The p62 P392L Mutation Linked to Paget’s Disease Induces Activation of Human Osteoclasts

Abstract

Mutations of the gene encoding p62/SQSTM1 have been described in Paget’s disease of bone (PDB), identifying p62 as an important player in osteoclast signaling. We investigated the phenotype of osteoclasts differentiated from peripheral blood monocytes obtained from healthy donors or PDB patients, all genotyped for the presence of a mutation in the p62 ubiquitin-associated domain. The cohort included PDB patients carrying or not the p62 P392L mutation and healthy donors carrying or not this mutation. Osteoclasts from PDB patients were more numerous, contained more nuclei, were more resistant to apoptosis, and had a greater ability to resorb bone than their normal counterparts, regardless of whether the p62 mutation was present or not. A strong increase in p62 expression was observed in PDB osteoclasts. The presence of the p62^P392L gene in cells from healthy carriers conferred a unique, intermediate osteoclast phenotype. In addition, we report that two survival-promoting kinases, protein kinase Cζ and phosphoinositide-dependent protein kinase 1, were associated with p62 in response to receptor activator of NF-κB ligand (RANKL) stimulation in controls and before RANKL was added in PDB osteoclasts. In transfected osteoclasts derived from cord blood monocytes, the p62 P392L mutation contributed to increased activation of kinases protein kinase Cζ/λ and phosphoinositide-dependent protein kinase 1, along with basal activation of NF-κB, independently of RANKL stimulation. These findings clearly indicate that the overexpression of p62 in PDB patients induces important shifts in the pathways activated by RANKL and up-regulates osteoclast functions. Moreover, the most-commonly reported p62 mutation, P392L, certainly contributes to the overactive state of osteoclasts in PDB.

Over-expression of p62 and a p62 P392L mutation induce important shifts in the pathways activated by RANKL and up-regulate osteoclast functions in Paget’s disease of bone.

Paget’s disease of bone (PDB) is the second most common skeletal disorder, after osteoporosis, and affects 1–2% of adults over 50. PDB is characterized by focal and disorganized increases in bone turnover, with an early excessively lytic phase making osteoclasts an important player in the development of PDB (1). In addition to their increased number and excessive bone-resorbing activity, in PDB, osteoclasts are phenotypically abnormal, being larger and containing more nuclei than normal osteoclasts (2). Although environmental factors, such as a viral infection, may contribute to the development of PDB, a strong genetic component has been demonstrated. Mutations in the gene encoding sequestosome 1 (SQSTM1), also known as p62, have been detected in a large proportion of PDB patients, up to 50% of familial Paget cases, with a high penetrance of about 80% in patients over 60 (3, 4). All the mutations identified to date in PDB patients cluster within or around the ubiquitin-associated (UBA) domain of the p62 protein (4). The P392L mutation, as well as the other deletions or substitutions that have been described, affect the interactions of the UBA domain with multiubiquitin chains, suggesting that a loss of ubiquitin chain binding by p62 is probably important for the development of PDB (5).

P62 mediates various cell functions, including controlling nuclear factor-κB (NF-κB) signaling, protein trafficking, and gene transcription (6). It has been described as a scaffolding protein that interacts with TNF receptor-associated factor 6 (TRAF6) and with atypical PKCζ after activation by receptor activator of NF-κB ligand (RANKL) (7), the key regulator of osteoclast differentiation, survival, and activity, which acts mainly by activating NF-κB (8). The crucial role of p62 during RANKL-induced osteoclastogenesis was demonstrated in p62−/− knockout mice, in which RANKL-induced osteoclast formation and NFATc1 expression were impaired (7). The role of the most common p62 mutation, P392L, has been studied in in vitro studies, using osteoclast precursors derived from peripheral blood from three PDB patients carrying the p62^P392L gene, and from bone marrow cells from normal subjects transfected with the p62^wt or p62^P392L gene. These osteoclast precursors were hyperresponsive to osteoclastogenetic factors, such as RANKL and TNFα, and the p62^P392L-transfected cells had an increased ability to resorb bone (9). In vivo, transgenic mice with targeted expression of the human p62^P392L gene in the osteoclast lineage developed osteopenia, with an elevated osteoclast perimeter, although no Paget-related bone lesions were observed. Once again, osteoclast precursors were hyperresponsive to RANKL and TNFα, and they demonstrated increased proliferation rates, but no change in either nuclei numbers, or apoptosis rates (9). Finally, overexpression of the p62^P392L mutant in human embryonic kidney 293 or Cos-1 cells increased basal and RANKL-induced NF-κB activation, compared with overexpression of the p62 wild-type mutant (10). All these studies demonstrate that the p62 P392L mutation affects the osteoclast phenotype, inducing overactive osteoclasts that display some characteristics of the Paget osteoclast phenotype. Most of these data have been obtained from murine models or transfected human cells; however, the mechanisms involved in such p62-driven mis-regulation are not fully understood in the human healthy or PDB osteoclasts.

Our aim was to explore the effects of the p62 P392L mutation on the osteoclast phenotype and on RANKL-induced intracellular signaling in a large cohort of PDB patients, all of whom had been genotyped for the presence of mutations in the p62 UBA domain, and to compare them with cells from healthy donors (HDs) with no p62 mutation, or healthy p62-mutation carriers. To further identify the mutation-related effects, we also overexpressed p62^wt and p62^P392L in osteoclasts derived from cord blood monocytes, thus precluding the pagetic background that could interfere with investigation of the resulting signaling events, because the healthy carriers were relatives of PDB patients.

Results

Characterization of the pagetic osteoclast phenotype

Mature osteoclasts were obtained by long-term cultures of peripheral blood mononuclear cells (PBMCs) from four groups of patients (HD^wt, HD^P392L, PDB^wt, and PDB^P392L) in the presence of RANKL and macrophage colony-stimulating factor (M-CSF). The percentage of multinucleated cells (MNCs) obtained at the end of the cultures was 8.9 ± 1% in the PBMCs from HD^wt, 21.6 ± 1.3% in those from HD^P392L (P < 0.001 vs. HD^wt), 37.9 ± 1% in PDB^wt, and 35.2 ± 1.2% in PDB^P392L (P < 0.001 each group vs. HD^wt and P < 0.001 each group vs. HD^P392L) (Fig. 1A). The number of nuclei per MNC was 4.2 ± 1.5 in PBMCs from HD^wt, 14.7 ± 6 in those from HD^P392L (P < 0.01 vs. HD^wt), 26.4 ± 15 in PDB^wt, and 22.3 ± 11 in PDB^P392L (P < 0.001 vs. HD^wt, P < 0.05 vs. HD^P392L) (Fig. 1B). Thus, the MNCs obtained in PBMC cultures from all the groups of PDB patients were significantly more numerous and contained more nuclei than those from healthy donors. A similar pattern was also observed in HD^P392L cells compared with HD^wt, although the numbers of MNCs and of nuclei per MNC remained lower than those of PDB patients.

