capacity for osteoclast differentiation with elevated c-Fos expression in the absence of Optn ex vivo. Mechanistically, OPTN’s inhibition of osteoclastogenesis is twofold. Optn–/– osteoclasts produce significantly decreased levels of IFNβ, a known negative regulator of c-Fos [10], as well as display a previously undescribed defect in signaling through the IFNα/βR. This defective IFNα/βR signaling also resulted in decreased cell death and hence increased survival. Thus, we have generated and characterized a novel and clinically relevant mouse model of PDB and identified a novel OPTN-type I IFN axis in PDB pathogenesis.

Materials and methods

Animals

Optnflox/flox mice on a C57BL/6 background were generated using a targeting vector-inserted LoxP site that flanks the first coding exon and a neomycin selection cassette (Fig. S1). We crossed Optnflox/flox mice with CMV-Cre mice (Jackson Laboratories, Bar Harbor, ME, USA) to generate CMV-Cre; Optnflox/wt mice, which were used as breeding pairs. Global Optn knockout (CMV-Cre; Optnflox/flox) mice are hereafter referred to as Optn−/− mice, and the primers used for genotyping are listed below.

Primer set

Forward

Reverse

A-B3

taggacctgttaccatgtccca

ccctgttcattcaggcccaaag

A-D1

taggacctgttaccatgtccca

cttggctggacgtaaactcctc

G2-H2

gcccggtaccatcaagtcta

aacacctctccagtgcaacc

Female Optn−/− mice were aged for up to 22 months, as aged male mice are more susceptible to spontaneous osteoarthritis, to minimize the influence of age-related osteoarthritis on the phenotype [20, 21]. Ifnar−/− mice were a generous gift from Dr. Michael Fessler (NIEHS). All animal procedures were approved by Institutional Animal Care and Use Committees at the University of North Carolina, Duke University, and NIEHS.

Micro-CT scanning and dual-energy X-ray absorptiometry (DEXA)

Bones were harvested, fixed in 4% paraformaldehyde, and scanned by MicroCT Scanco 40 (Scanco Medical, Bassersdorf, Switzerland) in 10-μm resolution (E = 70 kVa; I = 145 μA). Regions of interest of the cortical and trabecular bone were measured 0.7 mm proximal to the distal tibiofubular junction, and 0.7 mm distal to metaphysics of tibia, respectively. The reconstructed solid 3D images were used for visualizing bone morphology and microarchitecture. We measured bone mineral content (BMC) and bone mineral density (BMD) of lumbar spines and femurs by DEXA using LUNAR PIXImus bone densitometer (GE Healthcare, Fairfield, CT, USA).

Slide staining

Bones were decalcified in 10% EDTA for 3 weeks and then processed, paraffin embedded, and sectioned at thickness of 5 μm. Sections were stained with hematoxylin and eosin (H&E) for general histology, and with tartrate-resistant acid phosphatase (TRAP) activity to detect osteoclasts, with safranin O for cartilage. For immunohistochemistry staining, enzymatic antigen retrieval was performed on decalcified sections. After overnight incubation with the primary antibody, sections were incubated with the biotinylated secondary antibody (1:1000; Vector Laboratories, Burlingame, CA, USA). The sections were then immersed in a solution containing avidin–biotin peroxidase complex (Vector Laboratories, Burlingame, CA, USA), and 3,3′-diaminobenzidine was used as the chromogen.

Serum measurement of PDB biomarkers

Sera were isolated from whole blood by centrifugation of 3000 rpm at 4 °C for 15 min. Serum ALP level was measured by ALP Assay Kit (Abcam, Cambridge, UK). Serum concentrations of IL-6 was measured by Milliplex Multiplex Assay (EMD Millipore, Burlington, MA, USA). The C-terminal propeptide (sCTx) was measured using ELISA kits (Immunodiagnostic Systems, Boldon, UK).

Osteoclast differentiation and resorptive assay in vitro

After euthanasia by CO2, tibias and femurs were harvested from 8- to12-week-old mice. Bone marrow cells were flushed into the phenol-free α-MEM medium, supplemented with 10% FBS, L-glutamine, nonessential amino acids, and penicillin/streptomycin. Non-adherent cells were harvested after 24 h and re-plated at a density of 1.5 × 105 cells/cm2 with 30 ng/mL M-CSF (R&D Systems, Minneapolis, MN, USA). After 2 days, the medium was replenished with 30 ng/mL M-CSF and 10 ng/mL RANKL (R&D Systems Minneapolis, MN, USA) for osteoclast differentiation. After 3 days of culture, cells were fixed and stained with tartrate-resistant acid phosphatase (TRAP) to detect osteoclasts [22]. Equal numbers of bone marrow osteoclast precursors were plated and differentiated on Osteo Assay Surface plates (Corning Lifesciences, Tewksburg, MA, USA) for resorption assay. At day 5, the plate was bleached, and areas of resorption pits were quantified using NIH ImageJ.

