Glucocorticoids Inhibit Cell Cycle Progression in Differentiating Osteoblasts via Glycogen Synthase Kinase 3ß

Differentiating osteoblasts in culture undergo a commitment stage, during which cobblestone-like cells grow to high density past confluency. In contrast to earlier proliferative stages, the cell cycle during this commitment stage is inhibited by glucocorticoids (GC). Chronic GC treatment also impedes mineral deposition if steroid administration commences early enough during commitment. This study defines a role for glycogen synthase kinase-3beta (GSK3beta) and its target, c-Myc, in the GC-sensitive osteoblast persistent cell cycle. c-Myc levels decreased as cells reached confluence, but then increased during growth to high density. GC administration at this stage resulted in down-regulation of c-Myc. This was accompanied by GC-mediated attenuation of GSK3beta Ser(9) inhibitory phosphorylation and increased GSK3beta kinase activity. Down-regulation of c-Myc was attributable to enhanced Thr(58) phosphorylation, leading to accelerated degradation. In contrast, GC did not inhibit the c-Myc synthesis rate or the level of beta-catenin, a transcriptional coactivator of c-myc. The attenuated cell cycle and the reduced c-Myc level were returned to control levels by specific inhibition of GSK3beta using lithium chloride. These results suggest that tonal GSK3beta repression at the cobblestone stage of osteoblast differentiation permits osteoblast growth to high density. GC interfere with this growth-permissive axis by GSK3beta activation, resulting in c-Myc down-regulation and impediment of the G(1)/S cell cycle transition.


Results
GC activate GSK3ß by attenuating its inhibitory Ser 9 phosphorylation. We have recently shown that GC inhibit persistent cell cycle progression at a commitment stage during osteoblast differentiation, characterized by cobblestone appearance. We postulated that components of the Wnt signaling pathway might play a role during this commitment stage in mediating cell cycle persistence and its inhibition by GC. One of these components is GSK3ß, a cell cycle guardian, which is inactivated by a variety of growth stimuli. We evaluated GSK3ß in non-treated and GC-treated MC3T3-E1 cells at and immediately after the "cobblestone" stage. As determined by Western analysis, DEX-treatment of cells that had just become confluent resulted in a small but consistent increase in the total cellular levels of GSK3ß ( Figure 1A). Because nuclear localization of GSK3ß has recently been proposed to play a cell cycle-related role (22), we also performed Western analysis on nuclear extracts, and found that the increase in GSK3ß level was even more pronounced in the nuclear fraction of DEX-treated cells ( Figure 1A).
Notably, this effect was attenuated in cultures that had already grown to high density ( Figure 1A, day 9). Accordingly, we thereafter confined our work specifically to cells that just initiate growth to high density.
GSK3ß activity is subject to functionally critical post-translational modifications. We therefore measured GSK3ß activity by immunoprecipitation and kinase assays using a glycogen synthase-derived peptide as substrate. As exemplified in Figure 1B An important post-translational modification of GSK3ß is the phosphorylation of serine-9, which results in GSK3ß inhibition, such as seen following growth factor stimulation (14)(15)(16). We therefore evaluated the effect of GC on GSK3ß Ser 9 phosphorylation. Indeed, as shown in Figure 1B, DEX decreased the level of [phospho- Higher activity of GSK3ß, a cell cycle guardian, would result in cell cycle impediment, as observed in GC-treated post-confluent MC3T3-E1 osteoblasts (4).

GC down-regulate c-MYC in MC3T3-E1 cells in a ß-catenin-independent fashion.
An important downstream target of the Wnt signaling pathway is ß-catenin, which participates in the transcriptional activation of cell cycle stimulatory genes, such as c-myc (9) and cyclin D1 (10). GSK3ß-mediated phosphorylation promotes the degradation of ßcatenin. We therefore examined if ß-catenin was down-regulated in GC-treated MC3T3-E1 cells. However, as shown in Figure 1C, DEX did not decrease either the nuclear or the cytoplasmic ß-catenin level, suggesting that ß-catenin is resistant to GC-mediated activation of GSK3ß in MC3T3-E1 osteoblasts. This is consistent with the notion that some GSK3ß exists in pools that are physically and functionally distinct from those found complexed with APC, axin and ß-catenin (see ref. (25) for review).
