|Home | About | Journals | Submit | Contact Us | Français|
The orphan nuclear receptor, estrogen receptor-related receptor α (ERRα) is expressed in osteoblasts and osteoclasts (OCs) and has been proposed to be a modulator of estrogen signaling. To determine the role of ERRα in OC biology, we knocked down ERRα activity by transient transfection of an siRNA directed against ERRα in the RAW264.7 monocyte–macrophage cell line that differentiates into OCs in the presence of receptor activator of nuclear factor κB-ligands and macrophage colony-stimulating factor. In parallel, stable RAW cell lines expressing a dominant-negative form of ERRα and green fluorescent protein (RAW-GFP-ERRαΔAF2) were used. Expression of OC markers was assessed by real-time PCR, and adhesion and transmigration tests were performed. Actin cytoskeletal organization was visualized using confocal microscopy. We found that RAW264.7 cells expressing siRNA directed against ERRα and RAW-GFP-ERRαΔAF2 OCs displayed abnormal spreading, and decreased osteopontin and β3 integrin subunit expression compared with the corresponding control cells. Decreased adhesion and the absence of podosome belts concomitant with abnormal localization of c-src were also observed in RAW-GFP-ERRαΔAF2-derived OCs. In addition, RAW-GFP-ERRαΔAF2-derived OCs failed to transmigrate through osteoblast cell layers. Our data show that the impairment of ERRα function does not alter OC precursor proliferation and differentiation but does alter the adhesion/spreading and migration capacities of mature OCs.
Osteoclasts (OCs) are large multinucleated cells of hematopoietic origin formed by the differentiation and fusion of mononuclear monocyte–macrophage lineage precursors after stimulation by receptor activator of nuclear factor κB-ligands (RANKL) and macrophage colony-stimulating factors (M-CSF; Boyle et al. 2003). OC formation is associated with the expression of markers such as tartrate-resistant acid phosphatase (TRAP) and calcitonin receptor (CTR). Moreover, OCs are highly polarized cells, and several studies have identified a series of factors involved in OC spreading, polarity, adhesion, and migration, such as osteopontin (OPN), integrin subunits αv, β1, β3, β5, cluster of differentiation 44 (CD44), c-src, and tumor necrosis factor receptor-associated factor 6 (Traf6; Denhardt & Noda 1998, Chellaiah & Hruska 2003, Chabadel et al. 2007, Teitelbaum 2007). OCs are highly migratory on bone surfaces or through collagen (Domon et al. 2002, Delaisse et al. 2003, Saltel et al. 2004). Recently, we have found that OCs are also able to transmigrate through cell layers, a process that involves c-src and metalloproteinases (MMPs; Saltel et al. 2006). Moreover, OCs exhibit several features, such as podosomes, that are associated with cell adhesion, migration, and invasion. In fact, we have previously shown that mature OCs, when spread on glass or plastic, form highly dynamic podosome belts (Destaing et al. 2003, Jurdic et al. 2006) at their periphery; whereas, when cultured on apatite mineral (corresponding to the mineral phase of bone), they exhibit another actin-rich structure, the sealing zone, which seals off the resorption area (Vaananen et al. 2000, Saltel et al. 2004, 2008).
Nuclear steroid receptors are transcription factors involved in various physiological regulatory processes, including multiple processes in bone (Gennari et al. 2007). The superfamily to which nuclear receptors belong comprises both ligand-dependent molecules such as ERs and a large number of so-called orphan receptors, for which no ligands have yet been determined (Benoit et al. 2006). Three orphan receptors, ER-related receptor α (ERRα), ERRβ, and ERRγ (NR3B1, NR3B2, and NR3B3 respectively, according to the Nuclear Receptors Nomenclature Committee 1999), share similarities with ERα and ERβ (NR3A1 and NR3A2 respectively; Green et al. 1986, Kuiper et al. 1996), but they do not bind estrogen (Greschik et al. 2002, Kallen et al. 2004). Sequence alignment of ERRα and the ERs reveals a high similarity (68%) in the 66 amino acids (aa) of the DNA-binding domain, but only a moderate similarity (36%) in the ligand-binding domain, which may explain the fact that ERRα recognizes the same DNA-binding elements as ERs but does not bind estrogen (Giguere et al. 1988). Several studies have shown that ERRα may activate gene transcription constitutively essentially through its coactivator-binding domain AF-2 (Vanacker et al. 1999, Kallen et al. 2004). ERRα regulates fatty acid oxidation and the adaptative bioenergetic response (Luo et al. 2003, Huss et al. 2007). ERRα is highly expressed in skeletal (bone and cartilage) tissues (Bonnelye et al. 2001, 2007) and has been reported to regulate osteoblast development and bone formation in vitro (Bonnelye et al. 2001, 2007) and in vivo (Delhon et al. 2009, Teyssier et al. 2009). One study out of two has also described an ESRRA gene regulatory variant that is associated with bone mineral density and index in premenopausal women (Laflamme et al. 2005, Giroux et al. 2008).