Fig. 1.

Osteoclast formation in PBMC cultures. Human PBMCs were differentiated for 21 d with RANKL and M-CSF. At the end of the culture period, cells were stained with Evan’s blue and DAPI. A, The percentage of MNCs per total cell number was evaluated. Cells containing three or more nuclei were scored as MNCs. B, The number of nuclei per MNC was counted in 20 MNCs per well. C, Bone resorption was assessed at the end of the differentiation period by toluidine blue staining of the bone slices. The resorbed surface area was measured using BioQuant Nova Image analysis software and reported graphically. Results of three to five independent experiments are expressed as mean ± sem; *, P < 0.05; **, P < 0.01; ***, P < 0.001 vs. HD^wt; ^#, P < 0.05; ^##, P < 0.01; ^###, P < 0.001 vs. HD^P392L; °, P < 0.05; °°°, P < 0.001 RANKL 25 ng/ml vs. 100 ng/ml. HD^wt: healthy donors − no mutation in p62 gene (n = 3); HD^P392L: healthy donors − p62^P392L mutation (n = 3); PDB^wt: PDB − no p62 mutation (n = 4); PDB^P392L: PDB − with p62^P392L mutation (n = 5). OC, Osteoclast.

Bone resorption was quantified at the end of the 3-wk period to evaluate the terminal differentiation and activity of the osteoclasts. As pagetic osteoclasts are known to be more sensitive to bone-resorbing factors than normal osteoclasts, two concentrations of RANKL (25 and 100 ng/ml) were tested in these cultures. PBMCs from all four groups of patients (HD^wt, HD^P392L, PDB^wt, and PDB^P392L) were able to differentiate in bone-resorbing cells. As shown in Fig. 1C, when the cells were differentiated in the presence of low doses of RANKL (25 ng/ml), the percentage of resorbed bone area was 6.7 ± 0.9% in cultures from HD^wt, and 7.0 ± 3% in those from HD^P392L cells. The percentage of resorbed area was significantly higher in pagetic cell cultures, with 21.9 ± 7.3% in PDB^wt, and 19.0 ± 2.8% in PDB^P392L (P < 0.001 each PDB group vs. HD^wt). Moreover, using a higher dose of RANKL (100 ng/ml) resulted in an increase in the resorbed area in cultures from HD^wt (13.3 ± 1.5%), which was much higher in all other groups, with 31.0 ± 6.3% of resorbed area in HD^P392L, 29.8 ± 0.2% in PDB^wt, and 28.9 ± 11.8% in PDB^P392L (P < 0.001 each group vs. HD^wt). Thus, with both high and low concentrations of RANKL, the resorbed bone area was significantly higher in osteoclast cultures from PDB patients than in HD^wt. However, although cells from healthy carriers of p62^P392L were comparable to HD^wt when differentiated with low doses of RANKL, they became as effective as PDB osteoclasts in resorbing bone when cultured with higher doses of RANKL.

Pagetic osteoclasts are more resistant to apoptotic stimuli

We then studied osteoclast apoptosis in PBMC cultures. Fully matured cells were deprived of survival factors M-CSF and RANKL for 24 h. The percentage of apoptotic MNCs was 29.4 ± 2.4% in PBMCs from HD^wt and was significantly lower in cells from HD^P392L (14.3 ± 0.9; P < 0.001 vs. HD^wt). This percentage was further decreased in pagetic cells with 7.54 ± 0.6% in PDB^wt and 7.75 ± 0.6% in PDB^P392L (P < 0.001 for each of the PDB groups vs. HD^wt; and P < 0.01 for each of the PDB groups vs. HD^P392L) (Fig. 2A).

Fig. 2.

Induction of apoptosis in pagetic osteoclasts. At the end of the PBMC cultures, 24 h after the M-CSF and RANKL had been removed, apoptosis was detected by DAPI staining (A), or after adding various concentrations of hrTRAIL (10–400 ng/ml), a Fas-activating antibody (50–800 ng/ml); hrTGFβ (0.1–4 ng/ml) or M-CSF (1–50 ng/ml) where appropriate 24 h before apoptosis was assessed (B). Results of three independent experiments are expressed as the percentage of apoptotic MNCs, mean ± sem. The significance was indicated as follows: *, P < 0.05; **, P < 0.01; ***, P < 0.001 vs. untreated controls; significance of the dose-response trend-line: ^#, P < 0.05; ^##, P < 0.01; ^###, P < 0.001 vs. HD^wt; ^¤, P < 0.05; ^¤¤, P < 0.01 HD^P392L vs. PDB^P392L/wt; HD^wt: healthy donors − no mutation in p62 gene (n = 3); HD^P392L: healthy donors − p62^P392L mutation (n = 3), PDB^wt: PDB − no p62 mutation (n = 5); PDB^P392L: PDB − p62^P392L mutation (n = 6).

