Glucocorticoids inhibit the transcriptional activity of LEF/TCF in differentiating osteoblasts in a glycogen synthase kinase-3beta-dependent and -independent manner.

Glucocorticoids, widely used as immune suppressors, cause osteoporosis by inhibiting bone formation. In MC3T3-E1 osteoblast-like cultures, dexamethasone (DEX) activates glycogen synthase kinase-3beta (GSK3beta) and inhibits a differentiation-related cell cycle that occurs at a commitment stage immediately after confluence. Here we show that DEX inhibition of the differentiation-related cell cycle is associated with a decrease in beta-catenin levels and inhibition of LEF/TCF-mediated transcription. These inhibitory activities are no longer observed in the presence of lithium, a GSK3beta inhibitor. DEX decreased the serum-responsive phosphorylation of protein kinase B/Akt-Ser(473) within minutes, and this inhibition was also observed after 12 h. When the phosphatidylinositol 3-kinase (PI3K)/Akt pathway was inhibited by wortmannin, DEX no longer inhibited beta-catenin levels. Furthermore, DEX-mediated inhibition of LEF/TCF transcriptional activity was attenuated in the presence of dominant negative forms of either PI3K or protein kinase B/Akt. These results suggest cross-talk between the PI3K/Akt and Wnt signaling pathways. Consistent with a role for Wnt signaling in the osteoblast differentiation-related cell cycle, wortmannin partially negated the DEX inhibition of this cell cycle. DEX also induced histone deacetylase (HDAC) 1, which is known to inhibit LEF/TCF transcriptional activity. Overexpression of HDAC1 negated the inhibitory effect of DEX on LEF/TCF transcriptional activity. In the presence of trichostatin A, a deacetylase inhibitor, DEX-mediated inhibition of the differentiation-related cell cycle was partially negated. When administered together, wortmannin and trichostatin A completely negated the inhibitory effect of DEX on the differentiation-related cell cycle. These results suggest that inhibition of a PI3K/Akt/GSK3beta/beta-catenin/LEF axis and stimulation of HDAC1 cooperate to mediate the inhibitory effect of DEX on Wnt signaling and the osteoblast differentiation-related cell cycle.

GCs are used in combination with other drugs to reduce inflammation associated with hematologic and other cancers and to prevent implant rejection. A major side effect of GC treatment is rapid bone loss and increased risk of fracture (1). Although bone resorption has been suggested to contribute to GC-induced osteoporosis (2), it is now clear that the main mechanism underlying long-term GC-induced bone loss is the impairment of osteoblast function and bone formation (3)(4)(5).
One of the most effective culture systems for the investigation of inhibitory effects of GCs on osteoblasts is the MC3T3-E1 model. In fact, this is the only reported culture system in which GCs strongly inhibit mineralization (6,7). We have refined a commitment stage during osteoblast differentiation in MC3T3-E1 cultures, during which the cells are susceptible to GC inhibition (8). At the commitment stage, unique cell cycle machinery is activated. Unlike cell cycle progression in preconfluent cultures, the differentiation-related cell cycle is inhibited by GCs, concomitant with the abandonment of the differentiation program (7).
In our pursuit of mechanisms driving the differentiationrelated post-confluent cell cycle, we concentrated on the Wnt signaling pathway. The second messenger engaged by the Wnt signaling pathway to activate cell cycle-regulatory genes such as c-myc (9) is ␤-catenin, which also plays a role in cell-cell and cell-matrix interactions (10,11). Wnt signaling may therefore link cell-cell and cell-matrix interactions occurring in postconfluent cultures and the cell cycle. Importantly, Wnt proteins have been shown to promote post-confluent proliferation in cultured cells (12). We hypothesized that the Wnt signaling pathway plays a critical role in the osteoblast differentiationrelated cell cycle and in the aforementioned inhibitory effects of GCs (13). A key negative regulator in the Wnt signaling pathway is the serine/threonine kinase GSK3␤, which phosphorylates and promotes the degradation of ␤-catenin in quiescent cells. The tumor suppressors APC and axin form complexes with GSK3␤ and ␤-catenin, thus promoting the activity of GSK3␤ on ␤-catenin (14 -17).
