Homer1 mediates CaSR-dependent activation of mTOR complex 2 and initiates a novel pathway for AKT-dependent β-catenin stabilization in osteoblasts

The calcium-sensing receptor (CaSR) is critical for skeletal development, but its mechanism of action in osteoblasts is not well-characterized. In the central nervous system (CNS), Homer scaffolding proteins form signaling complexes with two CaSR-related members of the G protein–coupled receptor (GPCR) family C, metabotropic glutamate receptor 1 (mGluR1) and mGluR5. Here, we show that CaSR and Homer1 are co-expressed in mineralized mouse bone and also co-localize in primary human osteoblasts. Co-immunoprecipitation experiments confirmed that Homer1 associates with CaSR in primary human osteoblasts. The CaSR–Homer1 protein complex, whose formation was increased in response to extracellular Ca2+, was bound to mechanistic target of rapamycin (mTOR) complex 2 (mTORC2), a protein kinase that phosphorylates and activates AKT Ser/Thr kinase (AKT) at Ser473. siRNA-based gene-silencing assays with primary osteoblasts revealed that both CaSR and Homer1 are required for extracellular Ca2+-stimulated AKT phosphorylation and thereby inhibit apoptosis and promote AKT-dependent β-catenin stabilization and cellular differentiation. To confirm the role of the CaSR–Homer1 complex in AKT initiation, we show that in HEK-293 cells, co-transfection with both Homer1c and CaSR, but neither with Homer1c nor CaSR alone, establishes sensitivity of AKT–Ser473 phosphorylation to increases in extracellular Ca2+ concentrations. These findings indicate that Homer1 mediates CaSR-dependent AKT activation via mTORC2 and thereby stabilizes β-catenin in osteoblasts.

The calcium-sensing receptor (CaSR) is critical for skeletal development, but its mechanism of action in osteoblasts is not well-characterized. In the central nervous system (CNS), Homer scaffolding proteins form signaling complexes with two CaSRrelated members of the G protein-coupled receptor (GPCR) family C, metabotropic glutamate receptor 1 (mGluR1) and mGluR5. Here, we show that CaSR and Homer1 are co-expressed in mineralized mouse bone and also co-localize in primary human osteoblasts. Co-immunoprecipitation experiments confirmed that Homer1 associates with CaSR in primary human osteoblasts. The CaSR-Homer1 protein complex, whose formation was increased in response to extracellular Ca 2؉ , was bound to mechanistic target of rapamycin (mTOR) complex 2 (mTORC2), a protein kinase that phosphorylates and activates AKT Ser/Thr kinase (AKT) at Ser 473 . siRNA-based gene-silencing assays with primary osteoblasts revealed that both CaSR and Homer1arerequiredforextracellularCa 2؉ -stimulatedAKTphosphorylation and thereby inhibit apoptosis and promote AKTdependent ␤-catenin stabilization and cellular differentiation. To confirm the role of the CaSR-Homer1 complex in AKT initiation, we show that in HEK-293 cells, co-transfection with both Homer1c and CaSR, but neither with Homer1c nor CaSR alone, establishes sensitivity of AKT-Ser 473 phosphorylation to increases in extracellular Ca 2؉ concentrations. These findings indicate that Homer1 mediates CaSR-dependent AKT activation via mTORC2 and thereby stabilizes ␤-catenin in osteoblasts.
CaSR 3 is a family C G-protein-coupled receptor (GPCR) critical for the development of bone and cartilage (1)(2)(3) and contributes to the development of the CNS, in part by promoting the differentiation of oligodendrocytes and formation of cerebral white matter (4). When CaSR exon-7, encoding the receptor's heptahelical domain and intracellular C terminus, was selectively knocked out of early osteoblast lineage cells in mice, bone mass was greatly reduced (1,2), and selective deletion in chondrocytes was embryonic lethal (1). In a human case of neonatal severe hyperparathyroidism, in which early stop codons were present in both CaSR alleles, additional phenotypic disturbances included growth retardation, generalized osteopenia, and attenuated development of the cerebral white matter (5). The signaling pathways that underlie CaSR-mediated control of cell fate, however, are largely undefined.
We have previously reported that activation of CaSR results in the dual phosphorylation of AKT in osteoblasts (6). Phosphorylation of AKT at Thr 308 is mediated by 3-phosphoinositide-dependent kinase 1 (PDK1) (7,8) downstream of phosphatidylinositol 3-kinase (PI3K), whereas phosphorylation of AKT at Ser 473 is mediated by the mechanistic target of rapamycin (mTOR) complex-2 (mTORC2), also known as PDK2 (9 -11), a complex of proteins that specifically includes a rapamycin-insensitive companion of mTOR (Rictor), but not the regulatory-associated protein of mTOR (Raptor) (12). mTORC2-dependent Ser 473 phosphorylation of AKT occurs independently of, and is required for, Thr 308 phosphorylation (9). The upstream mechanism by which mTORC2 couples to the CaSR, however, is unknown. Recently it was shown that mTORC2 localizes to the plasma membrane, independently of PI3K, where it functions to phosphorylate AKT at Ser 473 (13).
Canonical Wnt signaling in osteoblasts is a key determinant of bone cell fate and function (14,15). The process depends on the binding of extracellular Wnt proteins, such as Wnt3a, to Frizzled-LDL receptor-related protein-5&6 (LRP5/6), inactivating glycogen synthase kinase-3 (GSK3) via plasma mem-brane sequestration, resulting in the stabilization of cytoplasmic ␤-catenin that subsequently translocates to the nucleus where it promotes expression of genes under the control of TCF/LEF transcription factors (16). Independently of canonical Wnt-dependent inhibition of GSK3, AKT inhibits the kinase activity of GSK3 via direct phosphorylation at Ser 21 (␣-subunit) (17) and Ser 9 (␤-subunit) (18). Similarly, direct phosphorylation of ␤-catenin at Ser 552 by AKT promotes its nuclear translocation and transcriptional activity (19).
We previously reported that activation of the CaSR in human osteoblasts resulted in the AKT-dependent translocation of ␤-catenin to the nucleus in primary human osteoblasts, thus potentially linking the AKT and Wnt/␤-catenin pathways in bone cells (6). In agreement, Wnt3a-dependent activation of LRP5/6 in stromal ST2 cells activated the mTORC2/AKT pathway and promoted cell differentiation (20), showing that crosstalk between the pathways can potentially occur in both directions. However, the underlying mechanism by which the CaSR links to the mTORC2/AKT pathway in bone cells has not been defined.
Homer proteins act as scaffolds at the plasma membrane for the assembly of signaling complexes that include the family C metabotropic glutamate receptors mGluR1 and mGluR5 (21)(22)(23). The outcomes include cell fate determination via AKT (22,24) and ERK 1/2 (23) and underlie neuroplasticity in the hippocampus and other areas of the CNS (25). Several Homer isoforms are recognized (26,27). Among these, the long isoforms Homer1b and -1c (but not the short isoform 1a), as well as Homer2 and -3, contain coiled-coil domains that mediate selfmultimerization and Ena/Vasp homology (EVH1) domains (present in all Homer isoforms) that mediate signaling interactions (28,29). Disruption of Homer-mGluR1 interactions in mouse neurons decreased PI3K activity (24). Whether, in some cell contexts, Homer also supports CaSR-dependent activation of mTORC2 upstream of AKT is unknown.
In this study, we tested the hypothesis that in osteoblasts the long isoforms of Homer1 (hence referred to as Homer1) containing coiled-coil regions may link CaSR to mTORC2, resulting in activation of AKT and ␤-catenin stabilization in these cells (hypothesis shown schematically in Fig. 1).

