Glucocorticoid hormones have complex stimulatory and inhibitory effects on skeletal metabolism. Endogenous glucocorticoid signaling is required for normal bone formation in vivo, and synthetic glucocorticoids, such as dexamethasone, promote osteoblastic differentiation in several in vitro model systems. The mechanism by which these hormones induce osteogenesis remains poorly understood. We demonstrate here that the coordinate action of dexamethasone and the osteogenic transcription factor Runx2/Cbfa1 synergistically induces osteocalcin and bone sialoprotein gene expression, alkaline phosphatase activity, and biological mineral deposition in primary dermal fibroblasts. Dexamethasone decreased Runx2 phosphoserine levels, particularly on Ser125, in parallel with the upregulation of mitogen-activated protein kinase (MAPK) phosphatase-1 (MKP-1) through a glucocorticoid-receptor-mediated mechanism. Inhibition of MKP-1 abrogated the dexamethasone-induced decrease in Runx2 serine phosphorylation, suggesting that glucocorticoids modulate Runx2 phosphorylation via MKP-1. Mutation of Ser125 to glutamic acid, mimicking constitutive phosphorylation, inhibited Runx2-mediated osteoblastic differentiation, which was not rescued by dexamethasone treatment. Conversely, mutation of Ser125 to glycine, mimicking constitutive dephosphorylation, markedly increased osteoblastic differentiation, which was enhanced by, but did not require, additional dexamethasone supplementation. Collectively, these results demonstrate that dexamethasone induces osteogenesis, at least in part, by modulating the phosphorylation state of a negative-regulatory serine residue (Ser125) on Runx2. This work identifies a novel mechanism for glucocorticoid-induced osteogenic differentiation and provides insights into the role of Runx2 phosphorylation during skeletal development.
Glucocorticoids (GCs) are steroid hormones secreted by the adrenal cortex that play a pivotal role in the regulation of a variety of developmental, metabolic and immune functions. The classic mechanism of GC action is primarily at the level of transcription, where the hormone forms a transcriptionally active complex with its cognate intracellular receptor. This complex can either enhance or attenuate gene expression by binding to GC response elements in the promoter region of target genes or by non-covalently associating with additional co-regulatory proteins (Brann et al., 1995).
Synthetic GC derivatives, such as dexamethasone (DEX), have complex stimulatory and inhibitory effects on skeletal metabolism and bone formation (Ishida and Heersche, 1998). DEX is widely utilized at pharmacological doses for the treatment of inflammatory and autoimmune diseases. However, long-term administration of this hormone has adverse side effects on the skeleton, inducing osteoporosis by impairing osteoblast activity (Canalis and Delany, 2002). The cellular and molecular mechanism(s) by which pharmacological doses of DEX induce bone loss include: (1) attenuated osteoblast proliferation (Shalhoub et al., 1995; Walsh et al., 2001); (2) impaired collagen synthesis (Delany et al., 1995); (3) increased osteoblast apoptosis (Weinstein et al., 1998); (4) inhibition of osteogenic growth factors (Canalis and Delany, 2002; Chevalley et al., 1996; Luppen et al., 2003); and (5) downregulation of osteogenic gene expression (Chang et al., 1998). In contrast to these catabolic effects, physiologic levels (10 nM) of DEX promote osteoblastic differentiation in vitro (Bellows et al., 1987; Cheng et al., 1994). Furthermore, it has been recently demonstrated that GC signaling is required for normal bone volume and architecture in transgenic models, suggesting that endogenously expressed GCs may have an anabolic effect on skeletal metabolism and bone formation in vivo (Sher et al., 2004). The mechanism(s) by which DEX promotes osteogenesis remains poorly understood, largely due to conflicting results associated with the various species and differentiation states of the model systems used to study this hormone in vitro (Brann et al., 1995; Chen et al., 1983; Pei et al., 2003; Prince et al., 2001; Rickard et al., 1994; Shui et al., 2003; Viereck et al., 2002).