In addition to the phosphorylation of ␤-catenin, GSK3␤ may also impede cell cycle progression by phosphorylating and marking for degradation substrates outside the Wnt signaling pathway, including the proto-oncogenes c-Myc and cyclin D1 (18,19). These substrates must first undergo priming by another kinase in order to correctly associate with GSK3␤ (20). Stimulation of the cell cycle with growth factors that activate the phosphatidylinositol 3-kinase (PI3K)/Akt pathway results in phosphorylation of GSK3␤ on Ser 9 (21)(22)(23). [pSer 9 ]GSK3␤ is unable to correctly associate with substrates that had been phosphorylated by the priming kinase and is thus inactive with respect to these substrates (24). Recently, it has been shown that, similar to substrates outside the Wnt signaling pathway, ␤-catenin also undergoes priming before it is phosphorylated by GSK3␤ (25).
The inhibition of the differentiation-related cell cycle in GCtreated MC3T3-E1 osteoblast cultures is associated with activation of GSK3␤, inhibition of its Ser 9 phosphorylation, and accelerated degradation of c-Myc (13). Furthermore, administration of lithium, a specific inhibitor of GSK3␤, to GC-treated MC3T3-E1 cultures rescued c-Myc and the differentiation-related cell cycle (13). Here we show that the GC-mediated activation of GSK3␤ also results in inhibition of the Wnt signaling pathway and LEF/TCF transcriptional activity. We demonstrate that GCs inhibit phosphorylation of protein kinase B (PKB)/Akt on Ser 473 and thus disrupt input from the PI3K/Akt pathway into the Wnt signaling pathway at the level of GSK3␤.

EXPERIMENTAL PROCEDURES
Cell Culture-Three culture models were employed in this study. A highly osteoblastic MC3T3-E1 subclone was isolated previously (7) and used between passage 7 and 11. The results did not depend on the passage number within this range. Newborn mouse calvarial osteoblast (NeMCO) cultures were prepared from 6-day-old mice by successive collagenase digestion (26). Human osteoblast cultures were obtained previously from vertebrae of subjects undergoing neurosurgical procedures (27), and frozen cell stocks were thawed for the present study. All cells were maintained in ␣-minimal essential medium supplemented with 10% fetal bovine serum (Omega, Tarzana, CA). At confluence, 50 M ascorbic acid and 10 mM ␤-glycerophosphate were added to promote differentiation (7). Dexamethasone (DEX; 1 M), wortmannin (230 nM), and trichostatin A (TSA; 50 nM) were administered with the last medium change.
Transfection and Luciferase Assays-Cells were plated in 6-well plates (80,000 cells/well) and transiently transfected with a total of 5 g of plasmid DNA using the calcium phosphate co-precipitation method of Chen and Okayama (28). The cells were harvested 20 -48 h after feeding and lysed in "reporter lysis buffer" (Promega). Luciferase activity was determined using a microtiter plate luminometer (MLX Dynex Technologies), and effects of DEX were tested for statistical significance using Student's t test.
Protein Extracts-Proteins were extracted using the following procedures. High salt extraction was performed according to the method of Ito et al. (29). Briefly, cells were suspended in buffer containing 10 mM Tris-HCl (pH 8.0), 500 mM NaCl, 0.25 mM EDTA, 10 mM 2-mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride, 1 mM dithiothreitol, 10 g/ml aprotinin, 10 g/ml leupeptin, 10 mM NaF, and 0.1 mM Na 3 VO 4 . Suspensions were homogenized and centrifuged at 20,000 ϫ g for 1 h to remove cell debris, and the supernatant was used for Western analysis. Triton extraction was performed according to the method of Su et al. (30). Cells were suspended in buffer containing 20 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1% Triton X-100, 10% glycerol, 10 mM EDTA, and protease and phosphatase inhibitors as described above. Suspensions were homogenized, and extracts were centrifuged at 20,000 ϫ g for 30 min to remove cell debris. Nuclear extracts were prepared essentially according to Verona et al. (31). Cell pellets were suspended in two packed cell volumes of hypotonic buffer containing 10 mM HEPES (pH 7.5), 10 mM KCl, 3 mM MgCl 2 , 1 mM EDTA (pH 8.0), and protease and phosphatase inhibitors as described above. Cells were left to swell on ice for 10 min and then vortexed for 10 s and spun at 500 ϫ g for 5 min. The nuclear pellet was washed twice with hypotonic buffer and lysed in two nuclear pellet volumes of lysis buffer containing 100 mM HEPES (pH 7.4), 0.5 M KCl, 5 mM MgCl 2 , 28% glycerol, and protease and phosphatase inhibitors as described above. The nuclear suspension was homogenized and centrifuged at 20,000 ϫ g for 1 h to remove cell debris. Protein concentrations were determined using the micro BCA protein assay kit (Pierce).