CaSR and Homer1-mediated extracellular Ca 2؉ -stimulated AKT signaling in osteoblasts
We first confirmed that activation of CaSR by its physiological ligand Ca 2ϩ resulted in dual phosphorylation of AKT with phosphorylation of Ser 473 preceding phosphorylation of Thr 308 ( Fig. 2A). Maximal phosphorylation at these sites was reached 15 min following exposure to elevated Ca 2ϩ and was the time point used for all subsequent signaling analyses ( Fig. 2A). Inhi-bitionofmTORC2withTorin1decreasedCa 2ϩ -stimulatedphosphorylation of AKT at both Ser 473 and Thr 308 in a concentration-dependent manner (Fig. 2B). We confirmed that Torin1 treatment had no effect on mTORC1 at this early time point as evidenced by no change in mTORC1-Ser 2448 phosphorylation, a marker of mTORC1 activity (Fig. 2B) (30). Therefore, changes in AKT phosphorylation in response to the inhibitor at this time point were not due to a previously reported phenomenon of negative feedback-dependent regulation of mTORC2 by mTORC1 (31, 32) but were due to inhibition of mTORC2 protein kinase activity per se.
We next hypothesized that Homer1 may mediate extracellular Ca 2ϩ -induced mTORC2-dependent AKT signaling downstream of the CaSR in osteoblasts. Consistent with this hypothesis, when the expression of either the CaSR (Fig. 2, C and D) or Homer1 (Fig. 2, E and F) proteins was reduced selectively by specific siRNA in human osteoblasts, we observed decreased Ca 2ϩ -dependent AKT phosphorylation at both Ser 473 and Thr 308 compared with control.