GSK3ß can down-regulate cell cycle regulatory factors, such as c-MYC and cyclin D1, in both ß-catenin-dependent and -independent manners. The ß-catenin-independent pathway entails direct phosphorylation of c-MYC on Thr 58 followed by ubiquitination and proteasomal degradation (26,27). Indeed, despite the persistence of ß-catenin in GCtreated MC3T3-E1 cells, Western blot analysis of c-MYC revealed reduced levels in DEX-treated compared to non-treated cultures ( Figure 1D). In three independent experiments, c-MYC levels were significantly decreased by 2.3±0.2 fold.
Osteoblast growth to high density is associated with c-MYC upregulation. c-MYC has been previously found to express in osteoblast and chondrocyte in vivo (28,29). We therefore analyzed c-MYC by Western blot analysis during MC3T3-E1 osteoblast differentiation in vitro. Expectedly, as cells reached confluence (Figure 2A . This was not seen with the NIH3T3 cells (data not shown). In contrast to c-MYC, cyclin A, which is down-regulated by GC in post-confluent persistently cycling osteoblasts (4), is expressed at steady levels during pre-confluence, confluence, and until at least two days after confluence ( Figure 3C).
Taken together, this expression pattern and the effect of DEX ( Figure 1D Figure 1D) is attributable to protein instability. We therefore studied c-MYC synthesis and degradation in GC-treated cultures. Initially, we evaluated c-MYC synthesis rates by blocking degradation using the proteasomal inhibitor MG132; this was immediately followed by concomitant MG132 withdrawal and administration of cyclohexemide to block protein synthesis during the measurement of c-MYC degradation rate. As shown in Figure  GSK3ß has been shown to target c-MYC for proteasomal degradation by phosphorylation on Thr 58 (26,27). We therefore tested the c-MYC-Thr 58 phosphorylation status in DEXtreated versus non-treated cells. Samples from 4-hour MG132-treated cultures (as in Figure 3A) were subjected to Western analysis using a [phospho-Thr 58 ]c-MYC-specific antibody. As shown in Figure 3C, DEX increased c-MYC Thr 58 phosphorylation level.
Corrected for total c-MYC, the relative DEX-mediated increase in c-MYC Thr 58 phosphorylation was 1.7-and 2.6-fold in two independent experiments. These results are consistent with the insensitivity of ß-catenin to GC ( Figure 1C), and strongly suggest that increased c-MYC phosphorylation by GSK3ß in GC-treated osteoblasts leads to accelerated degradation and reduced c-MYC steady state levels.

GC effects on GSK3ß and c-MYC precede alterations in the cell cycle profile. The
increased GSK3ß activity in GC-treated cells ( Figure 1B) may be a result, not a cause of cell cycle attenuation (30). We therefore tested GSK3ß activity following short periods of GC treatment, before an effect on the cell cycle is mounted. Triton extracts of 2-hour and 5 hour-treated cultures were subjected to GSK3ß immunoprecipitation and kinase assay. These short treatment periods did not alter the distribution of cells among the phases of the cell cycle (data not shown). As shown in Figure 4, top panel, DEX activated GSK3ß within as little as two hours of treatment, with somewhat lesser activation observed at 5 hours.