In vitro IFNβ and anti-IFNaR antibody treatment

Bone marrow osteoclast precursors were treated with 1 U IFNβ (R&D Systems, Minneapolis, MN, USA) in the osteoclastogenic medium (30 ng/mL M-CSF and 10 ng/mL RANKL) for 3 days. Similarly, bone marrow osteoclast precursors were treated with an anti-IFNAR antibody or control antibody with a working concentration of 5 μg (Leinco, Fenton, MO, USA) in the osteoclastogenic medium for 3 days. Cells were fixed and stained for TRAP activity to evaluate osteoclastogenesis.

Antibodies and western blot analysis

The following antibodies were from Cell Signaling (Beverly, MA, USA): c-Fos (Cat. #4384), LC3B (Cat. #2775), p62 (Cat. #5114), ATG5 (Cat. #12994), ATG7 (Cat. #2631), p65 (Cat. #8242), phos-p65 (Cat. #3033), p100/p52 (Cat. #4882), IκBα (Cat. #4814), phos-p38 (Cat. #9215), p38 (Cat. #9212), phos-ERK1/2 (Cat. #4370), ERK1/2 (Cat. #4695), phos-JNK (Cat. #4668), JNK (Cat. #9252), phos-CREB (Cat. #9198), CREB (Cat. #4820), phos-STAT1 (Cat. #9167), STAT1 (Cat. #9172), STAT2 (Cat. #72604), SOCS3 (Cat. #2932), phos-TBK1 (Cat. #5483), TBK1 (Cat. #3504), phos-IRF3 (Cat. #4947), IRF3 (Cat. #4302), and RIPK1 (Cat. #3493). The following antibodies were from Santa Cruz Biotechnology Inc. (Dallas, TX, USA): NFATc1 (Cat. #sc-7294), OPTN (Cat. #sc-166576), and actin (Cat. #sc-1616). Anti-OPTN antibodies were from ProteinTech (Rosemont, IL, USA; Cat. #10837–1-AP) and Cayman Chemical (Ann Arbor, MI, USA; Cat. #100000). The myosin VI antibody was from Proteus (Ramona, CA, USA; Cat. #25–6791); Huntingtin antibody was from Millipore (Burlington, MA, USA; Cat. #MAB2166); Rab8a antibody was from BD Biosciences (San Jose, CA, USA; Cat. #610845); phos-STAT2 antibody was from Abcam (Cambridge, MA, USA; Cat. #ab53132). Secondary antibodies were from Jackson Immuno-Research (West Grove, PA, USA).

Proteins were harvested, and protein concentration was measured as previously described [23]. In total, 5–10 µg of the total protein lysate was resolved by Criterion TGX precast gel (Biorad, Hercules, CA, USA) and transferred to the nitrocellulose membrane using the Trans-Blot Turbo Transfer System (Biorad, Hercules, CA, USA) and immunodetected using appropriate primary and peroxidase-coupled secondary antibodies (Jackson Immunoresearch, West Grove, PA, USA). Proteins were visualized by enhanced chemiluminescence (ECL, Amersham Bioscience, Little Chalfont, UK).

Osteoblast differentiation in vitro

MC3T3-E1 pre-osteoblasts (Subclone 14, CRL-2594) were obtained from ATCC (Manassas, VA, USA). The cells were seeded at 2 × 105 cells per 35 -mm dish, expanded in growth media (α-MEM containing 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin), and differentiated with growth media supplemented with 10 mM β-glycerophosphate and 0.2 mmol/L ascorbic acid. The media was changed every 3 days. After 7, 14, and 21 days, the cells were fixed with 75% ethanol for 30 min and then stained with 1% Alizarin Red (Acros Organics, Geel, Belgium) solution (pH 4.2) for 10 min at RT.

Fluorochrome labeling for in vivo bone formation

Two fluorochromes (Sigma-Aldrich, St. Louis, MO, USA), calcein (20 mg/kg) and alizarin red (30 mg/kg), were used to label in vivo bone formation by intraperitoneal injection. The calcein and alizarin red were administered 5 days and 2 days prior to euthanasia, respectively. Mineral apposition rate (MAR, µm/day), was calculated as the distance between two sequential labels divided by the interlabeling period (days).

RNA isolation and qPCR

The total RNA was isolated using the RNeasy plus mini kit (Qiagen, Hilden, Germany) and was reverse-transcribed using the iScript DNA synthesis kit (Biorad, Hercules, CA, USA). qPCR was performed using Taqman Universal PCR master Mix (ThermoFisher, Waltham, MA, USA) with Taqman probes (Gapdh: Cat. #4331182, Ifnb1: Cat. #4331182). The transcript level of Ifnb1 was normalized to the level of Gapdh within each sample using the ΔΔCt method.