Homer1 and CaSR co-localized and co-immunoprecipitated in vitro with mTOR and the mTORC2-specific protein Rictor in response to elevated extracellular Ca 2؉
Deconvolution microscopy of mouse bone sections revealed that CaSR and Homer1 were expressed in the same cells of mineralized bone (Fig. 3A, top row). As a comparison CaSR Ϫ/Ϫ -mineralized bone sections are also shown (Fig. 3A, bottom row). Confocal microscopy of a monolayer culture of primary human osteoblasts stained for CaSR and Homer1 is shown in Fig. 3B (top row). A high-resolution image of a single osteoblast (Fig. 3B, bottom row) indicated co-localization (Pearson's coefficient ϭ 0.41; below threshold ϭ Ϫ0.02).
Because Homer1 was required for extracellular Ca 2ϩ -dependent dual phosphorylation of AKT (Fig. 2, E and F), we hypothesized that Homer1 may interact with the CaSR and mTORC2 protein components, and we also tested whether any interaction might be dependent on an increase in extracellular Ca 2ϩ Figure 1. We propose Homer1 may link CaSR to mTORC2 in osteoblasts to promote AKT-dependent ␤-catenin stabilization. A proposal of this study is that Homer-1 may link CaSR to mTORC2 activation upstream of AKT (indicated by ?). AKT phosphorylation is known to occur via two discrete mechanisms: PI3K-dependent phosphorylation at Thr 308 , and mTORC2-dependent phosphorylation at Ser 473 . AKT subsequently inhibits GSK3 and directly phosphorylates ␤-catenin to promote ␤-catenin stabilization and allowing subsequent nuclear translocation leading to cellular differentiation. AKT also suppresses apoptosis and promotes growth in the cell. Kinase target of Torin1 is mTOR. PM ϭ plasma membrane.

Homer1 links CaSR to mTORC2
concentration. After immunoprecipitation with an anti-Homer1 antibody, Homer1 was itself readily detected, and this was unaffected by an increase in extracellular Ca 2ϩ concentration from 1.0 to 3.5 mM (Fig. 3, C and D, ns). Homer1 protein from nonimmunoprecipitated lysate is shown at the same molecular mass (ϳ45 kDa, Fig. 3C). CaSR protein, at ϳ170 kDa, a molecular mass representing glycosylated, plasma membrane-associated CaSR (33) (Uniprot ID P41180), was detected in Homer1 immunoprecipitates, and the amount of CaSR bound to Homer1 increased in response to elevated Ca 2ϩ (Fig.   Figure 2. CaSR and Homer1 were necessary for mTORC2-dependent dual phosphorylation of AKT in response to elevated extracellular Ca 2؉ in osteoblasts. A, phosphorylations of AKT at Thr 308 /Ser 473 from same osteoblasts by Western blotting in response to Ca 2ϩ o shift 1-2 mM over 1 h. B, Western blotting of AKT phosphorylated at Ser 473 /Thr 308 and mTOR phosphorylated at Ser 2448 in response to Ca 2ϩ o shift 1-2 mM for 15 min in the presence of Torin1 (pretreatment 5 min). Concentration is shown as fold times the half-maximal inhibitory concentration (IC 50 ). IC 50 Torin1 ϳ2 nM. C and E, Western blotting of phosphorylations of AKT at Thr 308 /Ser 473 , in response to Ca 2ϩ o shift 1-2 mM for 15 min after CaSR (siCaSR) (C) or Homer1 (siHomer1) (E) protein levels reduced (ϩ) compared with control transfected cells (Ϫ). D and F, densitometry of triplicate blots shown for silenced CaSR (C) and Homer1 (E), respectively. #, solid, vertical black lines on the presented Western blotting indicate where lanes have been spliced from the same Western blotting for presentation purposes. The data shown in C-F are also presented in Fig. 4 as part of the broader signaling cascade investigated in the study. *, p Ͻ 0.05; **, p Ͻ 0.01.
mTOR and the mTORC2-specific protein Rictor were present in nonimmunoprecipitated osteoblast lysates (Fig. 3C) and were detected in Homer1 immunoprecipitates. In the case of mTOR, however, no Ca 2ϩ -dependent increase in its level was observed (Fig. 3, C and D, ns), whereas Rictor protein levels, like those of the CaSR, increased in response to elevated extracellular Ca 2ϩ (Fig. 3, C and D, p Ͻ 0.05). In contrast, although the mTORC1-specific protein Raptor was present in lysates ( Fig.  3C), it was not detected in Homer1 immunoprecipitates at either control or elevated Ca 2ϩ (Fig. 3C). Thus, in the presence of an elevated extracellular Ca 2ϩ concentration, we observed increased association between the CaSR, Homer1, and the mTORC2-specific protein Rictor, but not the mTORC1-specific protein Raptor.