Western Analysis-Protein (60 -100 g) was subjected to SDS-PAGE and transferred to a 0.2-m nitrocellulose membrane using a Mini Trans-Blot Transfer Cell (Bio-Rad). Immunodetection was performed using ECL (Amersham Biosciences) according to the manufacturer's recommendations, followed by exposure of the membrane to x-ray film. Results were quantitated by photodensitometry using an AlphaImager 2000 system (Alpha Innotech Corp.). For each experiment, the control value was set at 1 to calculate the relative signal in DEX-treated cells. Effects of DEX were tested for statistical significance using the Mann-Whitney test.
Flow Cytometry-Cell cycle analysis was performed according to the method of Darzynkiewicz et al. (32). Briefly, cells were lightly trypsinized, resuspended in Hanks' buffer, fixed in cold 70% ethanol, washed with Hanks' buffer, and suspended in 1 ml of Hanks' buffer containing 20 g/ml propidium iodide and 5 Kunitz units of DNase-free RNase A. Cell cycle profiles were determined using an EPICS® Profile Analyzer, and effects of DEX were tested for statistical significance using Student's t test.

GCs Inhibit LEF/TCF-mediated Transcription via GSK3␤-
We have shown previously that GC-mediated inhibition of the osteoblast differentiation-related cell cycle occurs via activation of GSK3␤ and proteasomal degradation of c-myc (13). Because GSK3␤ is also a negative regulator of the Wnt signaling pathway, we decided to test whether GCs inhibit LEF/TCFmediated activity, the transcriptional outcome of the Wnt signaling pathway. MC3T3-E1 osteoblasts were transfected with TBE 4 -luc, which contains four LEF/TCF binding sites that control transcription of a luciferase reporter gene (9). Postconfluent proliferating cells undergoing commitment were treated with DEX and subjected to luciferase assay. As shown in Fig. 1A, DEX inhibited LEF/TCF-mediated transcription by 2-fold, whereas TBEm 4 -luc, which contains four mutated LEF/ TCF sites, was not inhibited (Fig. 1C). Unlike cultures undergoing commitment, administration of DEX to pre-confluent proliferating MC3T3-E1 cells did not inhibit LEF/TCF-mediated transcription (Fig. 1B).
It has been shown that GSK3␤ facilitates the transcriptional activity of p65-NFB (35). If DEX inhibition of LEF/TCF-mediated activity were a result of GSK3␤ activation, one would expect that under the same conditions DEX would increase the transcriptional activity of p65-NFB. Indeed, as shown in Fig.  1E, DEX stimulated p65-NFB-responsive luciferase activity by 2.4-fold, whereas no significant effect was observed in preconfluent cultures (Fig. 1F).
To confirm the role of GSK3␤ in the DEX-mediated inhibition of LEF/TCF in differentiating osteoblasts, MC3T3-E1 cells transfected with TBE 4 -luc were treated with DEX in the presence of lithium, a specific inhibitor of GSK3␤. Indeed, in seven independent experiments, one of which is shown in Fig. 1G, the inhibition of LEF/TCF transcription by DEX was reduced by an average of 5-fold in the presence of lithium as compared with potassium, which was used as control. These results suggest that GSK3␤ mediates the inhibition of LEF/TCF transcriptional activity in GC-treated differentiating osteoblasts.