Microarray

RNA was isolated from day 2 Optn+/+ and Optn−/− bone-marrow-derived osteoclasts as described above, and gene expression analysis was conducted using Affymetrix Mouse Genome 430 2.0 GeneChip® arrays (Affymetrix, Santa Clara, CA, USA). Arrays were scanned in an Affymetrix Scanner 3000 and preliminary analyses were performed with OmicSoft Array Studio (Version 9.0) software.

Flow cytometry

To assess cell death, osteoclast precursors were harvested 24 and 48 h after treatment with RANKL in the presence or absence of IFNβ and stained with Annexin V (1:50), Zombie-Red (1:1000), and CD45 (1:200), as previously described [24]. All antibodies were from Biolegend (San Diego, CA, USA).

Statistical analysis

For all in vitro studies, three independent experiments were performed. Data are presented as mean ± SD. Student’s T test or analysis of variance (ANOVA) was used to determine the differences among groups. A p-value less than 0.05 is considered statistically significant.

Results

Optineurin is highly expressed in the bone marrow and is upregulated during in vitro osteoclastogenesis

To explore the relevance of OPTN in bone biology, we first examined its expression pattern in the long bone by immunohistochemistry. OPTN was most highly expressed in the bone marrow (BM), but was also expressed in osteocytes and chondrocytes (Fig. S1A). We next examined the temporal expression of OPTN during in vitro osteoclastogenesis, using both bone marrow cells (Fig. S1B) and RAW 264.7 cells (Fig. S1C) as osteoclast precursors. Osteoclasts were fully differentiated on days 3 and 4 post RANKL stimulation of BM cells and RAW 264.7 cells, respectively, as determined by tartrate-resistant acid phosphatase (TRAP) staining. Expression of OPTN increased during osteoclast differentiation, peaking at days 3 and 2 post RANKL stimulation in BM cells and RAW 264.7 cells, respectively (Fig. S1B-C). Immunofluorescence staining further demonstrated that OPTN was localized in the cytoplasm and perinuclear space of differentiating osteoclasts (Fig. S1B-C).

Young Optn −/− mice do not display gross skeletal defects in vivo

To study the role of Optn during bone homeostasis in vivo, we generated CMV-Cre:Optnflox/flox (Optn global knockout mice), hereafter referred to Optn−/− mice, (Fig. S2A-B). Western blot analysis confirmed that OPTN was dramatically decreased in multiple organs of Optn−/− mice, confirming global deficiency of this protein (Fig. S2C). At 3 months of age, Optn−/− mice were phenotypically normal and did not have gross anatomical abnormalities or body size differences compared with Optn+/+ littermates (Fig. S3A). Furthermore, skeletal phenotyping using micro-computerized tomography (µCT) scanning revealed that 3-month-old Optn−/− mice did not display overt PDB-like lesions in the long bones (Fig. S3B). Young Optn−/− mice also had normal bone morphometric parameters of trabecular and cortical bones compared with Optn+/+ littermates (Fig. S3C). Similarly, serum levels of alkaline phosphatase (ALP) activity and the C-terminal telopeptide (CTX-1), both biomarkers indicative of bone turnover, were not significantly elevated between Optn+/+ and Optn−/− mice at 3 months of age (Fig. S3D).

Optineurin deficiency in vivo results in PDB-like lesions in aged mice

We next performed skeletal phenotyping of the mice at 8, 12, and 16 months of age. Similar to 3-month-old mice, neither genotype displayed any skeletal abnormalities or lesions at 8 or 12 months of age (Fig. S4A-B, Table 1). At 16 months of age, however, 50% of Optn−/− mice had developed incipient, monostotic, localized osteolytic lesions in the tibiae (Fig. S4C). While there was no significant difference in serum ALP levels between Optn+/+ and Optn−/− mice at 8, 12, and 16 months, serum ALP levels had significantly increased in Optn−/− mice between 8 and 16 months, signaling an age-induced onset of PDB in the absence of Optn (Fig. S4D).

Table 1 Phenotypic analysis of skeletal abnormalities in 3-, 8-, 12-, 16-, and 22-month old Optn+/+ and Optn-/- mice.
Full size table

Because PDB primarily affects the elderly population with a peak incidence between 70 and 80 years of age [25], we next performed skeletal phenotyping of 22-month-old mice, which is equivalent to humans at 70 years old [26]. Strikingly, 100% of aged Optn−/− mice had developed polyostotic, localized osteolytic lesions in femurs, tibiae (Fig. 1a, b), calvaria (Fig. 1c), lumbar vertebra (Fig. 1d), and fibulas (Fig. 1e), which are the most commonly affected bones of PDB, and these lesions phenotypically resembled the early or osteolytic stage of PDB (Table 1). In addition, 20% of Optn−/− mice exhibited facial deformities (Fig. 1c) and 40% of Optn−/− mice had bone hypertrophic lesions (Fig. 1e), pathologies associated with the intermediate or osteoblastic stage of PDB (Table 1). Furthermore, spinal cord/nerve root compression (Fig. 1d) and pathological fractures (Fig. 1e) were also seen in the affected bones of aged Optn−/− mice (Table 1). Aged Optn−/− mice also contained decreased cortical bone components in unaffected bones (Fig. S4F) and displayed a decrease in bone mineral density (BMD) (Fig. S4G). Taken together, these results demonstrate that Optineurin deficiency results in age-dependent, localized bone lesions that are phenotypically consistent with PDB.