CaSR and Homer1 were required for extracellular Ca 2؉ -dependent ␤-catenin nuclear translocation following AKT activation in osteoblasts
We previously reported that the CaSR agonist strontium (Sr 2ϩ ) induced nuclear translocation of ␤-catenin in human osteoblasts following activation of AKT and demonstrated that these effects were mediated by the CaSR (6). Consistent with this observation, exposure of cells to elevated extracellular Ca 2ϩ resulted in concentration-dependent phosphorylations of the inhibitory sites of GSK3␣/␤ at Ser 21 /Ser 9 , respectively, and activator site of ␤-catenin at Ser 552 (Fig. 4A), and nuclear trans- Taken together, these observations indicate that the CaSR and Homer1 were required for extracellular Ca 2ϩ stimulation of an AKT-dependent pathway that controlled ␤-catenin translocation. To separately confirm the role of AKT on the nuclear translocation of ␤-catenin in vitro, chemical inhibition of AKT with AKT-XI (20 M) reduced osteoprotegerin (OPG) expression, a marker of the LEF/TCF pathway (Fig. 5A, panel vi) (15). A degree of CaSR control of ␤-catenin stabilization and translocation was also apparent in vivo as CaSR Ϫ/Ϫ mice had reduced protein levels of the Wnt marker OPG compared with WT ( Fig. 5B). Similarly, qPCR analysis also showed that the mRNA levels of several Wnt-responsive genes in CaSR Ϫ/Ϫ mice were reduced compared with WT ( Table 1).
As a control for the signaling analysis and to assess potential off-target effects, siRNA was used to knock down G␣ 12 , a protein that does not affect AKT but inhibits ERK 1/2 in osteoblasts (34,35). Silencing G␣ 12 resulted in reduced G␣ 12 protein (Fig.  4D, panel i, bottom row, and panel ii, p Ͻ 0.001) and, as expected, further increased Ca 2ϩ -induced ERK 1/2 phosphory-    S1). However, knockdown of G␣ 12 had no effect on the phosphorylations of AKT at Ser 473 or Thr 308 or of its key downstream targets, including GSK3, ␤-catenin, or mTORC1 as revealed by p70 S6K-dependent phosphorylation of mTOR at Ser 2448 (Fig. 4D, panel i for Western, and panel ii for significance).

CaSR and Homer1 were required for extracellular Ca 2؉ -dependent AKT and GSK3␤ phosphorylation in MG63 osteosarcoma cells
The effect of Homer1 and CaSR on the phosphorylation of AKT-Ser 473 and GSK3␤-Ser 9 in response to elevated extracellular Ca 2ϩ was also examined in a second osteoblast cell type, the commonly used osteosarcoma cell line MG63. We found that Homer1 and CaSR were expressed in MG-63 cells (Fig. 6). In these cells, both AKT-Ser 473 and GSK3␤-Ser 9 phosphorylation were increased in response to elevated extracellular Ca 2ϩ , and siRNA-dependent knockdown of either Homer1 or CaSR clearly reduced the phosphorylation level of both proteins in response to Ca 2ϩ o (Fig. 6). These findings were consis-tent with our analyses performed in primary human osteoblasts (Fig. 4).

Effect of CaSR and Homer1 on cell fate of osteoblasts
Given the observed effect of the CaSR and Homer1 on Ca 2ϩstimulated nuclear ␤-catenin translocation (Fig. 5), which promotes differentiation in osteoblasts (15), we next investigated whether the CaSR and Homer1 modulated alkaline phosphatase (ALP) activity, another marker of osteoblast differentiation (36). Elevated extracellular Ca 2ϩ (1.0 -2.0 mM) stimulated ALP activity by 4 -5-fold after 3 days (Fig. 7A, p Ͻ 0.01). Silencing the CaSR abolished the stimulatory effect of elevated Ca 2ϩ (2.0 mM) on ALP activity but had no effect on basal ALP activity (Fig.  7B). Silencing Homer1 suppressed basal ALP activity in cells exposed to control extracellular Ca 2ϩ (1.0 mM, p Ͻ 0.05) and abolished the response to elevated Ca 2ϩ (2.0 mM) (Fig. 7C), demonstrating that Homer1 supports a key function of differentiated osteoblasts in the physiologically relevant extracellular Ca 2ϩ concentration range.
Finally, we investigated the roles of CaSR and Homer1 in AKT-dependent osteoblast survival in the presence of pro-apoptotic oxidative stress, an important determinant of viability in vivo (37), using caspase-3 activity as the readout. Elevated extracellular Ca 2ϩ suppressed oxidative stress-induced elevations of caspase-3 activity in a concentration-dependent manner from 1 to 2 mM (Fig. 7D), and the prosurvival effect of elevated Ca 2ϩ was reversed upon silencing either the CaSR (Fig.  7E) or Homer1 (Fig. 7F). Furthermore, downstream of CaSR and Homer1, mTORC2-dependent phosphorylation of AKT-Ser 473 (9) was clearly critical in this action of extracellular Ca 2ϩ , because the inhibitor Torin1 caused significant concentrationdependent increases in caspase-3 activity in the presence of 2 mM Ca 2ϩ and the oxidative stressor 125 M H 2 O 2 (Fig. 7G), whereas wortmannin, the inhibitor of PI3K upstream of AKT, had no effect (Fig. 7H).

Transfection of Homer-1c into HEK-293 cells stably expressing CaSR established AKT-Ser 473 sensitivity to extracellular Ca 2؉
It is possible that the CaSR-Homer1-mTORC2 complex identified in osteoblasts in this study operates more widely in other cell types that express both the CaSR and Homer1. To test this hypothesis, we performed experiments in the human embryonic kidney cell line, HEK-293, and HEK-293 cells stably expressing CaSR (HEK-CaSR), a well-established cellular system for studies of CaSR signaling (38,39).