Next, Triton extracts as above were subjected to Western blot analysis using antibodies to either GSK3ß or [phospho-Ser 9 ]GSK3ß. As shown in Figure 4, two-hour DEX treatment significantly reduced [phospho-Ser 9 ]GSK3ß, and this reduction amounted to 47±5% in three independent experiments. Corrected for total GSK3ß, the change in the relative Ser 9 phosphorylation level [calculated, as in Figure 1B, as the ratio between pGSK3ß(DEX)/pGSK3ß(control) and GSK3ß(DEX)/GSK3ß(control)] was significantly reduced within the 2-hr treatment period by 46±20% (n=3). The glucocorticoid effect on the relative Ser 9 phosphorylation of GSK3ß was similar at the 5-hr time point, amounting to a significant 50±18% reduction (n=3). Because Ser 9 phosphorylation is a major (albeit not the only (31)), determinant influencing GSK3ß activity, our data suggest that the fast decrease in GSK3ß Ser 9 phosphorylation following 2 hours of treatment may be the However, the lithium effect on the DEX-treated cells was more pronounced, bringing, within 9 hours, the cells in S+G2+M to virtually the same percentage as measured in the DEX-free cultures. Figure 6B shows representative cell cycle profiles from this 9-hour time point. As can be seen, 9-hour treatment was clearly sufficient for DEX to induce a GC-inhibition of GSK3ß Ser 9 phosphorylation requires receptor dimerization. As a first step towards elucidating mechanisms by which GC activate GSK3ß, we asked whether GC receptor (GR) dimerization was required for this activation. Receptor dimerization is a pre-requisite for DNA binding and classical transcriptional activation, but is not required for DNA-binding-independent action, such as AP-1 trans-repression (34). We employed SAOS-2 human osteosarcoma cell lines (originally GR-negative), which had been stably transfected with either the wild type or a dimerization defective (R479D/D481R) rat GR (35). Each of these cell lines was treated with DEX, and GSK3ß Ser 9 phosphorylation status evaluated by Western blot analysis. Similar to MC3T3-E1 osteoblasts ( Figure 1B), DEX significantly inhibited GSK3ß Ser 9 phosphorylation in SAOS-2 cells expressing wild type GR (Figure 7). In contrast, DEX did not inhibit GSK3ß Ser 9 phosphorylation in SAOS-2 cells expressing the dimerization-defective GR ( Figure 7). These results indicate that DEX-mediated GSK3ß activation requires receptor dimerization, suggesting that it occurs via classical transcriptional activation of genes yet to be identified.

Discussion
Cell proliferation and differentiation are traditional1y perceived as reciprocal processes, cell cycle withdrawal being a prerequisite for terminal differentiation. However, stages that precede expression of an ultimate cellular phenotype may occur as cells are cycling, such as during the differentiation of lymphocytes (36), adipocytes (5,6) and osteoblasts (4). Our earlier work with osteoblasts suggest that the differentiation-related cell cycle, How do GC exert their inhibitory effects on c-MYC and on the osteoblast persistent cell cycle? In search for upstream effectors, it is important to remember that the osteoblast cell cycle becomes sensitive to GC only at confluence. It therefore seemed logical that the GC-inhibition may involve the Wnt signaling pathway, which bridges cell-cell and cell-matrix interactions with the cell cycle machinery. Interestingly, what we observe in GC-treated cultures is increased intrinsic activity of GSK3ß, but not decreased nuclear ßcatenin levels, suggesting that GC primarily affect GSK3ß in the growth factor pathway.
However, we feel it is premature to rule out an effect of GC on GSK3ß in the Wnt pathway. In fact, GC decreased reporter activity in cells transiently transfected with a Tcf/LEF-controlled luciferase construct (data not shown). Thus, we believe that GC affect GSK3ß in both the Wnt and the growth factor pathways. Importantly, inhibition of GSK3ß by lithium neutralizes the inhibitory effects of GC, indicating causal relationship between GSK3ß activation and cell cycle attenuation.
The lack of GC-mediated decrease in nuclear ß-catenin is consistent with the unaffected c-MYC synthesis rate, which is regulated by ß-catenin (9. 42,43). However, GSK3ß may work directly on c-MYC, phosphorylating and accelerating its degradation (26,27). This mechanism clearly contributes to the antimitogenic effect of GC in differentiating osteoblasts, as indicated by the increased Thr 58 phosphorylation and the accelerated degradation of c-MYC in response to GC treatment. Preferential effect of GC on GSK3ß in the growth factor pathway versus the Wnt pathway may be explained by selective requirements for substrate priming (44,45).