In additional experiments, increasing concentrations of soluble TNF-related apoptosis-inducing ligand (TRAIL) (10–400 ng/ml), of a Fas-activating antibody (50–800 ng/ml), and of TGF-β1 (0.1–4 ng/ml), were added 24 h before apoptosis evaluation. These factors have previously been shown to induce osteoclast apoptosis in human models (11). We also tested increasing concentrations of M-CSF (1–50 ng/ml), a survival factor for osteoclasts. The results shown in Fig. 2B revealed that TRAIL induced osteoclast apoptosis in a dose-dependent manner. With the highest concentration of TRAIL (400 ng/ml), the percentage of apoptotic MNCs was 54.5 ± 9.1% in HD^wt, but only 40.5 ± 0.5% in HD^P392L, 19.1 ± 1.3% in PDB^wt, and 19.4 ± 3.7% in PDB^P392L (P < 0.001 for each group vs. HD^wt). Analysis of the shape of the curve revealed that the differences observed between groups were not only due to the reduced base level of apoptosis, but that increasing doses of TRAIL did not induce apoptosis as efficiently in the PDB groups as in the HD^wt (P < 0.001). The same observation was made for HD^P392L, which responded less efficiently to TRAIL-induced death signals than HD^wt (P < 0.001), but this difference was less marked than for PDB^wt or PDB^P392L osteoclasts (P < 0.01). A similar reduction in apoptosis induction was observed with Fas-activating antibodies and TGF-β1. Again, in addition to the lower base level of apoptosis, Fas and TGF-β1 had a reduced capacity to induce apoptosis in a dose-dependent manner in both PDB groups (P < 0.001 vs. HD^wt). The dose-response curve observed in HD^P392L cultures was intermediate, with a lower rate of death induction than HD^wt (P < 0.05 for Fas and P < 0.001 for TGF-β1), and a higher response than the PDB groups (P < 0.05 for both stimuli). In contrast, 50 ng/ml of M-CSF reduced deprivation-induced apoptosis, and the percentage of apoptotic osteoclasts was 13.8 ± 5.4% in HD^wt (P < 0.01 vs. nontreated cells). Apoptosis inhibition was even more efficient in mutated and pagetic cells, with 4.5 ± 0.5% in HD^P392L, 6.1 ± 1.1% in PDB^wt, and 5.8 ± 1.7% in PDB^P392L (P < 0.001 each vs. HD^wt). Thus, under all these experimental conditions, the rate of deprivation-induced or death factor-induced apoptosis observed in HD^P392L was intermediate between that of HD^wt and PDB osteoclasts. There was no significant difference between the two types of PDB osteoclasts. Finally, in HD^P392L as well as in PDB osteoclasts, the basal level of apoptosis was lower than that observed in HD^wt and decreased slightly in the presence of M-CSF.

Increased P62 expression in pagetic osteoclasts

To clarify the molecular basis of the strikingly different cell phenotypes found in cultures from PDB patients and healthy donors, and its relationship to the P392L mutation, we sought to evaluate p62 expression in the osteoclasts. Western blots showed that p62 was significantly more expressed in PDB patients than in healthy donors, regardless of whether the P392L mutation was present or not (Fig. 3, A and B). The immunofluorescence studies revealed that the pattern of expression was predominantly cytoplasmic in healthy donors, but exhibited clusters of denser protein expression at the membrane level in both PDB^wt and PDB^P392L (Fig. 3C, arrows). Thus, high p62 expression levels seem to depend mainly on the presence of PDB and are not greatly influenced by the presence of the mutation.

Fig. 3.

p62 protein expression in pagetic osteoclasts. Human PBMCs were differentiated for 21 d with RANKL and M-CSF. A, Whole-cell lysates of mature osteoclast cultures were subjected to SDS-PAGE followed by a Western blot with an antibody directed against p62. A control blot was done on the same membrane with an antibody directed against β-actin. B, ODs were measured with ImageJ software, and the ratio of p62 vs. actin was computed. Analyses are reported as mean ratio ± sem. *, P < 0.05 vs. HD^wt. The data shown are representative of three independent experiments conducted with different combinations of samples from various donors. C, Immunofluorescence studies were performed with an antibody directed against p62 and a secondary Alexa-647 antibody (red). After washing, cells were counterstained with DAPI and mounted. Arrows, show the membrane expression of protein p62. Images presented here were representative of cells seen in five independent experiments. HD^wt: healthy donors − no mutation in p62 gene (n = 3); HD^P392L: healthy donors − p62^P392L mutation (n = 3); PDB^wt: PDB − no p62 mutation (n = 5); PDB^P392L: PDB − with p62^P392L mutation (n = 5). NS, Nonspecific staining.

p62 transfected human osteoclasts

Expression of p62 in transfected osteoclasts

From the foregoing observations the possibility had emerged that, in addition to the mutation in p62, other PDB-related features might influence the impact on the phenotype of pagetic osteoclasts, or their resistance to apoptosis, that could overcome the mutation effects in all observations. Indeed, p62 expression appeared to be increased to a similar extent in both PDB^wt and PDB^P392L, as were osteoclast formation, survival, and activity. The intermediate phenotype observed in HD^P392L could suggest that the p62 P392L mutation may affect these processes, but then again the HD^P392L patients may have other predisposition factors in addition to the p62 mutation, because they were relatives of PDB patients. To identify the specific effects of the p62 P392L mutation, we used osteoclasts derived from cord blood monocytes (CBMs), transfected with constructs containing either an empty vector (EV), a plasmid containing p62^wt, or a plasmid containing p62^P392L, all of which were fused to a green fluorescent tag to make it possible to visualize the protein.

Western blots shown in Fig. 4, A and B, confirmed that, compared with nontransfected cells or cells transfected with an EV, the cells transfected with either p62^wt or p62^P392L displayed significantly greater expression of p62 (P < 0.01 and P < 0.001, respectively). The p62 protein was detected by immunofluorescence in tiny dot-like structures throughout the cytoplasm, and at the submembrane level in nontransfected osteoclasts. In cells transfected with plasmids containing p62^wt and p62^P392L, overexpression was detected in clusters of protein accumulated in the cytoplasm and membrane (Fig. 4C, arrows). The transfection process itself did not affect the general morphology of human osteoclasts, as shown by green fluorescent protein (GFP) expression in cells transfected with an empty pEGFP vector (Fig. 4C).