Fig. 1
figure 1

Aged Optn-/- mice develop Pagetic lesions. a–e Representative micro-CT 3D reconstructed images of osteolytic lesions in the distal femur (a, b), proximal tibia (b), calvaria (c), and lumbar vertebral body (d) of 22-month-old female Optn-/- mice (-/-), compared with age-matched female Optn+/+ mice (+/+). Irregular bone remodeling in the calvaria and facial deformity (c) was observed in Optn-/- mice (−/−). The affected L4 of the lumbar vertebral body displayed a chaotic trabecular structure, which causes spinal stenosis and spinal cord/nerve root compression (d). e Aged Optn-/- mice had mixed-stage lesions in the proximal tibia and fibula, and the affected fibula displayed a pathological fracture. (n = 5–7 mice per genotype)

Full size image

Pagetic lesions in Optn −/− animals are characterized by hyperosteoclastogenesis

The osteolytic lesions observed in Optn−/− mice are characteristic of disproportionate bone turnover, indicating that OPTN exerts its effect by either enhancing bone resorptive activity by osteoclasts or reducing bone-formation activity by osteoblasts. Although OPTN is induced during osteogenesis in vitro (Fig. S5A), we did not observe any difference in vivo bone formation as determined by mineral apposition rate (MAR) between young Optn+/+ and Optn−/− mice (Fig. S5B). Thus, we hypothesized that osteolytic lesions observed in aged Optn−/− mice were a result of altered osteoclast activity.

We next performed histological analysis of bones of aged Optn+/+ and Optn−/− mice. H&E and TRAP staining of the distal femur of Optn−/− mice confirmed osteolytic lesions and revealed increased bone resorption, with osteoclast-filled pits within the compact bone of Optn−/− mice, while bones from age-matched Optn+/+ mice lacked these pathological features (Fig. 2a). Cross-sectional analysis of a mixed osteoscleorotic–oste7olytic lesion in the proximal tibia of an Optn−/− mouse shows that these lesions comprise numerous osteoclasts located within the inner surface of cortical bone and adjacent trabecular surfaces (Fig. 2b, c). In addition, the mixed lesion also displayed secondary endochondrial bone formation (Fig. 2b). Focal lesions in the affected bones from Optn−/− mice contained significantly more osteoclasts per area examined compared with Optn+/+ mice (Fig. 2d), and each Optn−/− osteoclast contained more nuclei than Optn+/+ osteoclasts, indicative of increased osteoclast activity (Fig. 2e).

Fig. 2
figure 2

Increased osteoclastogenesis in aged Optn-/- mice. a H&E staining of the distal femur of 22-month-old female Optn+/+ (+/+) and Optn-/- (−/−) mice. b Staining of a mixed osteoscleorotic–osteolytic lesion in the proximal tibia of 22-month-old Optn-/- (−/−) mouse with H&E (left), TRAP (middle), and Safranin O/Fast Green (right). c, d TRAP staining in the distal femur (c) and parietal bone (d) in 22-month-old Optn+/+ (+/+) and Optn-/- (−/−) mice. Quantification of osteoclasts per mm2 of unaffected bones of Optn+/+ (+/+) and lesion sites of Optn-/- (−/−) mice (***p < 0.001, n = 5 mice). e Representative images of multi-nucleated osteoclast in the lesion of 22-month-old Optn+/+ ( + / + ) and Optn-/- (−/−) mice. Quantification of the number of nuclei per osteoclast in 22-month-old Optn+/+ (+/+) and Optn-/- (−/−) mice (****p < 0.0001, n = 60 osteoclasts). f Serum levels of alkaline phosphatase (ALP) and collagen type I C-telopeptide (CTX-1) in 22-month-old Optn+/+ (+/+) and Optn-/- (−/−) mice (*p < 0.05, ***p < 0.001, n = 4–6 per genotype)

Full size image

The gold standard for clinical diagnosis of PDB in human patients is increased serum levels of ALP, in response to increased osteoclastic bone resorption [27]. Similar to PDB patients, we found that aged Optn−/− mice had significantly increased ALP activity in their sera, compared with age-matched Optn+/+ mice, and thus recapitulate the classical diagnostic biomarker for PDB (Fig. 2f). Furthermore, serum levels of CTX-1, the bone resorption biomarker, were significantly increased in aged Optn−/− mice compared with Optn+/+ mice (Fig. 2f). However, serum levels of IL-6, a cytokine that has been associated with PDB [28], were not affected by loss of OPTN in aged animals (Fig. S5C). Taken together, our data demonstrate that Optn−/− mice exhibit serum and histological features of PDB associated with increased osteoclastogenesis, and the Optn−/− mice represent a novel and clinically relevant mouse model for PDB.