GCs Inhibit PKB/Akt Phosphorylation in Differentiating Osteoblasts-We have shown previously that GC activation of GSK3␤ is associated with a decrease in the inhibitory phosphorylation of GSK3␤-Ser 9 (13). The primary enzyme that phosphorylates GSK3␤ on Ser 9 is PKB/Akt, which is stimulated in response to serum and PI3K activation (23, 36 -38). Phosphorylation of PKB/Akt on serine 473 is a hallmark of enzyme activation (39). We therefore measured Akt-Ser 473 phosphorylation in DEX-inhibited osteoblast cultures. In MC3T3-E1 cells, the only reported osteoblast culture system in which GCs inhibit terminal differentiation, DEX decreased the phosphoryl-ation of Akt on Ser 473 by 2.3-fold ( Fig. 2A). Furthermore, serum stimulation of Akt Ser 473 phosphorylation was attenuated within minutes of DEX treatment (Fig. 2B). DEX also decreased Akt Ser 473 phosphorylation in newborn mouse calvarial osteoblast (NeMCO) and human osteoblast cultures ( Fig.  2A). These results suggest that in differentiating osteoblasts, the target of GCs upstream of GSK3␤ is the PI3K/Akt pathway.
GC Inhibition of LEF/TCF Transcriptional Activity Is Mediated through PKB/Akt-Taken together, the results described above led us to test whether the inhibition of the PI3K/Akt pathway contributes to the observed decrease in LEF/TCF transcriptional activity in GC-treated osteoblast cultures. Cells were transfected with TBE 4 -luc, along with expression plasmids encoding dominant negative forms of either PKB/Akt or PI3K. As shown in Fig. 2C, the DEX-mediated inhibition of LEF/TCF transcriptional activity in the differentiating osteoblasts was reduced from 3.6-fold to 2.4-fold when a dominant negative PKB/Akt was present. In the presence of dominant negative PI3K, the inhibitory factor was reduced to 1.9-fold (Fig. 2C, top panel). Specificity to LEF/TCF transcriptional activity is indicated by the results obtained in cells transfected with the TBEm 4 -luc reporter (Fig. 2C, bottom panel). Thus, inhibition of the PI3K/Akt pathway compromises, albeit partially, the inhibitory effect of DEX on LEF/TCF, indicating that the PI3K/Akt pathway mediates in part the inhibitory effect of DEX on LEF/TCF transcriptional activity.
The PI3K/Akt Pathway Mediates DEX-induced Decrease in ␤-Catenin-The observation that DEX inhibition of LEF/TCF transcriptional activity is counteracted by lithium (Fig. 1G) suggests that DEX may have abrogated Wnt signaling through activation of GSK3␤. To confirm this interpretation, we measured the effect of DEX on ␤-catenin levels in post-confluent MC3T3-E1 proliferating cultures. As shown in the left panels of Fig. 3, A and B, ␤-catenin levels were indeed reduced in response to DEX. Furthermore, ␤-catenin levels were not decreased by DEX in the presence of lithium (Fig. 3A, right  panel), indicating that GCs indeed reduce ␤ catenin via GSK3␤ activation.
The destructive activity of GSK3␤ on ␤-catenin is thought to be primarily regulated by Wnt ligands and disheveled, which destabilize the ␤-catenin destruction complex (14 -17). Here, we propose that the action of PKB/Akt on GSK3␤-Ser 9 also contributes to the regulation of ␤-catenin levels and would therefore mediate the inhibitory effect of GCs on the Wnt signaling pathway. To examine the interaction between the PI3K/Akt and Wnt signaling pathways, we tested whether DEX could still inhibit ␤-catenin levels in the presence of wortmannin, an inhibitor of the PI3K/Akt pathway. DEX no longer reduced ␤-catenin levels in the presence of wortamnnin, whereas inhibition was preserved in the presence of Me 2 SO used as vehicle (Fig. 3B, DMSO). These results demonstrate cross-talk between the PI3K/Akt and the Wnt signaling pathways, placing GSK3␤ at an intersection through which extracellular signals upstream of PI3K/Akt would cooperate with Wnt signaling. In differentiating osteoblasts, this intersection is blocked by GCs (see Fig. 7).