GSK3ß is a cell cycle guardian, which is inactivated in response to growth factors, typically by Ser 9 phosphorylation. Our results show that GC attenuate this inactivating phosphorylation. GC decreased GSK3ß Ser 9 phosphorylation in cultures 2 hours ( Figure   4B) as well as 48 hours ( Figure 1B) after feeding fresh medium with or without DEX.
The GC effect on GSK3ß Ser 9 phosphorylation was also apparent during the immediate response to serum stimulation ( Figure 5). Thus, GC activate GSK3ß by prevention of the serum-induced Ser 9 inhibitory phosphorylation, which is sustained for as long as 48 hours.
A limitation of our study is that it employs cell lines, either transformed (SAOS-2) or untransformed (MC3T3-E1), which may be regulated differently than osteoblasts in vivo.
Preliminary results with primary murine bone marrow-derived mesenchymal stem cells indicated that, like in the MC3T3-E1 cultures, DEX inhibited both condensation and calcium deposition (data not shown). Also, GSK3ß Ser 9 phosphorylation was decreased in the DEX-treated cultures, but this effect did not reach statistical significance. It is possible that a DEX effect on the bone marrow-derived osteoblasts was diluted by different responses of non-osteoblasts and/or osteoblasts at DEX-insensitive stages of differentiation. Thus, the role of GSK3ß in osteoblast differentiation and in glucocorticoid-induced osteoporosis remains to be confirmed in primary cultures and in vivo.
Among cyclins, cyclin-dependent kinases and cdi's that were analyzed, GC's most significant effect in MC3T3-E1 differentiating osteoblasts was down-regulation of cyclin A and its dissociation from E2F4/p130 complexes (4). It is possible that GC downregulate c-MYC and cyclin A via independent mechanisms, and that both of these effects contribute to attenuation of the cell cycle. However, the present study demonstrates rapid inhibitory effects of DEX on c-MYC, which occurred prior to any significant effect on cyclin A (data not shown). Furthermore, rescue of the GC-inhibited cell cycle by lithium was accompanied by restoration of c-MYC, but not cyclin A levels ( Figure 6B). These results suggest a more crucial role for c-MYC in GC-inhibition of the osteoblast persistent cell cycle. However, less than 2 hours of treatment was insufficient for DEX to mount any effect on c-MYC or GSK3ß (data not shown). This, and the requirement for GR dimerization for the Ser 9 -dephosphorylation of GSK3ß (Figure 7), suggests that GC- hyperparathyroidism, but no hypercalcemia or bone loss in these patients (46); one study reported increased levels of alkaline phosphatase (47), a serum marker of bone formation.
Preserved bone mass in the face of hyperparathyroidism has been attributed to enhancement of renal calcium reabsorption (46,48). Based on our in vitro studies, it is interesting to speculate that enhanced osteoblast function may also contribute to bone preservation in lithium-treated patients with hyperparathyroidism.
Maintenance of bone mass throughout life requires that bone formation equals bone resorption. GC impair this balance primarily by inhibiting osteoblastic bone formation.
The present study demonstrates that GC activate GSK3ß in osteoblasts. This may interfere with important differentiation cues generated by cell-cell contacts, cell-matrix interactions, or by growth factors signaling through GSK3ß, in particular IGF-I, known to play a role in osteoblast growth and differentiation (49). Our current findings suggest    MC3T3-E1 cells were treated on day 6 with either 1 µM DEX or vehicle and harvested following either 2 or 5 hours for preparation of Triton whole cell extracts. The same extracts were subjected to either GSK3ß kinase assays (Top) as in Figure 1B, or Western blot analysis using the indicated antibodies.