Fig. 4.

p62 protein expression in transfected osteoclasts. CBMs were differentiated for 3 wk in the presence of RANKL and M-CSF. Where appropriate, the cells were transfected on d 17 of culture with a pEGFP-C2 plasmid containing p62^wt, p62^P392L, or an empty pEGFP vector (EV), or were not transfected (NT). A, Cells were lysed 48 h after transfection, and whole-cell lysates were subjected to SDS-PAGE followed by a Western blot with an antibody directed against p62. A control blot was done on the same membrane with an antibody directed against β-actin. B, ODs were measured with ImageJ software, and the ratio of p62 vs. actin was computed. Analyses are reported as mean ratio ± sem. **, P < 0.01; ***, P < 0.001 vs. NT. The data shown are representative of four independent experiments. C, GFP-fused p62 was detected in transfected cells in green. As a control, endogenous levels of p62 were assessed in nontransfected osteoclasts using an anti-p62 antibody and a secondary antigoat Alex-546 antibody. After washing, the cells were counterstained with DAPI and mounted. Arrows, p62 protein was detected in the cytoplasm and membrane. The images presented here were representative of cells seen in six independent experiments. NS, Nonspecific staining.

P62^P392L increases osteoclast formation and the number of nuclei

In contrast to what we observed in cells from PDB patients, the overexpression of p62^wt alone, in a nonpagetic context, neither increased the formation of multinucleated cells nor increased the number of nuclei present in differentiated osteoclasts (Fig. 5, A and B). In contrast, overexpression of the P392L mutated p62 led to the formation of more osteoclasts (P < 0.01), which contained more nuclei (P < 0.05) than nontransfected cells or cells overexpressing wild-type p62.

Fig. 5.

Osteoclast phenotype in p62 transfected cells. CBMs were differentiated for 3 wk in the presence of RANKL and M-CSF. Where appropriate, the cells were transfected on d 17 of culture with a pEGFP-C2 plasmid containing p62^wt, p62^P392L, or an empty pEGFP vector (EV), or were not transfected (NT). A, At the end of the culture period, the number of multinucleated cells within the transfected population was evaluated. Among GFP-expressing cells, the cells containing more than three nuclei were scored as MNCs. B, The number of nuclei per MNC was counted. C, Bone resorption was assessed by toluidine blue staining. The resorbed surface area was measured using Simple PCI software, corrected for the number of MNCs under the same conditions and is reported graphically. Results are expressed as the percentage of the resorbed area. D, In apoptosis experiments, cells were deprived of RANKL and M-CSF for 24 h before apoptosis was evaluated. Apoptosis was detected by DAPI staining, which made it possible to discriminate between multinucleated transfected cells and other cells. Results are expressed as the percentage of apoptotic MNCs (or apoptotic tagged-MNCs where appropriate) over total MNCs. Results of four to six independent experiments are expressed as mean ± sem (*, P < 0.05; **, P < 0.01; ***, P < 0.001 vs. EV; ^#, P < 0.05 p62^P392L vs. p62^wt; °, P < 0.05; °°, P < 0.01 RANKL 25 vs. 100 ng/ml). OC, Osteoclast.

In bone resorption studies, because major differences were seen between cultures with regard to cell numbers, we corrected the resorbed area using the MNC counts. As shown in Fig. 5C, the corrected resorbed areas were similar for nontransfected and EV-transfected cells (28 ± 3% and 25 ± 2.8%, respectively; RANKL, 100 ng/ml). In contrast, the expression of p62^wt or p62^P392L significantly increased the corrected resorbed area (38 ± 2.1%, P < 0.05; and 46 ± 1.5%, P < 0.001 vs. EV, respectively; RANKL 100 ng/ml). Finally, in transfected cells, bone resorption was increased to a similar extent in the presence of both RANKL concentrations (25 and 100 ng/ml).

To determine p62 mutation effects on cell survival, apoptosis was evaluated in the transfected and nontransfected cells after withdrawing survival factors for 24 h. As shown in Fig. 5D, approximately half of the nontransfected osteoclasts (NT) and EV-transfected osteoclasts (50 ± 7.3% and 51 ± 3.16%) displayed condensed nuclei and the typical appearance of apoptotic cells. Apoptosis was significantly reduced to 25.3 ± 4.5% in osteoclasts overexpressing p62^wt (P ≤ 0.05 p62^wt vs. EV), and to 15.3 ± 6.9% in osteoclasts transfected with p62^P392L (P < 0.01 vs. EV). These results indicate that the P392L mutation in p62 promotes the formation of more multinucleated cells that contain more nuclei per cell than nonmutated p62 transfected cells. In addition, the overexpression of p62 (mutated or wild type) itself increases osteoclast activity and survival.

RANKL-induced signaling is increased in pagetic osteoclasts

The best described function of p62 is its role as a scaffold, and in rodent models p62 is involved in the formation of a signaling complex along with TRAF6 and activated PKCζ, leading to NF-κB activation in response to RANKL stimulation, the key regulator of osteoclast survival and differentiation. To further investigate the mechanisms responsible for enhanced osteoclast formation and activity in pagetic osteoclasts, and the contribution of the p62 P392L mutation, we investigated RANKL signaling, and particularly the association between p62 and PKCζ in response to RANKL stimulation. Immunoprecipitation studies were conducted in lysates from fully matured cells with or without RANKL stimulation, using anti-p62 antibodies to precipitate p62^wt or anti-PKCζ because anti-p62 antibodies, directed against the UBA domain, were not able to efficiently precipitate the mutated p62 (Fig. 6). In nonstimulated HD^wt cells, only low levels of PKCζ were detected in p62 precipitates. However, PKCζ detection increased significantly in p62 precipitates after RANKL was added to the culture medium. This pattern of activation was markedly changed in PDB^wt cells, which displayed a considerably higher level of p62-associated PKCζ before RANKL stimulation (P < 0.001 vs. HD^wt), and this did not further increase after RANKL addition (Fig. 6A). As in PDB^wt cells, in both HD^P392L and PDB^P392L cells, p62 was present in PKCζ precipitates before RANKL stimulation (Fig. 6B).

Fig. 6.

RANKL-induced PKCζ and p62 interactions in pagetic cells. Human PBMCs were differentiated for 21 d with RANKL and M-CSF. At the end of the culture, the cells were stimulated with RANKL (100 ng/ml) for 0, 30, 60, and 90 min. A, For HD^wt and PDB^wt, immunoprecipitations (IPs) were conducted with an anti-p62 antibody in whole-cell lysates. Because anti-p62 antibodies, directed against the UBA domain, were not able to efficiently precipitate the mutated p62, IPs were conducted with anti-PKCζ (panel B) or anti-P-PDK1 (panel C), for HD^P392L and PDB^P392L. Western blots (WB) using anti-PKCζ, anti-P-PDK1, and anti-p62 antibodies are shown. ODs were measured with ImageJ software, and the ratio of PKCζ or P-PDK1 to p62 was computed. Analyses are reported as mean ratio ± sem. ***, P < 0.001 vs. nontreated; ^###, P < 0.001 vs. HD^wt. The data are representative of three independent experiments. HD^wt: healthy donors − no mutation in p62 gene, HD^P392L: healthy donors − p62^P392L mutation, PDB^wt: PDB − no p62 mutation, PDB^P392L: PDB − with p62^P392L mutation.