Summary of results
We made several novel observations in this study. Homer1 protein was expressed in human osteoblasts. In mineralized mouse bone sections of WT mice, high levels of CaSR expression were associated with high levels of Homer1 expression. In agreement with this observation, confocal microscopy in primary human osteoblasts demonstrated co-localization of Homer1 and CaSR. Homer1 and the mature, glycosylated isoform of CaSR were shown to exist in a protein complex with mTOR and the mTORC2-specific protein Rictor by immunoprecipitation, and the level of CaSR and Rictor protein in the complex with Homer1 was significantly increased in response to increased extracellular calcium. Homer1 and CaSR modulated Ca 2ϩ -induced AKT phosphorylation and the attendant translocation of ␤-catenin to the nucleus within 15 min of treatment. In MG63 osteosarcoma cells, which expressed both CaSR and Homer1, phosphorylation of AKT-Ser 473 and GSK3␤-Ser 9 were increased in response to extracellular Ca 2ϩ , whereas siRNA directed at either Homer1 or CaSR attenuated these effects. As predicted from the above signaling studies, both Homer1 and CaSR positively modulated osteoblast differentiation and resistance to stress-induced apoptosis. Taken together, these studies have established a novel CaSR-Hom-er1-mTORC2-AKT-signaling pathway in osteoblasts with functional consequences for cellular maturation and survival. In addition, we were able to establish reconstituted AKT-Ser 473 sensitivity to extracellular Ca 2ϩ in a nonosteoblast cell line. Thus, independent transfections of either Homer1 or CaSR into HEK-293 cells showed no AKT-Ser 473 sensitivity to extracellular Ca 2ϩ , however co-transfection of Homer1 and CaSR established AKT-Ser 473 sensitivity to extracellular Ca 2ϩ .

Discussion
Phosphorylation of AKT at Ser 473 is mediated by mTORC2 (9), and our earlier studies showed that activation of the CaSR induces AKT Ser 473 phosphorylation and nuclear ␤-catenin translocation in osteoblasts (6). This study extends these findings to show that Homer1 acts as a previously unrecognized link between the CaSR and mTORC2 upstream of AKT-Ser 473 (Fig. 1). The proposal that Homer proteins, which primarily reside at the plasma membrane (21-23), promote AKT phosphorylation is consistent with the recent finding that mTORC2 assembles at the plasma membrane prior to phosphorylating AKT at Ser 473 (13).
In this study, we observed that in the mineralized bone of WT mice areas exhibiting high levels of CaSR expression also exhibited high levels of Homer1 expression, and confocal microscopy of primary human osteoblasts demonstrated co-localization at the cellular level. Homer1 mediates PIKE-L-dependent activation of PI3K upstream of AKT-Thr 308 phosphorylation in hippocampal neurons (22), but a mechanism dependent on Homer1 and mTORC2 has not been described