Runx2 (also known as Cbfa1, Osf2, AML3 and PEBP2αA) is an essential transcriptional regulator of osteoblast differentiation and bone formation. Homozygous deletion of Runx2 arrests osteoblast maturation, resulting in the absence of endochondral and intramembranous ossification (Komori et al., 1997). Moreover, Runx2 haploinsufficiency causes the pathogenic skeletal phenotype cleidocranial dysplasia in mice and humans, characterized by short stature, hypoplastic clavicles and dental abnormalities (Mundlos et al., 1997; Otto et al., 1997). Runx2 directs osteogenic differentiation by binding to an osteoblast-specific cis-acting element, termed OSE2, in the promoter region of skeletal target genes and regulating their expression (Ducy et al., 1997). We and others have demonstrated that forced expression of Runx2 upregulates osteoblast-specific gene expression and induces mineralization in a cell-type-dependent manner (Byers et al., 2002; Ducy et al., 1997; Hirata et al., 2003; Yang et al., 2003). Intriguingly, both postnatal disruption of Runx2 by dominant-negative expression and overexpression of Runx2 from the pro-α (I) collagen promoter induce bone fragility and osteopenia in transgenic mice (Ducy et al., 1999; Liu et al., 2001). These studies collectively demonstrate that cellular regulation of Runx2 is crucial for normal skeletal development and bone formation.
Runx2 is regulated at multiple levels by a complex spatiotemporal cascade of growth factors, hormones, transcription factors and cell-matrix interactions (Banerjee et al., 2001; Franceschi, 2003; Lian and Stein, 2003; Sudhakar et al., 2001). In particular, DEX has differential effects on Runx2 mRNA transcript expression, protein levels, and DNA-binding activity depending on the species, osteogenic cell type and culture conditions used to study this hormone in vitro (Chang et al., 1998; Prince et al., 2001; Viereck et al., 2002). These conflicting results suggest that DEX may regulate Runx2 by modulating its post-translational modification. The mitogen-activated protein kinase (MAPK) pathway has been shown to phosphorylate Runx2 on residues within the C-terminal proline-serine-threonine-rich (PST) domain (Franceschi and Xiao, 2003; Xiao et al., 2000). This increase in phosphorylation strongly correlates with enhanced Runx2 transactivation and is stimulated by signaling through the extracellular matrix (ECM), fibroblast growth factor-2 (FGF-2) and mechanical loading (Wang et al., 2002; Xiao et al., 2002; Xiao et al., 1998). By contrast, Wee and colleagues have reported that the activity of the human Runx2 type I isoform is negatively regulated by phosphorylation of two serine residues, Ser104 and Ser451 (corresponding to Ser125 and Ser472 in the murine type II Runx2 isoform) (Wee et al., 2002). The putative signaling cascades or effector molecules that regulate these inhibitory phosphorylation events, as well as the functional significance of these residues, remain poorly understood.
In the present study, we investigated the effect of DEX on Runx2 serine phosphorylation and the functional role of this phosphorylation state during osteoblastic differentiation. Runx2-transduced primary dermal fibroblasts were utilized as the experimental model in order to investigate the Runx2-dependent molecular pathway(s) involved in DEX-mediated osteogenesis. This reconstituted model system allowed for the direct examination of DEX and Runx2 interactions in the absence of native osteoblastic components, such as endogenous Runx2 isoforms or Runx2-independent signaling pathways, which may confound the analysis. We show that DEX induces osteogenesis, at least in part, by modulating the phosphorylation state of a negative-regulatory serine residue (Ser125) on the Runx2 type II isoform. We demonstrate that the phosphorylation state of this specific serine residue plays a crucial role in both early osteoblastic differentiation and late-stage mineralization induction. Interestingly, the mutation of Ser125 to arginine, which possibly mimics the steric hindrance caused by phosphorylation of this residue, has been identified in a human patient with cleidocranial dysplasia (Quack et al., 1999). Thus, this work assists in elucidating a mechanism of GC-mediated osteogenesis and provides insights into the functional importance of Runx2 phosphorylation during skeletal pathogenesis.