To investigate the complex signaling involving p62 and PKCζ in greater detail, we hypothesized that activated phosphoinositide-dependent protein kinase (PDK1) [phospho (P)-PDK1], involved in PKCζ activation (12), could also interact with p62 in response to RANKL stimulation. In immunoprecipitation studies, P-PDK1 was only weakly detected in p62 precipitates from HD^wt cells not stimulated with RANKL, and RANKL stimulation increased the detection of P-PDK1 in p62 precipitates in a time-dependent manner. In PDB^wt cells, the level of P-PDK1 was significantly higher in p62 precipitates than in HD^wt cells before adding RANKL (P < 0.001), and the P-PDK1/p62 ratio remained high after RANKL stimulation (Fig. 6A). As in PDB^wt cells, in both HD^P392L and PDB^P392L cells, p62 was present in P-PDK1 precipitates before RANKL stimulation (Fig. 6C). Taken together, these findings show that, in addition to PKCζ, the PDK1 kinase is also part of the signaling complex that involves p62. In addition, these findings could suggest that there was a basal activated state of osteoclast signaling in PDB osteoclasts, even before RANKL was added, and this could be partly related to the presence of the p62 P392L mutation as the association between p62 and PKCζ or PDK1 was observed in healthy carriers before RANKL stimulation.

RANKL-induced signaling in p62-transfected osteoclasts

Interactions between p62 and PKCζ or PDK1

To identify the contribution of the P392L mutation in the signaling complex involving p62 in osteoclasts, we evaluated p62 interactions with the activated kinases (phospho-PKCζ and phospho-PDK1) in human osteoclasts derived from cord blood transfected either with a plasmid containing p62^wt or p62^P392L or with an EV. RANKL stimulation was followed by immunoprecipitation with an anti-GFP antibody (to precipitate the GFP-fused p62 variant) or an anti-p62 antibody (nontransfected cells) (Fig. 7). In nontransfected cells, p62 was not associated with P-PDK1 before RANKL stimulation, but P-PDK1 was strongly associated with p62 10 min after stimulation. P-PKCζ/λ was detected in p62 precipitates before RANKL stimulation, and this association increased 10 min after RANKL addition, the peak level of association being detected after 30 min (Fig. 7A). The overexpression of p62^wt did not alter the pattern of P-PKCζ/λ or P-PDK1 interaction with p62, the association occurring after 30 and 10 min, respectively. In osteoclasts overexpressing p62^P392L, the base level of P-PKCζ/λ or P-PDK1 found associated with p62 was significantly higher than in cells overexpressing p62^wt but did not significantly increase after RANKL stimulation (Fig. 7B). These results confirm the association between p62 and activated kinases P-PKCζ/λ or P-PDK1 and argue in favor of a role of the p62 PDB-related mutation to induce an osteoclast-activated phenotype as, even in the absence of RANKL stimulation, the P392L p62 mutation enhanced the association between p62 and activated kinases.

Fig. 7.

RANKL-induced p62 interactions in transfected osteoclasts. At the end of CBM cultures, cells were transfected on d 17 of culture with a pEGFP-C2 plasmid containing p62^wt, p62^P392L, or an empty pEGFP vector (EV), or were not transfected (NT). At the end of CBM cultures, cells were stimulated with RANKL (100 ng/ml) for 0, 10, and 30 min. A, Immunoprecipitations (IPs) were conducted in nontransfected cells with an anti-p62 antibody in cell lysates. Western blots (WBs) with anti-P-PKCζ/λ, anti-P-PDK1 and anti-p62 antibodies are shown. Whole cell lysates, not submitted to IP, were run in addition to IP fractions and blotted with the same antibodies. B, In transfected cells, IPs were conducted with an anti-GFP antibody in cell lysates, to precipitate the GFP-fused p62 variant, and Western blots (WBs) with anti-P-PKCζ/λ, anti-P-PDK1, and anti-GFP antibodies are shown. C, Before immunoprecipitation, a sample of whole-cell lysate was kept and run through SDS-PAGE in parallel to the IP fractions. After transfer, PVDF membranes were blotted successively with an anti-P-PKCζ, an anti-P-PDK1, or an anti-actin antibody. The data shown are representative of four independent experiments. ODs for bands corresponding to P-PKCζ/λ and P-PDK1 were corrected with the OD obtained for bands corresponding to p62, GFP, or actin, and computed in graphical representations. Analyses are reported as mean ratio ± sem. *, P ≤ 0.05; **, P ≤ 0.01; *** , P ≤ 0.001 vs. nonstimulated cells; ^#, P < 0.05; ^##, P < 0.01 vs. p62^wt.

RANKL-induced NF-κB activation

To further explore the mechanisms involved in the osteoclast activation stage in PDB, and the physiological relevance of the presence of p62-associated activated kinases, we wondered whether these associations were part of a signaling process that leads ultimately to NF-κB activation.