OPTN deficiency enhances in vitro osteoclastogenesis

We next explored the mechanisms by which deficiency of OPTN augments osteoclast formation. Optn−/− bone marrow osteoclast precursors generated significantly increased numbers of osteoclasts ex vivo, compared with Optn+/+ bone marrow precursors (Fig. 3a). Furthermore, Optn−/− osteoclasts differentiated ex vivo on calcium phosphate-coated plates displayed increased resorptive activity, as indicated by larger resorption pit areas compared with Optn+/+ osteoclasts (Fig. 3b). Finally, Optn−/− precursor cells expressed significantly higher levels of c-Fos and NFATc1, two important osteoclast differentiation factors, post RANKL treatment (Fig. 3c). Collectively, these data demonstrate that OPTN is a key negative regulator of osteoclastogenesis.

Fig. 3
figure 3

OPTN deficiency promotes osteoclastogenesis in vitro. a Optn+/+ and Optn−/− bone marrow-derived precursors from 8- to12-week-old mice were cultured under osteoclastogenic conditions, and osteoclast (OC) differentiation was assessed by quantification of the number of multi-nucleated TRAP-positive cells on day 3 post RANKL treatment. Representative images are shown. (***p < 0.001, n = 4 wells). b Optn+/+ and Optn−/− bone marrow-derived precursors from 8- to 12-week-old mice were cultured under osteoclastogenic conditions on Osteo Assay Surface plates, and resorptive surfaces were quantified on day 5 post RANKL treatment. Representative images are shown. (*p < 0.05, n = 9 wells). c Western blot of OPTN, c-Fos, and NFATc1 in Optn+/+ and Optn−/− OC at days 0–3 of OC differentiation. Densitometry was calculated with ImageJ (*p < 0.05, n = 3 blots). d, g Optn+/+ and Optn−/− bone marrow-derived precursors were cultured under osteoclastogenic conditions, and autophagic activity (d) and noncanonical NF-kB signaling (g) was assessed by Western blot on days 0–3 post-RANKL treatment from the sample protein lysates. In D, cells were probed for OPTN, LC3, p62, ATG5, ATG7, and β-actin. In G, the Western blots were reprobed for p100/p52 with the same OPTN and β-actin bands shown for reference. (e–f) Optn+/+ and Optn−/− bone marrow-derived precursors were cultured with RANKL for 0-60 minutes or 0-3 days, to assess the activation of canonical NF-kB signaling by Western blot. Cells were probed for OPTN (e–f), phos-p65 (e–f), Total p65 (f), Total IkB (e), and β-actin (e–f).

Full size image

As autophagy has been shown to be involved in osteoclast differentiation [29] and OPTN is an autophagy receptor, we first asked if the absence of OPTN alters autophagic induction during osteoclastogenesis. Intriguingly, Optn−/− osteoclasts had a similar transformation of the autophagy marker LC3-I to LC3-II and degradation of p62, as well as expression of ATG5 and ATG7, compared with Optn+/+ osteoclasts, suggesting that autophagy is unaffected in Optn−/− osteoclasts under osteoclastogenic conditions (Fig. 3d). Activation of the c-Fos-NFATc1 axis for osteoclastogenesis involves the NF-κB pathway [30]. As OPTN has been shown to suppress NF-κB activation [12], we examined the phosphorylation of p65, a member of the canonical NF-κB pathway, in Optn+/+ and Optn−/− osteoclast precursors after RANKL treatment. The level of phosphorylated p65 in Optn−/− osteoclast precursors was similar to that in Optn+/+ osteoclast precursors at both early time points (Fig. 3e) and late time points (Fig. 3f). Similarly, the degradation of the IκBα, a negative regulator of canonical NF-κB activation, was equivalent in Optn+/+ and Optn−/− cells after RANKL treatment (Fig. 3e). While previous studies utilized the OptnD477N/D477N model, which results in OPTN protein that is unable to bind Ly63-linked ubiquitin chains, but can still interact with other proteins via its other un-mutated domains [10], our constitutive knockout model displayed no differences in NF-κB activation during osteoclastogenesis. Collectively, these data indicate that while defects in OPTN’s ability to bind ubiquitin can result altered NF-κB activation, the absence of OPTN does not.