Homer1 links CaSR to mTORC2
previously for the activation of AKT-Ser 473 . We observed that CaSR-mediated AKT-Ser 473 phosphorylation, the resultant downstream phosphorylations of GSK3␤ and ␤-catenin, and the attendant nuclear translocation of ␤-catenin are all dependent on Homer1. These observations provide a molecular explanation for CaSR-stimulated stabilization of ␤-catenin in osteoblasts ( Fig. 1) (6). The requirement for CaSR and Homer1 to observe increased phosphorylation of both AKT-Ser 473 and GSK3␤-Ser 9 in response to extracellular Ca 2ϩ was also shown to be present in another commonly used osteoblast cell type, MG63 osteosarcoma cells. It is not yet clear whether the interaction between the CaSR and Homer1 arises from a direct binding interaction or indirectly via an intermediate protein.
In addition to the class C GPCRs mGluR1 and mGluR5 (21)(22)(23)29), Homer1 interacts with the scaffolding protein SHANK1 as well as components of the actin cytoskeleton and cell signaling apparatus (27). For some of these interacting proteins, the Homer1-binding sequences have not yet been identified. A widely recognized consensus binding sequence PPXXF (25,40), which is present in the C termini of both mGluR1 and mGluR5, is not present in the CaSR. However, some related sequences are present in the CaSR, including PTSF in intra-loop2 (residues 711-714; Uniprot P41180), and PENF in intra-loop3 (residues 799 -802), which conform to a proposed variant of the consensus Hom-er1-binding sequence PXXF (22). Interestingly, the latter sequence is conserved in other members of GPCR family C, including mGluR1 and mGluR5 (Uniprot Q13255 and P41594, respectively). Two other proteins that interacted with the Homer1-CaSR complex in this study, mTOR and Rictor (Fig.  2), also contain Homer-binding sequences. mTOR contains the classical Homer1-binding sequence, PPKDF (residues 464 -468, Uniprot P42345), and Rictor contains the variant sequence, PKGF (residues 812-815, Uniprot Q6R327).
We previously found that AKT stimulation in osteoblasts, downstream of the activated CaSR, promoted ␤-catenin stabilization and subsequent nuclear translocation that was dependent, in part, on inhibition of GSK3␤ activity via Ser 9 phosphorylation and that chemical inhibition of AKT by AKT-XI halted this process (6). We confirmed these findings in this study and observed that the inhibitory phosphorylation sites of GSK3 were also suppressed by the mTOR kinase inhibitor Torin1 but not the PI3K inhibitor wortmannin (Fig. S2), highlighting the relative importance of mTORC2versus PI3K-dependent control of AKT.
Increases in the osteoblast differentiation marker ALP in response to extracellular Ca 2ϩ were suppressed by silencing both CaSR and Homer1 in this study. It is of note that basal levels of ALP were suppressed by Homer1 silencing but not CaSR silencing. One interpretation of this finding is that Homer1 performs functions essential to cell differentiation, other than those that respond to increases in extracellular Ca 2ϩ . Clearly, CaSR is essential to normal osteoblast differentiation (1, 2); however, the role of Homer1 is not known in bone cells. To further investigate, we performed a direct comparison between silencing CaSR and Homer1 on osteoblast viability in response to hydrogen peroxide-induced stress as measured by the viability dye Cell Titer Blue (Fig. S3). Silencing Homer1 reduced osteoblast viability, and the effect exceeded that observed in cells in which the CaSR had been silenced. The results of this experiment are consistent with the hypothesis that Homer1 is a key mediator of prosurvival pathways in bone cells and warrants further investigation.
Resistance to oxidative stress promotes bone quality in vivo (41). Reducing the level of either CaSR or Homer1 with siRNA significantly reduced the ability of elevated extracellular Ca 2ϩ to support cell survival in the context of oxidative stress or sparse culture conditions. Hydrogen peroxideinduced apoptosis increased in response to Torin1, which inhibited mTORC2 and thus AKT-Ser 473 phosphorylation but was unaffected by wortmannin that inhibits PI3K, supporting the idea that mTORC2-induced phosphorylation of AKT, downstream of CaSR and Homer1, is a determinant of osteoblast survival.
The key observations of this study indicate the existence of a receptor-scaffolding protein-intracellular protein kinase mechanism in support of cell survival and differentiation in Ca 2ϩ -stimulated osteoblasts that requires the CaSR, Homer1, mTORC2, and AKT. To further explore whether these interactions might be generalizable to other cell types, we investigated whether it might be possible to reconstitute extracellular Ca 2ϩstimulated AKT-Ser 473 phosphorylation in HEK-293 cells. Consistent with our primary cell findings, Ca 2ϩ -stimulated AKT-Ser 473 phosphorylation was not observed in HEK-293 cells that had been stably transfected with the CaSR or in HEK-293 cells that had been transiently transfected with Homer1c alone. However, in HEK-CaSR cells that were transiently transfected with Homer1c for the optimal period of 48 h, we observed reconstitution of extracellular Ca 2ϩ -stimulated AKT-Ser 473 phosphorylation that exhibited an EC 50 for Ca 2ϩ of 1.0 -2.0 mM and a maximal response at Ca 2ϩ o concentrations around 3-5 mM. HEK-293 cells have been reported previously to endogenously express both Rictor (42) and mTOR (42,43), and we used Western blot analysis to confirm this in this study (data not shown). These observations, together with the observed impact of Homer1 knockdown in osteoblasts and osteosarcoma cells, indicate that expression of Homer1 in CaSR-expressing cells is required for the establishment of Ca 2ϩ -stimulated AKT-Ser 473 phosphorylation.
In conclusion, this study demonstrates that in response to extracellular Ca 2ϩ , a novel protein complex, composed of CaSR, Homer1, and mTORC2, forms in human osteoblasts and mediates a pathway that controls the phosphorylation of AKT leading to increased differentiation and survival (Fig.  1). AKT phosphorylation supports an alternative pathway to the stabilization and nuclear translocation of ␤-catenin, a key driver of bone anabolism (15). This pathway may provide alternative targets for the development of new osteoporosis therapies. The findings demonstrate that in addition to roles in supporting neuromodulation and neuroplasticity by type-1 metabotropic glutamate receptors, Homer1 supports the control of bone cell survival and differentiation downstream of the CaSR, which monitors the levels of Ca 2ϩ in the bone microenvironment.