Because p62 mutations modified early RANKL-induced events, we assessed NF-κB activation in response to RANKL stimulation in p62-transfected osteoclasts. The activation of NF-κB was evaluated both by immunofluorescence (nuclear relocalization of NF-κB p50 subunit) and by Western blot (IκB degradation). In nontransfected human osteoclasts, p50 was detected in the cytoplasmic and nuclear area of nonstimulated cells, suggesting a basal level of activation in healthy surviving cells, and 30 min after RANKL stimulation, most of the p50 had been relocalized in the nuclei (Fig. 8A). IκB is bound to the inactive form of NF-κB and undergoes degradation after stimulation, thus freeing the active form of NF-κB. In nontransfected cultures, IκB was found in nonstimulated cells, but levels started to fall 15 min after RANKL stimulation (Fig. 8B). Like nontransfected cells, osteoclasts transfected with an EV displayed a nuclear translocation of p50, and IκB degradation occurred after RANKL stimulation. Surprisingly, in osteoclasts overexpressing p62^wt, p50 staining was barely detectable before or after stimulation and had a significantly lower base level of the inhibitor, which almost completely dissipated over time after the addition of RANKL. The lower expression of p50 subunit in p62^wt transfected osteoclasts was confirmed by Western blot analysis (supplemental Fig. 1 published as supplemental data on The Endocrine Society’s Journals Online web site at http://mend.endojournals.org). In contrast, osteoclasts expressing the p62 P392L mutation displayed high p50 nuclear staining under basal conditions, and this remained unchanged after the addition of RANKL (Fig. 8A and supplemental Fig. 1). The base level of IκB was similar to the one observed in cells transfected with an EV, but only limited degradation occurred after the addition of RANKL (Fig. 8B). As before, these results suggest that the p62^P392L osteoclast underwent an activation stage before the addition of RANKL, illustrated by the basal level of NF-κB activation.

Fig. 8.

Activation of NF-κB in transfected osteoclasts. At the end of CBM cultures, cells were transfected on d 17 of culture with a pEGFP-C2 plasmid containing p62^wt, p62^P392L, or an empty pEGFP vector (EV), or were not transfected (NT). A, Nuclear translocation of the p50 subunit. Cells were either left untreated or stimulated with RANKL (100 ng/ml) for 30 min and then immunostained for NF-κB p50 subunit. DAPI was added at the end of the procedure to make it possible to distinguish multinucleated cells from other cells. B, Degradation of IκB. Nontransfected cells were stimulated for 5, 10, 15, 30, or 60 min with 100 ng/ml RANKL or left untreated. C, Degradation of IκB. Cells were transfected as above, then stimulated for 10 or 45 min with RANKL or left untreated. Western blots were performed with anti-IκB and antiactin antibodies. Graphic representations of average ODs measured from blots are presented. ODs for bands corresponding to IκB were corrected using the OD obtained for bands corresponding to actin. Analyses are reported as mean ratio ± sem. *, P ≤ 0.05; **, P ≤ 0.01 vs. nontreated cells. Results are representative of three experiments. WB, Western blot.

Discussion

Mutations in the gene encoding p62 have been shown to be involved in the pathogenesis of PDB (3). Although the effects of p62 mutations in the development of pagetic osteoclasts are still unclear, recent studies, mainly performed in murine models or transfected human cells, have suggested their potential role in the excessive formation of osteoclasts, and in their altered phenotype (7, 9, 10). The major contribution of our work is that we carried out cellular studies in a large cohort of PDB patients and healthy donors with known genetic status with regard to p62, including, for the first time, healthy carriers of the p62^P392L gene.

Osteoclasts obtained from cultures of peripheral monocytes or bone marrow cells from PDB patients are increased in number and size and have increased numbers of nuclei, increased bone-resorbing activity, and are hyperresponsive to osteoclastogenetic factors, such as RANKL, TNFα, or 1,25-(OH)₂D (13, 14, 15, 16). These were also characteristics of the osteoclasts obtained from PDB patients in our study, regardless of whether the p62 mutation was present or not, and contrasted with the osteoclasts from healthy controls. Moreover, we found lower rates of apoptosis induced by the deprivation of survival factors or by death inducers like TRAIL, Fas, or TGF-β1 in osteoclast cultures derived from PDB patients. This resistance to apoptosis could contribute to the increased number of osteoclasts in pagetic bone lesions and strongly supports the notion that there is a general increase in the activation of survival signals in these cells (17). An interesting observation is that osteoclasts from healthy donors bearing the p62^P392L gene had an intermediate phenotype. Indeed, they displayed increases in osteoclast formation, in the numbers of nuclei per cells, and in the resistance to apoptosis, which were all higher than in osteoclasts from normal controls, but lower than in osteoclasts from PDB patients. This could suggest that the p62 P392L mutation may affect these processes, although predisposing factors to PDB other than mutation of the p62 gene could have overcome the effects of the mutation, because these patients were relatives of PDB patients. To identify the specific effects of the p62 P392L mutation, we used osteoclasts derived from cord blood monocytes transfected with p62^wt or p62^P392L genes. We showed that, compared with p62^wt transfected osteoclasts, the p62 P392L mutation was indeed associated with increased osteoclast formation and greater nuclei numbers per osteoclast, but not with the increased bone resorption after correction for osteoclast numbers, or with the apoptosis rate.

The increased numbers of nuclei per osteoclast is a well-known feature of osteoclasts in PDB. The notion that the p62 P392L mutation may enhance the fusion process is consistent with a study in which murine osteoclasts generated from the RAW 264.7 line and transfected with p62 that lacked the UBA domain displayed greater expression of dendritic cell-specific transmembrane protein (DC-STAMP) (18), a molecule involved in preosteoclast fusion that is up-regulated after RANKL stimulation (19). In addition, osteoclasts from healthy carriers of p62^P392L were more sensitive to high doses of RANKL than normal osteoclasts, because their bone resorption activities were significantly higher than those in normal subjects when the RANKL concentration increased, reaching levels comparable to those seen in pagetic osteoclasts. This may reflect hyperresponsivity to RANKL related to the presence of the PDB-related p62 mutation, as previously reported in pagetic osteoclasts bearing this mutation (9).