In addition, we also assessed noncanonical NF-κB activation by examining the processing of p100 to p52 during osteoclastogenesis. Optn−/− osteoclast precursors displayed similar levels of the cleaved p52 protein compared with Optn+/+ osteoclast precursors (Fig. 3f). These results demonstrate that OPTN regulates osteoclast differentiation independently of autophagy, canonical, or noncanonical NF-κB activation.

We next evaluated other possible signaling pathways implicated in osteoclastogenesis and observed equivalent phosphorylated levels of JNK1/2, ERK1/2, p38, and CREB (Fig. S6A, B) in Optn+/+ and Optn−/− osteoclast precursors post RANKL treatment. We further examined the expression of known OPTN-binding partners, RAB8A, Myosin VI, and Huntingtin (HTT) in Optn−/− osteoclasts relative to Optn+/+ osteoclasts, and we found that their expression levels were comparable (Fig. S6C). Therefore, OPTN deficiency does not affect classical osteoclastogenic pathways in osteoclast precursors during RANKL treatment.

OPTN is required for both the production of type I IFN and signaling via the IFNα/βR

To investigate potential transcriptional differences between Optn+/+ and Optn−/− osteoclasts, genome-wide transcriptional profiling was performed at day 3 of osteoclast differentiation ex vivo. Microarray revealed that many critical pathways were upregulated in Optn−/− osteoclasts, including genes regulating cytoskeletal rearrangements, cell cycle, cytokine and chemokine expression, and osteoclast differentiation and function, compared with Optn+/+ osteoclasts (Fig. S6D). Consistent with our protein results, the transcription of Rab8a/b, Myosin VI, and Htt was equivalent between Optn+/+ and Optn−/− osteoclasts (Fig. S6D). In addition, there was no difference in transcription of other components known to interact with OPTN, such as Ripk1, Cyld, Sqstm1, and Hace1, between Optn+/+ and Optn−/− osteoclasts (Fig. S6D).

We did, however, observe a significant difference in the expression of genes associated with the type I IFN signature. Genes such as Ifitm10, Ifitm5, Ifna1, Irf7, and Ifnar2 were dramatically reduced in Optn−/− osteoclasts compared with Optn+/+ osteoclasts (Fig. 4a). Conversely, expression of Socs3, a negative regulator of type I IFN signaling [31], was significantly upregulated in Optn−/− osteoclasts compared with Optn+/+ osteoclasts (Fig. 4a). Engagement of the IFNα/βR with type I IFN (typically IFNβ in myeloid-derived cells, such as osteoclasts) results in the phosphorylation of STAT1 [32]. In response to RANKL treatment, Optn−/− osteoclast precursors displayed lower levels of the total and phosphorylated STAT1 and higher levels of SOCS3, suggesting that Optn deficiency confers a reduced activation of the type I IFN response in response to RANKL (Fig. 4b).

Fig. 4
figure 4

OPTN deficiency affects RANKL-induced IFNβ production and type I IFN activation. a Microarray analysis of Optn+/+ and Optn−/− OC on day 2 post RANKL treatment. Heatmap representing color-coded expression levels of differentially expressed genes in the type I IFN signature (log2 values) of biological triplicate samples of Optn+/+ and Optn−/− OC. b Western blot of OPTN, phos-STAT1, Total STAT1, and SOCS3 expression in Optn+/+ and Optn−/− OC precursors treated with RANKL at 4 and 6 h. c The mRNA level of Ifnb1 in OC precursors treated with RANKL at 4 h was determined using qPCR. (*p < 0.05, n = 3 wells). d Optn+/+ and Optn−/− OC precursors from 8- to12-week-old mice were treated with IFNβ and assessed for type I IFN signaling via western blot for phos-STAT1, phos-STAT2, Total STAT1, and Total STAT2. Densitometry was calculated with ImageJ (*p < 0.05, n = 3 blots)

Full size image

We next explored whether this reduced type I IFN signaling in Optn−/− osteoclasts resulted from decreased IFNβ production or defective IFNα/βR signaling, or both. We first observed that Optn−/− osteoclast precursors generated significantly lower levels of Ifnb1 in response to RANKL treatment (Fig. 4c), compared with Optn+/+ precursors. RANKL-induced IFNβ production does not require canonical TBK1 and IRF3 machinery [33], and we indeed observed that RANKL failed to activate IRF3 in osteoclast precursors and induced a comparable level of TBK1 phosphorylation, which is required for NF-κB activities, between Optn+/+ and Optn−/− precursors (Fig. S6E). Next, we examined signaling downstream of IFNβ engagement of IFNα/βR during IFNβ treatment. Strikingly, Optn−/− osteoclast precursors displayed defective IFNα/βR signaling in response to IFNβ stimulation, as evidenced by decreased total and phosphorylated STAT1/2 (Fig. 4d). As expected, IFNα/βR-deficient (Ifnar−/−) osteoclast precursors displayed a completely abolished type I IFN signaling in response to RANKL or IFNβ (Fig. S6F-G). Taken together, these results demonstrate that OPTN is critical for both the production of type I IFN and efficient signaling via the IFNα/βR pathway.