In addition to its expression in osteoblasts, we have previously described ERRα expression in multinucleated TRAP-positive OCs in rat femurs after ovariectomy (Bonnelye et al. 2002). We therefore decided to address the possible function of ERRα in osteoclastogenesis, including OC differentiation and adhesion–transmigration activities, using RAW264.7 cells transfected with an siRNA directed against ERRα or with a wild-type (WT) or a dominant-negative form of ERRα.
A pEGFP vector (Clontech) was used to obtain the stable RAW-GFP(Ct) and RAW-GFP-ERRαΔAF2 (clones 1, 2, and 3) cell lines. Mouse ERRα cDNA (WT and ΔAF2) was obtained from mRNA extracted from primary OCs derived from spleen cells on day 6, using reverse transcription (RT)-PCR with specific primers ((NM_007953.1): ERRα upstream (177bp): TCC AGC ACC ATG TCC AGC; ERRα downstream (WT) (1–1436bp): CCA CCC CTT GCC TCA GTC CAT; ERRα downstream (ΔAF2): TCA TGC TGT CTG GCG GA (1–1341bp); helix 11–12 deletion (42 aa); Bonnelye et al. 2007). Amplicons were sequenced for verification. Anti-phospho-src (pTyr418) was from Abcam (ab4816); anti-integrin β3 chain was from BD (San Jose, CA, USA); anti-GFP was from Roche (monoclonal antibody mixture, clones 7.1 and 13.1). CY5-conjugated secondary antibodies (anti-rabbit, -mouse, -goat, and -hamster; 1/3000 final dilution) were from Jackson ImmunoResearch Laboratories (West Grove, PA, USA). The F-actin distribution was revealed with Alexa Fluor-546-Phalloidin from Molecular Probes (Eugene, OR, USA).
MC3T3-E1 cells were cultured in α-MEM containing 10% fetal bovine serum (FBS; Zirngibl et al. 2008). On the day of transfection, cells were transfected using Lipofectamine 2000 (Invitrogen). Spleen cells from 6-week-old OF1 male mice, none modified RAW264.7 and RAW264.7 (GFP(Ct) and GFP-ΔAF2) cells were cultured for 8 and 6 days respectively in differentiation medium: α-MEM containing 10% FCS (BioWest, Nuaille, France), 20ng/ml M-CSF and 20ng/ml soluble recombinant RANKL (Destaing et al. 2003).
The ERRα siRNA sequence was GCU AGU GCU CAG CUC UCU ACC CAA (403–426bp) (Invitrogen). This sequence was rigorously analyzed by BLAST search, and no homologies were found; this included no homology with any of the ERα, ERβ, ERRβ, or ERRγ, or with any other currently known nuclear receptors. The recommended negative control (Sc) by the manufacturer was used (Medium GC stealth RNA interference negative control duplexes; Invitrogen). ERRα siRNA and Sc were transfected twice into RAW cells following the manufacturer's protocol on days 3 and 4 after the addition of RANKL. mRNA was extracted on day 5.