DEX Targets the Wnt Signaling Pathway in Primary Osteoblast Cultures-Although the MC3T3-E1 cell culture model is ideal for the investigation of adverse effects of GCs in osteoblasts, we also tested whether DEX reduces ␤-catenin levels in NeMCO cultures under conditions established by Noh and Frenkel. 2 DEX decreased ␤-catenin levels in NeMCO cultures (Fig. 3C) and in differentiating bone marrow-derived human mesenchymal stem cell cultures (data not shown) in a manner comparable with that seen in the MC3T3-E1 model (Fig. 3, A  and B). Furthermore, GSK3␤ mediated DEX attenuation of cell cycle progression in NeMCO cultures: after 9 h of treatment, the percentage of S-phase cells was insignificantly reduced by only 10%, followed by a 27% inhibition at the 15 h time point and a 7-fold inhibition after 5 days of treatment (Fig. 3D). When the NeMCO cultures were co-treated with lithium, a GSK3␤ inhibitor, DEX no longer attenuated cell cycle progression (Fig. 3D). Thus, the Wnt signaling pathway is intercepted by DEX in both MC3T3-E1 and NeMCO cultures.
Histone Deacetylase I Contributes to GC Inhibition of LEF/ TCF Transcriptional Activity-Although our data suggest that GCs inhibit LEF/TCF transcriptional activity through a PI3K/ Akt/GSK3␤/␤-catenin axis, an additional mechanism of GC action is suggested by the fact that DEX still has some inhibitory activity on LEF/TCF in cells treated with lithium ( Fig.  1G) or transfected with dominant negative forms of PI3K or PKB/Akt (Fig. 2C). Such a mechanism could involve acetylases and deacetylases, which directly activate and inactivate LEF/ TCF transcriptional activity, respectively (40 -45).
Initially, we tested the effect of the acetyltransferase CBP on DEX-mediated repression of LEF/TCF transcriptional activity. As shown in Fig. 4A, co-transfection of CBP did not influence the DEX-mediated repression of TBE 4 -luc. We also tested 2 T. Noh and B. Frenkel, manuscript in preparation. whether the DEX-mediated repression would be compromised when endogenous CBP is unable to co-activate LEF/TCF. To this end, cells were co-transfected with TBE 4 -luc along with p53, which has been shown to sequester CBP from interaction with LEF/TCF transcription factors (46). However, in the presence of p53, DEX still repressed LEF/TCF transcriptional ac-tivity to the same extent seen with the empty vector control (Fig. 4B), although the basal activity was reduced 2.2-fold by p53 (data not shown). These results suggest that CBP does not play a significant role in GC inhibition of LEF/TCF transcriptional activity in differentiating osteoblasts.
We next asked whether HDAC1, a LEF/TCF co-repressor (42)(43)(44)(45), could be involved in the DEX-mediated inhibition of LEF/TCF transcriptional activity. Co-transfection of HDAC1 suppressed TBE 4 -luc activity and attenuated the inhibitory effect of DEX (Fig. 5A). This suggested to us that GCs could have induced HDAC1 in differentiating osteoblasts. Indeed, as demonstrated by Western analysis, DEX-treated cells had 1.9fold more HDAC1 than did untreated cells (Fig. 5B). Thus, induction of HDAC1 likely contributes to DEX inhibition of LEF/TCF transcriptional activity.