Type I IFN signaling can also promote RIPK3-mediated necroptosis [24], therefore we next examined if defects in IFNβ production by Optn−/− osteoclasts conferred a survival advantage compared with Optn+/+ osteoclasts. Treatment of Optn−/− osteoclast precursors with RANKL, a member of the TNF superfamily [34], resulted in significantly increased survival at both 24 and 48 h post RANKL treatment, compared with Optn+/+ precursors (Fig. 5a). In order to determine if exogenous IFNβ treatment during osteoclastogenesis could modulate cell death in Optn−/− osteoclast precursors, we differentiated Optn+/+ and Optn−/− precursors into osteoclasts with RANKL in the absence or presence of IFNβ. The survival advantage observed in Optn−/− osteoclast precursors was rescued at 48 h post RANKL/IFNβ treatment, compared with Optn+/+ precursors (Fig. 5b, S7A).

Fig. 5
figure 5

Recombinant IFNβ can rescue hyperosteoclastogenesis observed during OPTN deficiency. a Optn+/+ and Optn−/− OC precursors from 8- to 12-week-old mice treated with RANKL, and cells were harvested at 0, 24, and 48 h post RANKL treatment and analyzed by flow cytometry for cell death by Annexin V and Zombie-Red. %Dead cells calculated as percentage Annexin V+ Zombie-Red+ of CD45+ cells (*p < 0.05, **p < 0.01, n = 6 wells). b Optn+/+ and Optn−/− OC precursors treated with RANKL with or without IFNβ, and cells were harvested at 24 and 48 h post RANKL/ IFNβ treatment and analyzed by flow cytometry for cell death by Annexin V and Zombie-Red. %Dead cells calculated as percentage Annexin V+ Zombie-Red+ of CD45+ cells (*p < 0.05, ***p < 0.001, n = 3 wells). c TRAP staining of Optn+/+ and Optn−/− osteoclast precursors treated with RANKL with or without IFNβ for 3 days to assess OC formation (**p < 0.01, n = 4 wells). h TRAP staining of osteoclast precursors under osteoclastogenic conditions with control or IFNα/βR blocking (αIFNAR) antibody for 3 days (*p < 0.05, *p < 0.01, ***p < 0.001, n = 4 wells)

Full size image

As IFNβ is also an established negative regulator of osteoclastogenesis [33], we next asked if exogenous IFNβ treatment during osteoclastogenesis could rescue the hyperdifferentiation observed in Optn−/− osteoclasts. Recombinant IFNβ was sufficient to inhibit the number of osteoclasts generated by Optn−/− precursors, as evidenced by TRAP staining (Fig. 5c). Furthermore, antibody-mediated inhibition of IFNα/βR signaling (αIFNAR) resulted in a significant increase in osteoclastogenesis of both Optn+/+ and Optn−/− precursors (Fig. 5c). Importantly, IFNα/βR signaling blockade had a larger effect on Optn+/+ precursors, essentially normalizing the Optn−/− osteoclastogenic phenotype (Fig. 5d). This suggests that the excess OC differentiation of Optn−/− cells is mediated by a deficit in native IFN signaling.

Collectively, we have described the Optn−/− mouse as a novel and clinically relevant model for PDB in vivo and have demonstrated that OPTN is required to maintain homeostatic levels of osteoclastogenesis in vivo and ex vivo, the absence of which results in hyperactive osteoclast differentiation and activity. In addition, we demonstrate that OPTN functions on two levels to regulate osteoclast differentiation—the production of type I IFN and signaling of type I IFN through the IFNα/βR (Fig. 6), and the net effect of this defect is increased osteoclastogenesis, as well as defects in cell death, opening the door for possible interferon intervention for the treatment of PDB.

Fig. 6
figure 6

Schematic of the dual role of OPTN in mediating negative regulation of osteoclastogenesis. During RANKL-mediated osteoclastogenesis, OPTN is required for optimal IFNβ production, which is critical for inhibition of c-Fos, promotion of the IFNα/βR signaling cascade, and induction of cell death

Full size image

Discussion

Recent advances in the underlying genetic influences on the onset and severity of PDB have shed light onto the root causes of this disease and opened up avenues for deeper biological understanding. Here, we performed extensive in vivo skeleton characterization of the Optn-deficient mouse, which exhibits full penetrance and multiple key clinical manifestations of PDB in an age-dependent manner, in contrast to the OptnD477N/D477N mutant mice, in which only ~10% of 15–18-month-old mice develop PDB lesions [10, 35]. Therefore, this Optn-deficient mouse represents a novel and useful mouse model to study PDB pathogenesis, and potentially, other pathologies associated with OPTN dysfunction, such as primary open-angle glaucoma (POAG) and amyotrophic lateral sclerosis (ALS).