HeLa cells were plated at 40000cells/well on a 24-well plate in DMEM (Sigma D5796) plus 10% FBS the day before transfection and incubated in a humidified incubator at 37°C in a 95% air–5% CO2 atmosphere. Media were changed just prior to transfection and again 24h after transfection. Cells were transfected with Lipofectamine 2000 (Invitrogen) following the manufacturer's protocol using 100ng expression vector, 200ng luciferase reporter vector, and 10ng pRL tk luc (Promega) for normalization. The pGL3 BSP reporter contains the rat bone sialoprotein promoter from −448 to −1 relative to the transcription start site cloned into pGL3B (Promega). Luciferase activities were measured 48h later on an EG&G Berthold Microplate Luminometer LB96V (EG&G Berthold GmbH & Co., Oak Ridge, TN, USA) using the Dual-Luciferase reporter assay system (Promega). Transfections were repeated at least three times, and a representative experiment is shown.
After 6 days, GFP(Ct) and GFP-ΔAF2 (1, 2, and 3) cells were fixed with paraformaldehyde (PFA; Merck), then stained for TRAP using a histochemical kit (no. 387; Sigma–Aldrich), following the manufacturer's instructions. Multinucleated TRAP-positive cells containing three or more nuclei were counted as OCs. To quantify OCs spreading, OCs derived from GFP(Ct) and GFP-ΔAF2 (1, 2, and 3) cells were fixed and labeled with Alexa Fluor-546-Phalloidin to visualize, by fluorescence, specific F-actin structure-containing podosome belts that characterize spread OCs. Spread OCs were quantified by counting the podosome belts versus the total number of OCs, under a X40 (NA 1.0) Zeiss Plan-Achromat objective and expressed as percentage. Results were plotted as the mean percentage of podosome belt number ±s.d. of triplicate wells for GFP(Ct)- and GFP-ΔAF2 (1, 2, and 3)-derived OCs and for siRNA-transfected cells and are representative of two independent experiments each.
OCs were detached and re-seeded on plastic in 12-well plates. The OCs were incubated at 37°C in PBS containing 0·25mM EDTA (PBS–EDTA) 4h later. The incubation was stopped either immediately (T=0) or after 35min (T=35). Cultures were then fixed with PFA and stained for TRAP. Adhesion was quantified by counting OCs that were still adherent at T=35min versus the total number of OCs at T=0 and expressed as percentage. Results were plotted as the mean percentage of adherent OCs±s.d. of triplicate wells for GFP(Ct)- and GFP-ΔAF2 (1, 2, and 3)-derived OCs and are representative of two independent experiments.
A transmigrating OC is an OC that had been detached and re-seeded on cell multilayers (for instance, osteoblasts) and that in 12h is able to cross the osteoblast multilayers without destroying them and to spread on the plastic substratum beneath the osteoblast. We used the osteoblastic cell line MC3T3-E1, which we cultured for at least 1 week to obtain multilayers of osteoblasts. At the end of the OC differentiation process, OCs derived from GFP(Ct) and GFP-ΔAF2 (1, 2, and 3) were removed and re-seeded on the top of the multilayers of MC3T3-E1 cells as described previously (Saltel et al. 2006). Primary OCs were also treated with echistatin at 10−8M (Sigma–Aldrich) for 24h, then seeded on MC3T3 in the presence of echistatin at 10−8M for 12h. Cells were then fixed after 20min or 12h, then permeabilized with 0·2% Triton X-100 for 7min, and stained with Alexa Fluor-546-Phalloidin. OC transmigration efficiency was quantified using a X40 (NA 1.0) Zeiss Plan-Achromat objective by counting the transmigrated OCs versus the total number of OCs and expressed as percentage. Results were plotted as the mean of the number of the transmigrated OCs±s.d. of three wells for GFP(Ct)- and GFP-ΔAF2 (1, 2, and 3)-derived OCs and as the mean number of the transmigrated OCs±s.d. of three wells for non-treated OCs and OCs treated with echistatin. They are representative of three and two independent experiments respectively.