Abrogation of the PI3K/Akt Pathway and Stimulation of HDAC Cooperate to Inhibit the Differentiation-related Cell Cycle in GC-treated Osteoblasts-
We have shown previously that in the MC3T3-E1 osteoblast model, DEX inhibits cell cycle progression specifically in post-confluent cultures undergoing commitment to mineralization (7). We hypothesized that this GC-sensitive differentiation-related cell cycle was primarily driven by the Wnt signaling pathway (13). In the preceding experiments, we showed that the inhibition of the PI3K/Akt pathway and induction of HDAC each contributed to GC-mediated inhibition of LEF/TCF transcriptional activity. If these activities of GCs are causally related to the inhibition of the cell cycle, then manipulation of the PI3K/Akt pathway and HDAC activity could be expected to negate the effect of GCs on the differentiation-related cell cycle at least partially. We therefore tested the effect of DEX on the cell cycle profile in the absence and presence of wortmannin, a PI3K inhibitor, and TSA, a HDAC inhibitor. As shown in Fig. 6A, wortmannin and DEX each reduced the percentage of post-confluent MC3T3-E1 cells in S phase from 15% to 2.9%. Importantly, whereas DEX reduced the percentage of post-confluent S-phase cells by 5.2-fold when the PI3K/Akt pathway was intact, DEX inhibition was attenuated to only 2.4-fold in the presence of wortmannin (Fig.  6C). These results suggest that the PI3K/Akt pathway is critically important for DEX repression of the osteoblast differentiation-related cell. Interestingly, wortmannin did not inhibit the percentage of cells in S-phase in pre-confluent MC3T3-E1 cultures (Fig. 6B). The residual DEX inhibition of the cell cycle in wortmannin-treated post-confluent cells (Fig. 6C) could have resulted from induction of HDAC1 (Fig. 5B). To address this possibility, we measured the effect of DEX on the cell cycle profile in the presence of both wortmannin and TSA. DEX no longer decreased the percentage of S-phase cells in the presence of both inhibitors (Fig. 6C), suggesting that HDAC stimulation contributes to DEX-mediated inhibition of the cell cycle. We conclude that inhibition of the PI3K/Akt pathway and stimulation of HDAC cooperatively mediate the inhibition of the osteoblast differentiation-related cell cycle by GCs (Fig. 7). DISCUSSION Mechanisms underlying GC-induced osteoporosis are poorly understood. The present study focuses on a commitment stage during development of the osteoblast phenotype in the nontransformed MC3T3-E1 model, in which cell cycle progression is sustained in post-confluent cultures (7). We have shown previously that specifically at this stage, GCs inhibit both the . Through PI3K and PKB/Akt, these growth factors stimulate the inhibitory phosphorylation of GSK3␤ on Ser 9 (P enclosed in a black octagon). In the absence of these growth factors, GSK3␤ phosphorylates c-Myc, marking it for degradation, and this is enhanced by GCs. GCs also inhibit Wnt signaling. First, they interfere with cross-talk between the PI3K/Akt and Wnt signaling pathways at the level of GSK3␤. Second, they stimulate HDAC1, which directly inhibits LEF/TCF transcriptional activity. CK-I␣, casein kinase-I␣; LRP-5, low density lipoprotein-related protein 5; DVL, disheveled. cell cycle and the progression toward terminal differentiation (7). We suggested that the Wnt signaling pathway plays an important role in the commitment-associated cell cycle (13). Here we show that abrogation of Wnt signaling mediates the inhibitory effect of GCs on the commitment-associated cell cycle during osteoblast differentiation. Two parallel mechanisms are employed by GCs: 1) stimulation of HDAC, and 2) inhibition of the PI3K/Akt pathway and activation of GSK3␤. Both mechanisms contribute to suppression of LEF/TCF transcriptional activity and attenuation of the G 1 -S cell cycle transition GSK3␤ may attenuate cell cycle progression by phosphorylating substrates in and outside of the Wnt signaling pathway (Fig. 7). Within the Wnt signaling pathway, GSK3␤ phosphorylates and promotes the degradation of ␤-catenin (14 -17). Outside the Wnt signaling pathway, GSK3␤ phosphorylates and promotes the degradation of substrates such as the protooncogene c-Myc (18). Progression of the osteoblast differentiation-related cell cycle can be promoted by interference with the activity of GSK3␤ on these substrates. This can be achieved in two ways (Fig. 7). First, Wnt signaling results in disassembly of the ␤-catenin destruction complex and rescue of ␤-catenin from proteasomal degradation. Second, growth factors that activate the PI3K/Akt pathway inhibit GSK3␤ by phosphorylating its Ser 9 residue (Fig. 7). Our previous work suggested that GCs inhibit the osteoblast differentiation-related cell cycle by stimulating the inhibitory activity of GSK3␤ on substrates outside the Wnt signaling pathway, e.g. phosphorylation of c-Myc (13). However, under the same conditions, we also observed that GCs inhibited the activity of LEF/TCF transcription factors, which transmit Wnt signals to target genes (13). The tight association observed between GC-induced GSK3␤ activity and the inhibition of its Ser 9 phosphorylation (13) prompted us to investigate the potential for cross-talk between the PI3K/Akt and Wnt signaling pathways through GSK3␤ Ser 9 phosphorylation.