Methods
Cell culture-Human osteoblasts were grown from the minced trabecular ends of the fetal long bone in accordance with the National Health and Medical Research Council guidelines and with the approval of the University of Sydney Human Ethics Committee (approval number 2012/043), as described previously (6,(44)(45)(46). Each experiment was conducted using human osteoblasts from at least two different donors, and cells were maintained routinely in DMEM containing 10% (v/v) FCS. These bone-derived cells were moderately differentiated as described previously (6,44). Because the bone cells were from different donors, some biological variations in responses were observed, as expected (6,44,45). HEK-293 (ATCC CRL-1573) and MG-63 osteosarcoma cells (ATCC CRL-1427) were maintained in DMEM containing 10% (v/v) FCS. The stablyexpressed WT human CaSR in HEK-293 cells was the kind gift of Dr. Karen Krapcho and Dr. Edward Nemeth (NPS Pharmaceuticals).
CaSR knockout mice-Male and female mice with ablation of both alleles of the Casr gene in a C57/B6 background were generated by breeding the floxed-CaSR mice and the mice expressing Cre recombinase transgene under the control of a 2.3-kb promoter of rat type I collagen ␣-subunit as described previously (1). All mice were housed in a pathogen-free climate-controlled room (22°C; 50% relative humidity) with a 12:/ 12-h dark/light cycle, given filtered water and standard chow containing 1.0% calcium and 0.7% phosphate, and euthanized at 3 months of age by isoflurane overdose for subsequent harvests of serum and bone samples. All animal procedures were approved by the Institutional Animal Care and Use Committee at the San Francisco Veterans Affairs Medical Center. Femurs and tibiae from WT and heterozygous and homozygous osteoblast-specific CaSR knockout mice (1) were used for immunofluorescent (IF) and Western blotting analyses of either CaSR, Homer1, or OPG. Femurs were snap-frozen and kept at Ϫ80°C prior to protein extraction for Western blot analysis. Tibiae were cleaned, fixed in 10% formalin, decalcified, and sectioned at 8 m-thick for IF. Sections were stained for IF as described below.
HEK-293 transient transfection-Cells were transiently transfected using the X-tremeGENE HP DNA transfection reagent (Roche Applied Science) under serum-free conditions and were used for subsequent experiments 48 h after transfection. The pRK5-Homer1c construct was the kind gift of Dr. RaoZhu Lin and Prof Paul Worley (Johns Hopkins School of Medicine).
siRNA transfection-Each siRNA used in this study was a commercially supplied (Santa Cruz Biotechnology) pool of three target-specific 19 -25-nucleotide siRNAs designed to knock down expression of the specific gene of interest. siRNA sequences are listed in Table S1. siRNAs against all target transcripts were used at a concentration of 50 nM in the presence of siRNA transfection reagent (Santa Cruz Biotechnology) according to the manufacturer's instructions. Control transfections containing a nondirected siRNA sequence were carried out simultaneously, on the same plate, for all experiments. Briefly, osteoblasts were transfected overnight under serumfree conditions and permitted to grow for an additional 24 h in complete growth media after transfection, which allowed efficient reduction to the level of the targeted protein. Protein knockdown was confirmed by Western blotting.
Immunofluorescence (human osteoblasts)-Human osteoblasts were grown on poly-L-lysine-coated coverslips and were fixed with 4% (w/v) paraformaldehyde followed by 100% icecold methanol and were processed with the following antibodies: anti-CaSR (Sigma, mouse monoclonal, clone HL1499), anti-Homer1 (Santa Cruz Biotechnology, rabbit polyclonal, clone H-174), and isotype controls. Coverslips were then washed and incubated with anti-rabbit Alexa Fluor 488 (1:750; Santa Cruz Biotechnology) and anti-mouse Cy3 (1:750; Life Technologies, Inc.) at room temperature for 60 min. To visualize the nuclei, coverslips were mounted with UltraCruz TM mounting medium containing DAPI. Slides were observed under a LSM510 Meta confocal laser microscope (Zeiss) at the Advanced Microscopy Facility (Bosch Research Institute, University of Sydney).
Images for co-localization analysis were taken using the ϫ63 objective under oil. Background threshold levels were determined using ImageJ software (W. Rasband, National Institutes of Health), and Pearson's co-localization coefficients (47,48) were calculated using the Coloc2 plugin of ImageJ.