Total RNA was extracted with TRIzol reagent from OC derived from spleen cells and RAW cells (GFP(Ct) and GFP-ΔAF2 (1, 2, and 3)) and from siRNA-transfected RAW cells. Samples of total RNA were reverse-transcribed using SuperScript II retrotranscription kit (Invitrogen; Bonnelye et al. 2001). Semi-quantitative PCR was performed for all studied genes. Real-time PCR was carried out to confirm regulated genes found by semi-quantitative PCR. Real-time RT-PCR was carried out by using the LightCycler system (SYBR Green; Qiagen) according to the manufacturer's instructions. PCRs were performed with the primers listed in Table 1, and amplimers were all normalized to corresponding L32 values. Data normalization: in real-time PCR, each replicate average gene's CT was normalized to the average CT of L32 by subtracting the average CT of L32 from each replicate to give the ΔCT. Results are expressed as with ΔΔCT equivalent to the ΔCT of the genes in GFP-ΔAF2 (1, 2, and 3) or siRNA subtracting to the ΔCT of the endogenous control (GFP(Ct), Scramble respectively).
Total proteins were extracted (pools from three 35mm dishes for each condition) from RAW cells transfected with GFP(Ct) or GFP-ERRα (ERRαWT or ERRαΔAF2) constructs according to the standard methods (Ausubel et al. 1996). Western blot analyses were performed using a semidry system. Immunoblotting was performed with anti-GFP (Roche; monoclonal antibody mixture, clones 7.1 and 13.1). Blots were incubated overnight at room temperature with the polyclonal antibody diluted at 1/100, and binding was detected using HRP-conjugated rabbit anti-mice antibodies (1/3000; Bio-Rad Laboratories) and chemiluminescence.
Cells were fixed in 4% paraformaldehyde, pH 7·2 for 10min, permeabilized with 0·2% Triton X-100 for 7min, then incubated with antibodies and AlexaFluor-546-Phalloidin as indicated. Cells were imaged with a confocal Zeiss LSM 510, using a X63 (NA 1.4) Plan Neofluor objective. To prevent contamination between fluorochromes, each channel was imaged sequentially, using the multitrack recording module, before merging. Z-cut pictures were obtained using Zeiss LSM 510 software.
Results for the real-time PCR analysis, the quantification of multinucleated TRAP-positive OCs, podosome belt numbers, adhesion, and transmigration were expressed as mean±s.d., and analyzed statistically by one-way ANOVA with the treatment group as variance and by Student's t-test with InStat software (version 2.01, GraphPad Software, San Diego, CA, USA). Statistical significance was taken as P<0·05 (Student's t-tests *P<0·05, **P<0·01, ***P<0·001).
Consistent with our previous findings on the presence of ERRα protein in multinucleated TRAP-positive cells in rat femur in vivo (Bonnelye et al. 2002), we found, by real-time PCR, ERRα mRNA expressed throughout all the stages of OC differentiation: in primary leukocytes extracted from spleen (day 0), proliferation (day 2), mononucleated progenitor fusion (day 4), and maturation (days 6–8) with significant increases on days 4–8, when multinucleated OCs were starting to form until they reached maturity (Fig. 1A and B). For comparison, mRNA levels for ERRβ, ERRγ, and the late OC differentiation marker CTR were also assessed (Fig. 1A and B). Interestingly and in contrast to ERRα, ERRγ mRNA expression decreased dramatically during OC differentiation, while ERRβ was not detected. As previously shown (Bonnelye et al. 2002), ERRα is also expressed during OC differentiation of the RAW264.7 macrophage cell line, although fold changes in the expression level are less than that observed in primary OC cultures, which is probably due to the fact that RAW264.7 cells are already committed into the macrophage lineage (Fig. 1C). Given its expression in mature OCs, we then checked whether ERRα was involved in their function.