The results of the present study indeed suggest that the PI3K/Akt pathway plays a role in the GC inhibition of the Wnt signaling pathway. GCs decreased ␤-catenin levels, and this inhibition was negated not only by inhibition of GSK3␤ (lithium), but also by inhibition of the PI3K/Akt pathway (wortmannin) (Fig. 3). Furthermore, we demonstrated that DEX inhibits Ser 473 -Akt phosphorylation in MC3T3-E1 as well as NeMCO and human osteoblast cultures. When examined at the level of LEF/TCF transcriptional activity, again we observed that the DEX-mediated inhibition of the Wnt reporter TBE 4 -luc was abrogated not only by lithium, but also by dominant negative forms of both PI3K and PKB/Akt (Figs. 1G and 2C). The notion that the PI3K/Akt pathway influences the activity of GSK3␤ not only on substrates such as c-Myc, but also on ␤-catenin, is consistent with the recent discovery that, like c-Myc, ␤-catenin undergoes priming phosphorylation before it is phosphorylated by GSK3␤ (25). It remains to be seen what determines the ability of the PI3K/Akt pathway to influence the Wnt signaling pathway. In MC3T3-E1 cells, the PI3K/Akt/ GSK3␤/␤-catenin/LEF axis (Fig. 7) appears to be specific for the commitment stage. It may also be operative in other cell systems, where tight association between activation of the PI3K/Akt pathway and ␤-catenin-mediated activation of LEF/ TCF has been demonstrated (47)(48)(49)(50). It is noteworthy that the response of ␤-catenin to GCs in our studies was transient, and the decrease seen up to ϳ12 h after GC administration (Fig. 3) was no longer observed at the 24 or 48 h time point (13). Similar to the MC3T3-E1 model, DEX also decreased ␤-catenin levels in primary osteoblast cultures (Fig. 3C).
The suggestion that GCs inhibit the osteoblast differentiation-related cell cycle by interference with the Wnt signaling pathway infers that manipulations of the Wnt signaling pathway would influence the ability of GCs to inhibit this cell cycle. We have shown previously that GC action on the differentiation-related cell cycle in the MC3T3-E1 model is attenuated in the presence of lithium, a GSK3␤ inhibitor, and this observation is now extended to the NeMCO culture system (Fig. 3D). These results can be explained based on the ability of lithium to counteract the GC-mediated decrease in c-Myc (13) as well as in ␤-catenin levels (Fig. 3A) and LEF/TCF transcriptional activity (Fig. 1G). Treatment with wortmannin, which mimicked the action of GCs on the PI3K/Akt/GSK3␤/␤-catenin/LEF axis (Figs. 2, 3, and 7), partially attenuated the ability of GCs to inhibit the differentiation-related cell cycle (Fig. 6). Finally, treatment with both wortmannin and TSA (Fig. 6C) completely blunted the GC effect on the differentiation-related cell cycle. This result can be explained based on the ability of TSA to counteract the GC-induced HDAC, which apparently contributed to inhibition of LEF/TCF transcriptional activity directly (Fig. 5). Taken together, our results suggest that GCs inhibit the osteoblast differentiation-related cell cycle in a Wnt-dependent and -independent manner and that the Wnt-dependent manner involves actions both upstream and downstream of GSK3␤. Possible interactions between the various activities of GCs described here as well as the stimulation of Dickkopf-1, a Wnt antagonist (51), remain to be explored. It is not unlikely, however, that inhibition of the PI3K/Akt pathway is a primary effect of GCs because it is demonstrable within 7.5 min of treatment (Fig. 2B).
The role of the Wnt signaling pathway in the regulation of bone mass in vivo has been well demonstrated by the altered bone mass phenotypes in mice and humans carrying mutations in the Wnt co-receptor LRP5 (26,(52)(53)(54)(55). The consequences of LRP5 inactivation could be very similar to the effect of GCs emerging from our studies: inhibition of a cell cycle that becomes critically dependent on Wnt signaling during a distinct stage of osteblast differentiation. Understanding how this cell cycle is controlled may provide unique opportunities in drug design for the treatment of medical conditions in which Wntdependent regulation of the cell cycle goes awry. These conditions may include GC-induced and other osteoporoses as well as cancer.