Homer1 links CaSR to mTORC2
Protein extraction from mouse bones-Following the removal of bone marrow, total protein was extracted from mouse bones using a combination of de-mineralization and chemical denaturation as described previously (49). Protein extracts were precipitated from their chemical denaturation solutions, and equal amounts of protein were subjected to Western blot analysis, and densitometry was carried out on triplicate bands.
Alkaline phosphatase activity-Following treatment, human osteoblasts in triplicate wells of 96-well plates were lysed in PBS containing 0.1% (v/v) Triton X-100. An equal volume of 0.2 M glycine (pH 9.5) containing 1 mg/ml 4-pNPP (ALP substrate) was added to start the enzymatic reaction. The reaction was simultaneously carried out with calf intestinal ALP (0 -1 unit) under the same conditions to construct a standard curve. Identically treated wells on the same plate were used for cell number correction determined by total cellular protein (BCA).
Caspase-3 activity-Following treatment, apoptosis was assessed in human osteoblasts in triplicate wells of 96-well plates via caspase-3 activity with the peptide substrate Ac-DEVD-AFC (BD Biosciences) used according to the manufacturer's instructions. Identically treated wells on the same plate were used for cell number correction determined by total cellular protein (BCA).
Nuclear localization of ␤-catenin-Detection of ␤-catenin localized to the nucleus was performed as described previously (6). Briefly, following treatment, osteoblast monolayers were lysed in 10 mM Tris (pH 7.5), 10 mM NaCl, 3 mM MgCl 2 , 0.5% (v/v) Nonidet P-40, with protease inhibitors for 5 min on ice. Lysates were centrifuged at 500 ϫ g for 30 min at 4°C, and the pellet washed three times with lysis buffer. Insoluble pellet was resuspended in a high salt buffer (10 mM Tris (pH 7.5), 0.4 M NaCl for 30 min on ice, centrifuged at 12,000 ϫ g for 30 min at 4°C, and the resulting supernatant represented the nuclear fraction. Protein concentrations were normalized between treatments (BCA), and equal amounts of protein were subjected to Western blot analysis. LaminB1 was used as a loading control for the nuclear fraction.
Co-immunoprecipitation-Human osteoblasts in triplicate wells of 6-well plates were exposed to control (1.0 mM) or elevated (3.5 mM) Ca 2ϩ concentration for 2 min, immunoprecipitated with anti-Homer-1-b/c using a variation of a protocol previously described (50). Briefly, following treatment, the cell monolayer was lysed on ice in 50 mM Tris (pH 7.5), 250 mM NaCl, 1% (v/v) Nonidet P-40 with protease inhibitors, and lysates from triplicate wells were combined and processed through a 21-gauge needle five times. Insoluble material was removed by centrifugation, and the lysate was pre-cleared with protein-A/G PLUS-agarose beads (20 l/ml). Following protein concentration measurement, immunoprecipitation of 1 mg of total cellular lysate with 1 g/ml monoclonal anti-Homer1 (Santa Cruz Biotechnology, rabbit polyclonal, #H-174) (or isotype control) was carried out overnight at 4°C with protein-A/G PLUS-agarose (20 l/ml) with orbital mixing. Beads were extensively washed in lysis buffer prior to detection of immunoprecipitated protein by Western blotting under reducing conditions using VeriBlot TM secondary antibodies (Abcam). As with other Western blottings, this experiment was performed three times for quantification.
OPG protein-OPG protein levels in vitro were detected by whole-cell ELISA. OPG protein levels in the supernatant of primary osteoblasts were routinely at the lower limit of detection by capture ELISA, but the protein was readily found to be associated with the cell monolayer. For this reason, a whole-cell ELISA was developed. Following treatment, cells were fixed with 10% neutral buffered formalin for 5 min and then permeabilized with three incubations with PBS, 0.1% (v/v) Triton X-100 for 5 min each. Endogenous peroxidase was inactivated with 0.6% H 2 O 2 in PBS for 15 min. Wells were blocked with 5% (w/v) BSA in PBS for 30 min. Detection was with anti-OPG mAb (mouse monoclonal, clone #69127, R&D Systems) or an isotype control for background correction, in blocker at 4°C on a rocking platform overnight followed by anti-Ig-horseradish peroxidase conjugated at 1 g/ml in blocker for 1 h at room temperature and subsequent 3,3Ј,5,5Ј-tetramethylbenzidine detection. All steps were separated by five PBS, 0.01% (v/v) Tween 20 washes.
qPCR protocol for Wnt SuperArray-We used the reverse transcription (RT 2 ) Profiler PCR Array Mouse Wnt signaling pathway array purchased from SuperArray Bioscience (Frederick, MD). The complete gene list is available online. Total RNA from marrow-flushed mouse bones was isolated using the RNA Stat-60 (Amsbio, Cambridge, MA). First-strand cDNA synthesis reaction was performed as follows: 2 g of extracted RNA was mixed with 10 l of the SuperArray RT mixture mix. The products were then incubated at 37°C for 1 h and heated at 95°C for 5 min. Real-time-based SYBR Green PCR was performed using a ViiA7 real-time PCR system (Thermo Fisher

Homer1 links CaSR to mTORC2
Scientific), and the following thermal cycling condition was used: 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 60 s. Data analysis and the cycle threshold (CT) values, which were defined as the fractional cycle number at which the fluorescence passes an arbitrarily set threshold, were analyzed using the built-in ViiA7 analytic software. The CT value of each gene was normalized to that of ACTB, which is included in this commercially available kit.
CellTiter-Blue-The cell viability dye CellTiter-Blue (Promega) was used according to the manufacturer's instruction. Fluorescence of the dye was measured at Ex 560 /Em 590 .
Statistics and data analysis-Experiments were performed in triplicate. Each experiment was repeated at least three times, with cells from different donors, and the data are reported as means Ϯ S.D. One-way analysis of variance with the Tukey post-test were used to determine significant differences between treatments using Prism (GraphPad Software Inc).