To check the involvement of ERRα in OC function, we used a truncated version of ERRα lacking the coactivator-binding domain AF2, ERRαΔAF2, which acts as a dominant-negative form of the receptor (Vanacker et al. 1999, Bonnelye et al. 2007). We first confirmed that GFP-ERRαΔAF2 acts dominant-negatively on the expression of OPN, an ERRα target gene in osteoblasts (Vanacker et al. 1998, Zirngibl et al. 2008, Teyssier et al. 2009), in the MC3T3-E1 cell model (Fig. 2A). GFP-ERRαWT was also made and transiently transfected with GFP-ERRαΔAF2 into the RAW264.7 macrophage cell line. Protein size was evaluated by western blotting with a GFP antibody, which detected a protein of ~80kDa resulting from the fusion of the GFP (30kDa) with the ERRαWT protein (around 53kDa) used as a size control (Fig. 2B). As expected, a band of slightly lower molecular weight was detected in GFP-ERRαΔAF2 cells, corresponding to the truncation of the AF2 domain (42 aa; Fig. 2B). Finally, we confirmed the efficacy of GFP-ERRαWT and GFP-ERRαΔAF2 constructs by luciferase reporter assay with a (−448 to −1bp) BSP promoter fragment (pGL3BSPLuc), an ERRα target gene in osteoblasts (Bonnelye et al. 2001), in HeLa cells (Fig. 2C). RAW264.7 cells were then stably transfected with the GFP vector alone, which served as control (Ct), and the dominant-negative GFP-ERRαΔAF2 constructs. Three independent RAW-GFP-ERRαΔAF2 (1, 2, and 3) and one GFP alone clone were obtained, namely GFP-ΔAF2 (1, 2, and 3) and GFP(Ct) respectively. GFP expression was tested with specific GFP primers after 3 days of culture at the end of the proliferation stage (Fig. 2D and Table 1). GFP mRNA expression was observed in GFP(Ct) and all the three stable clones GFP-ΔAF2 (1, 2, and 3), with lower GFP expression in GFP-ΔAF2 (1, 2, and 3) versus GFP(Ct) cells (Fig. 2D). As expected, a band of 1·7kb was amplified with GFP (5′) and ERRα (3′) primers in the RNA isolated from GFP-ΔAF2 (1, 2, and 3) cells, whereas no band was detected in GFP(Ct) cells (Fig. 2D). GFP-ERRαΔAF2 expression was estimated to be tenfold less than the expression of the endogenous ERRα. Moreover, by using a GFP antibody, ERRα was localized by immunofluorescence in mature OCs derived from GFP(Ct) and GFP-ΔAF2 (1, 2, and 3) cells (Fig. 4C and G). As expected, ERRαΔAF2 was detected in both the nucleus and the cytoplasm of GFP-ΔAF2-2-derived OCs (Fig. 4G; similar results were obtained for GFP-ΔAF2-1 and -3), whereas no GFP was detected in the nucleus of GFP(Ct)-derived OCs (Fig. 4C). Also as expected, the expression level of the ERRα target genes, cytochrome c-1 (cytc; Huss et al. 2007) and OPN, was significantly reduced in all the three GFP-ΔAF2 clones in comparison with GFP(Ct) cells (Fig. 2E–G), confirming the dominant-negative function of the truncated ERRαΔAF2 construct. Interestingly, the expression of ERRγ, which was expressed at only undetectable to very low levels in RAW264.7 cells, was markedly increased in two of three GFP-ΔAF2 clones in comparison with GFP(Ct) cells (Fig. 2G).
To determine whether ERRα is involved in OC formation, GFP-ΔAF2 (1, 2, and 3) and GFP(Ct) cells were differentiated into mature multinucleated OCs in the presence of M-CSF and RANKL. Based on the cell counts after 3 days of culture, at the end of the proliferation stage, reduction of ERRα transactivation capacity had no effect on cell proliferation (data not shown). Concomitantly, the expression of early markers of OC differentiation, such as TRAP, c-fms, and RANKL (data not shown), and of the proliferation marker cyclin D1 was not affected (Fig. 2E and F). Later during differentiation, while the OPN expression was still down-regulated, we found that integrin β3 chain expression was also decreased in GFP-ΔAF2 (1, 2, and 3) compared with GFP(Ct) cells (Fig. 3A). On the other hand, the expression of αv, β1 and β5 integrin chain subunits, c-src, CD44, and Traf6 was not affected (Fig. 3A, and data not shown). Moreover, no differences were observed in the expression of OC factors involved in OC fusion (dendritic cell-specific transmembrane protein DC-STAMP (DCS)) or apoptosis (Bcl2 and Bax) in GFP-ΔAF2 (1, 2, and 3)- versus GFP(Ct)-derived OCs (Fig. 3A).