While multiple studies have explored the molecular mechanisms by which OPTN dysfunction underlies disease pathogenesis, these studies have mainly focused on POAG and ALS [19]. Studies have demonstrated that the POAG-associated OPTN mutant (E50K) could selectively induce ROS-associated cell death of retinal ganglion cells in vitro [17], and OPTN E50K knock-in mice exhibit loss of retinal ganglion cells in vivo [36]. Furthermore, in response to TNFα signaling, OPTN deficiency leads to persistent RIPK1 signaling and uncontrolled necroinflammation in the spinal cord, resulting in axonal degeneration and ALS [18]. However, in contrast to the pathophysiology of Optn-dependent POAG and ALS, which is characterized by the loss of ganglion cells or neurons [18, 36], PDB is highlighted by a “gain” of osteoclasts [37], suggesting that the molecular mechanisms implicated in the pathogenesis of PDB is dissimilar to POAG and ALS. While Optn-deficient osteoclasts did not display any difference in RIPK1 expression at either mRNA and protein levels, it is possible that activation of necroptotic machinery is altered in Optn-deficient osteoclasts. A recent study has identified MK2 has a critical negative regulator of RIPK1 activity, wherein MK2 phosphorylates RIPK1 at Ser321 to inhibit its ability to induce RIPK1-dependent apoptosis and necroptosis [38]. Interestingly, MK2−/− mice display a significant decrease in osteoclastogenesis, which could suggest that an increase in functionally active RIPK1 results in decreased osteoclast survival [39]. The role that OPTN plays in modulating the MK2-RIPK1 axis in response to type I IFN remains to be elucidated.

Although OPTN has been shown to be a potent inhibitor of NF-κB activity [11], our data showed that Optn-deficient osteoclasts do not have altered NF-κB activation ex vivo, consistent with recent reports [15, 16, 35]. However, failure to detect alterations in NF-κB activity in our study may be due to experimental limitations, as the previous study found increased NF-κB responses in the OptnD477N/D477N model using the NF-κB luciferase reporter assay [10], which may be more sensitive compared with western blotting. Indeed, we did observe a significant and two-pronged defect in the execution of the negative feedback loop mediated by IFNβ. In the absence of OPTN, both production of IFNβ and signaling downstream of the IFNα/βR are impaired. While previous studies have described a role for OPTN in IFNβ production during osteoclastogenesis [10], here we demonstrate for the first time that OPTN is also required for optimal signaling downstream of IFNβ engagement of the IFNα/βR (Fig. 6). This defect manifests itself in a failure to upregulate anti-osteoclastogenic factors, as well as promote cell survival. Thus, OPTN limits osteoclastogenesis at two distinct nodes, and is a critical factor in restricting uncontrolled bone resorption.

Moreover, a defect in type I IFN production could in fact account for the defect in IFNα/βR signaling, as type I IFN upregulates expression of key components of the IFNα/βR signaling pathway [40]. It is possible that Optn-deficient osteoclasts fail to produce amounts of type I IFN necessary to properly establish an IFNα/βR signaling platform, as indicated by decreased levels of the total STAT1 and STAT2 in Optn-deficient precursors. In addition, supplementation with recombinant IFNβ rescues the hyperosteoclastogenic phenotype in Optn-deficient osteoclasts as it upregulates components of the IFNα/βR signaling pathway, such as STAT1 and STAT2, as well as inhibit c-Fos-mediated osteoclast differentiation. Furthermore, blocking IFNα/βR signaling with anti-IFNAR Ab had a more profound effect on Optn+/+ osteoclasts, suggesting that defects in IFNα/βR signaling prevented antibody blockade from being fully effective in Optn−/− osteoclasts.

Interestingly, while Optn-deficient osteoclasts exhibit a defective type I IFN pathway, the measles virus, which elicits a robust type I IFN response, is strongly correlated to PDB development [41]. Indeed, this association could support why individuals harboring the OPTN variant are prone to develop PDB, as their intrinsic defects in the type I IFN pathway permit them to be more susceptible to measles virus infection. Because of Optineurin’s dual function in maintaining bone homeostasis and antiviral type I IFN signaling, IFNβ could be a novel pharmacotherapy for PDB, as it provides restoration of dysregulated bone resorption in addition to antiviral functions.

Taken together, our results allow us to propose a possible mechanistic model of OPTN-associated PDB pathogenesis. At the molecular level, OPTN plays a dual role in the negative regulation of osteoclastogenesis—the production of type I IFN and signaling of the type I IFN through the IFNα/βR to maintain bone homeostasis. Deficiency in OPTN fails to execute the proper inhibitory functions of type I IFN signaling, leading to hyperactivation of osteoclastogenesis and development of osteolytic lesions observed in PDB.