In a murine model, a ternary complex involving TRAF6, p62, and the atypical PKCs (aPKCs), is formed in response to RANKL stimulation, identifying p62 as an important mediator during osteoclastogenesis (7). Consistent with these data, our results demonstrate, for the first time, that RANKL stimulation induces the formation of a multiprotein complex containing not only PKCζ and p62, but activated PDK1 as well, the substrates of which include PKCζ. The activation of PDK1 depends on phosphatidylinositol-3-kinase, which is generally activated by tyrosine kinase receptors (12), but also by TRAF6 (20). The recruitment of P-PDK1 by p62 may provide the subsequent activation of downstream PKCζ-dependent cascades during RANKL stimulation, as suggested by the kinetics observed in nonmutated healthy cells in which activated PDK1 appeared before phosphorylated PKCζ. Both PDK1 and PKCζ kinases have previously been shown to be involved in increased survival (12), and in the inhibition of apoptosis (21, 22) in various models, and the only overexpression of PKCζ may deregulate growth control (23). Interestingly, we found an association of p62 with both activated PDK1 and PKCζ or activated PKCζ/λ in PDB osteoclasts, in osteoclasts from healthy donors harboring the p62^P392L gene, and in p62 transfected osteoclasts, even before RANKL stimulation. Moreover, the signaling complex formed in response to RANKL stimulation normally results in NF-κB activation, which was actually observed in the nontransfected osteoclasts. A repressive effect of the overexpression of wild-type p62 on NF-κB activation was observed; this has been reported previously, although no clear explanation has been provided so far (10, 18). This effect may be related to increased degradation of both IκB and NF-κB as a result of p62 overexpression, but further studies are needed to clarify this. We found that mutated p62 overexpression promoted the formation of higher levels of activated kinase-p62 complex than wild-type p62, along with an increase in the basal level of NF-κB activation. Our findings strongly suggest that p62^P392L contributes, at least in part, to the induction of an activated stage in osteoclasts by stimulating signaling pathways involving PDK1-PKCζ/λ that could lead to NF-κB activation. Because PKCζ/λ and p62 belong to the RANKL-signaling complex, aPKCs could be potential IκB kinase (IKK) kinases in the p62-driven activation of NF-κB in osteoclasts. However, p62 may also contribute to the activation of NF-κB independently of aPKC through TRAF6 polyubiquitination (24). In addition, as the lost of PKCζ does not impair osteoclastogenesis (7), it has been speculated that p62/PKCλ might be necessary for NF-κB activation, whereas PKCζ could act through NF-κB-independent pathways (24). These aspects need further investigation to explain the mechanistic relationship between the activation of the PDK1/PKCζ/λ and NF-κB pathways that we observed in the presence of p62^P392L mutant.

Every PDB p62 mutation identified to date clusters within the C-terminal region of the protein, the UBA domain (residues 387-436). P392L substitution, as well as the other mutations described in PDB, alter the interactions of the UBA domain with multiubiquitin chains (5); therefore, a defective UBA domain could prevent p62 degradation and subsequently promote constitutive activation. Our results seem to confirm this hypothesis, because increased expression of p62 protein was detected in pagetic osteoclasts, as previously described at a transcriptional level (25). This increased expression of p62 was not influenced neither by the presence of the P392L mutation in PDB patients, nor was it detected in cells from healthy carriers of the p62^P392L gene, suggesting that the P392L mutation itself is not involved in the overexpression of p62 by osteoclasts. However, these findings also strongly suggest that even without a mutation, p62 is by some way involved in the overactivated phenotype of these cells, because bone resorption and survival were also increased in osteoclasts overexpressing the p62^wt protein. Moreover, RANKL stimulates p62 expression in osteoclasts from murine bone marrow cultures (7), and high levels of RANKL are present in the bone microenvironment in pagetic bone lesions, and thus around osteoclasts that are hyperresponsive to this factor (15). p62 is ubiquitously expressed, and p62 mutations may also affect cells other than osteoclasts, particularly stromal cells or osteoblasts. This hypothesis has been investigated in a recent study, in which p62^P394L (the murine equivalent of human p62^P392L) knock-in mice were generated. The expression of p62^P394L in stromal cells was associated with increased RANKL expression in response to 1,25-dihydroxyvitamin D₃ (26). Thus the presence of the p62 mutation in bone cells other than osteoclasts may be part of a vicious cycle leading to an increase in RANKL production in the bone microenvironment, and increased p62 expression in osteoclasts, indirectly enhancing osteoclast formation and activity. In addition, and according to studies in other systems, increased expression of p62 could in itself have a protective effect against cell death by preventing the buildup of potentially cell-damaging proteins (27, 28) and could act as a molecular switch to activate intracellular cascades (29, 30).

Growing data support that p62 may act as a conformational adapter linking ubiquitinated substrates bound to its UBA domain to the proteasome and also to signaling partners linked to other domains, such as the TRAF6-binding domain (29, 31, 32). Thus, it appears that the consequences of mutations in the UBA domain of p62 may affect pathways other than the degradation of p62. It could affect the turnover of other proteins, or the signaling complexes induced by RANKL, TNFα, or IL1 that lead to NF-κB activation and include p62 as a critical scaffold protein (33). These possibilities remain to be further explored, and our findings provide considerable support to the concept that activated signaling pathways are not solely the result of p62 overexpression but also depend on the presence of UBA p62 mutations, such as P392L, in osteoclasts.

Data from transgenic mice concluded that the p62 P392L mutation leads to increased osteoclastogenesis but does not induce the complete Pagetic phenotype (9). Our findings support this conclusion, because part of the osteoclast phenotype was related to the presence of the p62 P392L mutation. Our results provide strong evidence that the p62 P392L mutation plays an important role in driving osteoclasts into a persistent state of activation. They also strongly highlight the major cellular changes observed in carrier patients who have no sign of the disease, but are at an increased risk of developing it (3, 4). As already suggested (34, 35, 36), UBA p62 mutations could lead to activated osteoclasts primed for further events, or environmental or other genetic factors that would eventually lead to the development of PDB.

Materials and Methods

Clinical investigation and phenotype classification

This research has been approved by the Ethics Committees of the Centre Hospitalier de l’Université Laval and the Centre Hospitalier de l’Université de Sherbrooke. All participants, Paget sufferers or not, signed an informed-consent form before entering the study. The clinical assessment consisted of a physical examination (JB), and measurement of the total serum alkaline phosphatase, x-rays of the skull and pelvis (enlarged view including the last three lumbar vertebrae, the pelvis, and the femoral heads), and a whole-body bone scan.

A blood sample (50 ml) was obtained from healthy donors and PDB patients. For every donor and patient, exons 7 and 8 of the gene encoding SQSTM1 were sequenced, as previously described (3). For our present study, we analyzed data from 33 subjects (43–81 yr of age; mean age, 65.7 yr; 70% females) who were divided into four groups as follows. Healthy donors: p62 gene sequenced and exempt from any known mutation (HD^wt, n =5); healthy donors: p62 gene with the P392L mutation (these donors usually came from families with a history of PDB but had no signs or symptoms of PDB at the time of the study) (HD^P392L, n = 3); patients with clinical manifestations of PDB: p62 gene sequenced and exempt from any known mutation (PDB^wt, n = 8); patients with clinical manifestations of PDB: p62 gene shown to have the P392L mutation (PDB^P392L, n =17).