To confirm the putative functional role of ERRα in OC as suggested by the results with the dominant-negative form of ERRαΔAF2, we then checked whether the expression of OPN and integrin β3 chain would be similarly altered by the overexpression of ERRαWT or by an siRNA directed against ERRα. Transient transfection of ERRαWT in RAW264.7 cells at day 3 (transfection efficiency was very low at earlier stages) up-regulated the expression of OPN and integrin β3 chain at 48h (Fig. 3B). On the other hand, ERRα siRNA elicited a 60% decrease in the ERRα expression level at day 5 (Fig. 3C). OPN as well as integrin β3 expression was also down-regulated in OCs derived from ERRα siRNA-transfected RAW cells compared with control-transfected cells (Sc). Moreover, ERRγ expression was higher in ERRα siRNA-treated cells (Fig. 3C), consistent with the results in GFP-ΔAF2 (1, 2, and 3)-derived OCs (cf. Fig. 2). Taken together, these data suggest that ERRα does not play a role in the proliferation/differentiation process of macrophages to OCs, but may be involved in their adhesion, spreading, or migration activities.
To unravel the function of ERRα in OCs, GFP-ΔAF2 (1, 2, and 3) and GFP(Ct) RAW cells were differentiated into mature multinucleated OCs in the presence of M-CSF and RANKL. After 6 days, no difference was observed between the number of large multinucleated TRAP-positive OCs and between the nuclei number in GFP-ΔAF2 (1, 2, and 3) versus GFP(Ct) cells, suggesting that ERRα does not affect OC differentiation from macrophages (Fig. 4I–J). However, TRAP-positive GFP-ΔAF2-2-derived OCs (Fig. 4E) appeared less spread than TRAP-positive GFP(Ct)-derived OCs (Fig. 4A). This morphological difference was well correlated to their actin cytoskeletal organization. Indeed, well spread GFP(Ct)-derived OCs exhibited a typical podosome belt (Fig. 4B, D and K), whereas in less spread GFP-ΔAF2 (1, 2, and 3)-derived OCs, podosomes were organized as clusters or small rings (Fig. 4F, H and K). Similar results were obtained in RAW cells transfected with ERRα siRNA versus Sc siRNA or non-transfected cells (Fig. 4L). Together with the down-regulation of both OPN and β3 integrin chain, these latter results strongly suggest that ERRα is involved in OC adhesion.
Down-regulation of β3 expression in GFP-ΔAF2-2-derived OCs at day 6 (Fig. 3A) was confirmed by immunofluorescence using an antibody against integrin β3 (Fig. 5A and D). As expected, intense β3 integrin subunit staining was localized around the actin core of podosomes organized in a belt in GFP(Ct)-derived OCs at day 6 (Fig. 5A and B, see merged in Fig. 4C). In contrast, labeling for β3 integrin subunit protein was diffusely distributed within GFP-ΔAF2-2-derived OCs at day 6 (Fig. 5D and E, see merged in F). Since we, and others, have shown the importance of c-src in integrin signaling for actin organization, we looked for phosphorylated c-src localization in OCs expressing the dominant-negative form of ERRα. Whereas phosphorylated c-src was distributed around the podosome belt in GFP(Ct)-derived OCs on day 6 (Fig. 5G and H, see merged in I), it was diffusely localized in GFP-ΔAF2-2 cells (Fig. 5J and K, see merged in L). Similarly, vinculin localization was disrupted in GFP-ΔAF2 cells (data not shown) compared with GFP(Ct)-derived OCs at day 6. These results suggest that disturbing ERRα function impairs a signaling pathway linked to OPN, integrin β3, activated c-src, and actin organization. Moreover, and in agreement with the decrease in mature OC spreading, we observed a decrease in OC adhesion in GFP-ΔAF2 (1, 2, and 3)-derived OCs compared with GFP(Ct)-derived cells (by 40% in ΔAF2-1, 50% in ΔAF2-2, and 68% in ΔAF2-3 compared with GFP(Ct); Fig. 5M). It was observed that the differences in OC viability, as reflected by OC number, cannot account for the differences in spreading between GFP(Ct)- and GFP-ΔAF2 (1, 2, and 3)-derived OCs (T=0; Fig. 5N).
To address further the consequences of disrupting the actin structure due to the expression of a dominant-negative form of ERRα in OCs, we then checked whether OC transmigration was affected (Saltel et al. 2006). To this end, mature GFP-ΔAF2 (1, 2, and 3)- and GFP(Ct)-derived OCs were seeded onto confluent osteoblastic MC3T3-E1 cells. After 20 min, large rounded OCs (identified by GFP expression in green, white arrows) derived from GFP-ΔAF2 (1, 2, and 3) (Fig. 6G, see Z-cut section E) and GFP(Ct) (Fig. 6C, see Z-cut section A) cells were localized on the top of confluent osteoblastic MC3T3-E1 cell layers (red, see white asterisk). After 12h, OCs derived from GFP(Ct) cells were well spread and attached to the substratum beneath the osteoblasts (Fig. 6D, and Z-cut section in B), showing that they had transmigrated through the MC3T3-E1 cell layers. In contrast, after 12h of culture, OCs derived from GFP-ΔAF2-2 cells appeared ‘frozen’ on the top of the MC3T3-E1 cell layers where they maintained a rounded shape (Fig. 6H, see GFP-positive OCs (white arrow) and Z-cut section in Fig. 6F). Similar observations were obtained for GFP-ΔAF2 clones 1 and 3 (data not shown). Quantification of transmigrated OCs versus non-transmigrated OCs confirmed a significant decrease (72% in ΔAF2-1, 63% in ΔAF2-2, and 67% in ΔAF2-3) in the transmigrated OC numbers in GFP-ΔAF2 (1, 2, and 3)- compared with GFP(Ct)-derived OCs (Fig. 6I). Viability of the transmigrated GFP(Ct)- and GFP-ΔAF2 (1, 2, and 3)-derived OCs is equivalent (total OC number at T=12h; Fig 6J). Interestingly, the inhibition of transmigration observed with the knockdown of ERRα correlated with that observed after blocking integrins with echistatin, an RGD disintegrin, in primary OCs re-seeded onto confluent osteoblastic MC3T3-E1 cells (Fig. 6K).
We report that the overexpression or knockdown of ERRα function in the RAW264.7 macrophage cell line either by stable transfection of a dominant-negative form of ERRα, GFP-ERRαΔAF2 or by ERRα siRNA leads to modulation of the expression of OPN and β3 integrin subunit with concomitant actin cytoskeletal disorganization in the podosome belt, reduced adhesion on tissue culture plastic or glass, and reduced transmigration activity. Although ERRα knockout mice and cells are now available, our work was initiated when these were not available, and both our stable and transient knockdown approaches gave comparable and highly reproducible results. It is therefore not clear why no detectable alteration in the OC parameters was observed in ERRα knockout mice (Delhon et al. 2009, Teyssier et al. 2009), but it seems likely that compensatory mechanisms involving ERRγ may be operating, since we found that ERRγ is expressed in RAW264.7 cells when ERRα function is disrupted.
The molecular basis for the ERRα effect on cell mobility is of interest because OPN, which is expressed in several motile cell types including OCs and many tumor cell lines (Standal et al. 2004), has been described as a target gene of ERRα in promoter reporter assays in vitro (Vanacker et al. 1998, Zirngibl et al. 2008). OPN is involved in cell adhesion and migration and promotes cell attachment to bone (Denhardt & Noda 1998). Adhesion receptors αvβ1, αvβ3, and αvβ5 integrins and the hyaluronan receptor (CD44) are all receptors for OPN, and their interactions are crucial for OC adhesion and motility (McHugh et al. 2000, Chellaiah & Hruska 2003). Notably, OPN is known to stimulate glass-adherent OC migration through the αvβ3 integrin-induced activation of c-src and actin cytoskeletal reorganization into podosome belts (Duong et al. 1998). While αv, β1, and β5 integrin subunits and CD44 were not affected by the changes in ERRα activity, the expression of the β3 integrin subunit was down-regulated in siRNA and GFP-ΔAF2-derived OCs. In addition, more detailed studies in GFP-ΔAF2-derived OCs showed that the disruption of podosome belt formation was accompanied by an aberrant/diffuse distribution of the β3 integrin subunit, activated c-src, and vinculin. Thus, it is likely that the inhibition of adhesion and disruption of the podosome belt observed in GFP-ΔAF2-derived OCs is a consequence of the reduced expression of both OPN and β3 that consequently induced the disrupted localization of activated c-src. Moreover, although neither of the latter is known to be involved in OC transmigration, the fact that αvβ3 is involved in monocyte transendothelial migration, that echistatin, an RGD disintegrin, induces a decrease in OC transmigration, and that c-src is involved in OC transmigration through MC3T3-E1 cell layers suggests that the β3 integrin subunit may be involved in OC transmigration (Weerasinghe et al. 1998, Saltel et al. 2006). Indeed, in the presence of PP2, a c-src inhibitor, OCs maintain a rounded shape and sit on the top of MC3T3-E1 cell layers (Saltel et al. 2006) in a manner similar to GFP-ΔAF2-derived OCs. These data suggest that the disruption of activated c-src localization may, at least in part, be due to down-regulated OPN and/or integrin subunit β3 expression and may be responsible for the inhibition of transmigration in GFP-ΔAF2-derived OCs.