Materials

Opti-MEM, penicillin, streptomycin, fungizone, glutamine, and fetal calf serum were purchased from Wisent (Montreal, Quebec, Canada). Ficoll-Paque was purchased from Amersham Biosciences (Montreal, Quebec, Canada). Human recombinant (hr) M-CSF, hrGM-CSF, and hrTRAIL were purchased from R&D Systems (Minneapolis, MN); hrTGFβ1 was purchased from Peprotech (Rocky Hill, NJ), and soluble hrRANKL was produced in our laboratory. Goat polyclonal anti-p62 (P-15; directed against the C terminus), rabbit polyclonal anti-NFκB (p50 subunit), and anti-inhibitor of NF-κB (IκB) antibodies were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). We used rabbit polyclonal antibodies to detect phospho-Ser241-PDK1, PKCζ and phospho-PKCζ/λ (Cell Signaling Technology, Danvers, MA). Mouse monoclonal Fas-activating antibody (CD95) was purchased from Medical & Biological Laboratories (MBL; Watertown, MA).

Cell cultures

PBMCs were isolated from heparinized blood by density-gradient centrifugation, washed, and suspended in Opti-MEM with the antibiotics, glutamine, and 2% fetal calf serum. They were plated at a density of 3 × 10⁶/ml on eight-well chamber/slides (Lab-Tek, Biosciences, Bedford, MA). After incubating overnight, the cells were washed to remove any nonadherent cells. The selected PBMCs were cultured in medium supplemented with GM-CSF (100 pg/ml) for the first 3 d, and then for a further 3 wk in the same medium supplemented with M-CSF (25 ng/ml) and RANKL (75 ng/ml). The medium was changed every 2–3 d. Alternatively, blood was harvested from human umbilical cord at delivery after obtaining informed consent from parturient women, as approved by our institution’s review board. Monocytes were then obtained and processed as described above. These culture conditions yielded multinucleated cells (MNCs) that expressed osteoclast markers and possessed the ability to resorb bone (11).

Constructs and transfections

The full coding sequence of SQSTM1 (p62^wt) was obtained from IMAGE clone no. 2906264 and subcloned into a pEGFP-C2 vector. The P392L mutation (p62^P392L) was generated from the wild-type sequence by PCR-directed mutagenesis. These GFP-fused constructs have previously been shown not to interfere with the normal subcellular localization of p62 (37). Differentiated human osteoclasts were transfected in serum-free Opti-MEM containing LipoLTX for 6 h. The transfection efficiency was assessed by fluorescence microscopy 24 h after the procedure and ranged between 40% and 70% in all cultures. The effective expression of the plasmid was assessed by RT-PCR with specific primers of p62 (expressed 4 to 7 times more than endogenous mRNA, depending on the culture) and Western blot against GFP.

Evaluation of bone resorption

CBMs or PBMCs were allowed to settle on 150-μm devitalized bovine bone slices and then cultured for 3 wk as described above. Where appropriate, cells were transfected on d 17 of culture and allowed to resorb bone until d 22. The bone slices were then removed, washed with sodium hydroxide and distilled water, sonicated to remove cell debris, and stained with 1% toluidine blue containing 1% sodium borate. The resorption pits then appeared blue/violet on the slice. Optical light microscopy was used to determine bone resorption. Pictures were taken and analyzed using BioQuant Nova version 5.50.8 software to calculate the percentage of the resorbed area on the slice.

Immunofluorescence

At the end of the maturation period, osteoclasts were washed quickly with cold PBS and fixed with 1% paraformaldehyde. After permeabilization and autofluorescence quenching, nonspecific binding sites were blocked with 5% BSA. Specific antibodies directed against p62 or NF-κB (p50) were incubated in 5% BSA overnight at 4 C. After washing, Alexa-594 (red) or Alexa-546 (green) antigoat antibodies were incubated for 1 h and then washed several times with PBS. Cells were then counterstained with 0.5 ng/ml 4′,6-diaminido-2-phenylindole (DAPI) (Sigma, Oakville, Ontario, Canada) to visualize the nuclei. Alternatively, transfected cells expressing the GFP-fused p62 were counterstained with DAPI and visualized. Cells containing more than three nuclei were pictured using specific filters for the visualization of each color.

Study of osteoclast apoptosis

At the end of the maturation period, cells were deprived of M-CSF and RANKL for 24 h, and factors previously reported to affect osteoclast apoptosis were then added for another 24 h. Various concentrations of TRAIL (10–400 ng/ml), a Fas-activating antibody (50–800 ng/ml), TGFβ1 (0.1–4 ng/ml), and M-CSF (1–50 ng/ml) were added as appropriate. After 24 h, cells were fixed in 1% formaldehyde solution, washed, and permeabilized using 0.1% Triton. Condensed nuclei, typical of apoptosis, were detected using DAPI. Osteoclasts were defined as cells containing more than three nuclei, and apoptotic cells were defined as having condensed nuclei and/or apoptotic bodies.

Immunoprecipitation and Western blot

Cells were cultured and stimulated as described above. After stimulation, the cells were lysed for 20 min in a buffer containing Nonidet P40 and a cocktail of protease/phosphatase inhibitors. Lysates were precleared with goat or rabbit serum followed by A/G agarose bead precipitation. Lysates were then submitted to overnight immunoprecipitation at 4 C with antibodies directed against p62 or GFP. Immune complexes were recovered with Protein A/G agarose beads and loaded onto SDS-PAGE. Western-blot were performed by incubating specific primary antibodies overnight at 4 C. Horseradish peroxidase-conjugated secondary antibodies were used to achieve detection with a chemiluminescent system. Resulting films were scanned and densitometric analyses were performed with ImageJ software.

Statistical analysis

Results were expressed as means ± sem, and the significance was evaluated using one-way or two-way ANOVA with Tukey or Bonniferroni posttests where applicable. The statistical significance was defined as P < 0.05.