We have previously shown that ERRα positively regulates osteoblast formation in rat calvarial cells in vitro, acting at multiple developmental times to regulate both proliferation and differentiation events (Bonnelye et al. 2001). However, more recent data with ERRα knockout mice and with human bone marrow stromal cells suggest that ERRα may play a negative regulatory role in other models or at other developmental times (Delhon et al. 2009, Teyssier et al. 2009). Further complicating the issue, one study suggests that an ESRRA gene regulatory variant leading to increased ERRα expression is associated with increased bone mineral density in premenopausal women (Laflamme et al. 2005), although a recent report has failed to confirm the observation (Giroux et al. 2008). Such apparently complex and multiphasic activities, complicated by potential compensatory mechanisms involving other ERR family members, are in keeping with our current data showing that down-regulation of ERRα decreases OC transmigration, suggesting that ERRα may positively regulate OC migration. Altogether, the data suggest that ERRα may contribute to bone homeostasis not only through effects on osteoblast lineage cells but also on OCs.
ERRα expression in other highly motile cells such as dendritic cells and macrophages (Sonoda et al. 2007, Bonnelye et al. 2008) supports a role for ERRα in inflammatory diseases associated with osteoclastic bone degradation such as rheumatoid arthritis and other diseases, where αvβ3 and OPN are also implicated (Wilder 2002, Yumoto et al. 2002). In this regard, recently, and similarly to our data on OC migration, siRNA-mediated knockdown of ERRα in the highly aggressive breast carcinoma MDA-231 cell line resulted in a dramatic decrease in their migration (Stein et al. 2008). We thus hypothesize that the ERRα-regulated OPN and integrin subunit β3 pathway that we described here may also be operative in cancer cells.
In conclusion, ERRα is expressed in OCs where it regulates the expression of OPN and integrin subunit β3, and plays a role in the spreading, organization, and/or stabilization of podosomes at the cell periphery, as well as in the adhesion and transmigration in a RAW cell culture model in vitro. Further understanding of the mechanisms by which ERRα modulates migration in the normal development and disease is required. In the meantime, our data suggest that the regulation of ERRα activity may provide a new therapeutic strategy for inflammatory and bone diseases associated with an imbalance in bone remodeling due to excessive OC resorption.
The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.
This work was supported by the ‘Ligue Contre le Cancer’ (Rhône, Drome), ARC and CNRS, and the Canadian Institutes of Health Research (CIHR FRN 88104 to JEA). AC was the recipient of an ‘allocation couplée’ grant from ENS Lyon.
We are very grateful to Fabienne Simian and Claire Lionnet from PLATIM (IFR 128 Lyon Biosciences) for their kind help with imaging experiments, and to Olivier Destaing for helpful discussions.
**(F Saltel and A Chabadel contributed equally to this work)
†(F Saltel is now at European Institute of Chemistry and Biology, Unité INSERM 889, Université Victor Segalen Bordeaux 2, IFR66, 2, rue Robert Escarpit, 33 600 Pessac, France)