Runx2 and DEX synergistically induce osteoblastic differentiation
In order to assess the effects of DEX on Runx2-mediated osteogenesis in the absence of native osteoblastic signals, primary dermal fibroblasts were genetically engineered to constitutively express the Runx2 type II isoform and cultured in differentiation media supplemented with and without 10 nM DEX. We hypothesized that a combination of forced Runx2 expression and DEX supplementation would induce osteoblastic differentiation in primary dermal fibroblasts. Skeletal gene expression was investigated at 1, 3 and 7 days post-transduction by quantitative RT-PCR (Fig. 1). Runx2 mRNA levels were upregulated by two orders of magnitude in transduced cultures compared with control cells at day 3, and this relative difference decreased to one order of magnitude as cells reached confluence at day 7. eGFP transgene expression was detectable by flow cytometry for 21 days, demonstrating integrated and sustained expression of the Runx2 transgene. Runx2 primers utilized in this study (Byers and Garcia, 2004) were designed to detect both type I and type II Runx2 isoforms (Banerjee et al., 2001; Harada et al., 1999). Thus, the absence of Runx2 mRNA transcripts in untransduced cultures indicates that the Runx2 type II isoform is the predominant isoform expressed in transduced cultures. Osteocalcin (OCN) is the most abundant non-collagenous ECM protein in bone and a marker of mature osteoblasts (Ducy et al., 1996). Bone sialoprotein (BSP) and osteopontin (OPN) are ECM glycoproteins implicated in the regulation of mineralized nodule nucleation (Seibel, 2000). BSP mRNA levels were upregulated by three orders of magnitude in Runx2-transduced cells, whereas OCN and OPN demonstrated a ten-fold increase at 7 days post-transduction compared with control cultures. Notably, addition of DEX to Runx2-expressing cultures resulted in significant enhancements in OCN, BSP and OPN mRNA levels at days 3 and 7 compared with untreated Runx2-expressing cultures. DEX supplementation alone significantly increased OPN mRNA compared with untreated controls, but had no effect on Runx2, OCN and BSP mRNA levels. These results demonstrate that sustained expression of Runx2 upregulates osteoblastic gene expression and treatment with DEX enhances this effect.
Alkaline phosphatase (ALP) is a membrane-bound enzyme that hydrolyzes phosphate esters, thereby making inorganic phosphate available for incorporation into mineral deposits (Seibel, 2000). The activity of this enzyme was examined at 7 days post-transduction (Fig. 2A). Runx2 overexpression stimulated a 15-fold increase in ALP activity compared with control cultures. Moreover, addition of DEX to Runx2-transduced cultures resulted in a synergistic increase in ALP activity. Matrix mineralization was assessed at 14 and 21 days post-transduction by von Kossa staining and image analysis (Fig. 2B,C). Notably, co-treatment with Runx2 and DEX synergistically induced matrix mineralization in primary dermal fibroblasts, whereas Runx2-expressing cultures alone displayed minimal staining and sparse nodule formation. No mineral was detected in empty vector or DEX-treated control cultures. Fourier transform infrared (FTIR) spectroscopy was utilized to analyze the chemical composition of the mineral phase because osteogenic culture conditions can lead to non-biological calcium phosphate precipitation (Bonewald et al., 2003; Boskey et al., 1996) (Fig. 2D). Runx2 and cranial bone samples displayed amide I and II peaks indicative of ECM proteins, an enhanced phosphate peak at 1100 cm–1, a doublet split phosphate peak at 560 and 605 cm–1, and a carbonate peak at 870 cm–1, which represent the characteristic bands of carbonate-containing, poorly crystalline hydroxyapatite (Sauer and Wuthier, 1988). These bands were absent in empty vector and DEX-only control cultures. Collectively, these results demonstrate that a combination of constitutive Runx2 overexpression and DEX supplementation synergistically induces osteoblastic differentiation in primary dermal fibroblasts.
DEX decreases Runx2 serine phosphorylation
Because DEX treatment did not alter Runx2 gene expression, we examined the effects of this hormone on Runx2 total protein and phosphoserine levels by immunoprecipitation and western blot analysis. Significant amounts of Runx2 protein were detected in Runx2-transduced samples compared with unmodified control cells (Fig. 3A,B). Surprisingly, addition of DEX to Runx2-transduced cells significantly decreased Runx2 serine phosphorylation, whereas total Runx2 protein levels remained unchanged (Fig. 3C,D). Omission of immunoprecipitation antibody or cell lysates in negative controls demonstrated the stringency and specificity of the immunoprecipitation procedure. These findings indicate that DEX regulates Runx2 phosphorylation, with no net effect on Runx2 mRNA or protein levels. This DEX-induced decrease in Runx2 serine phosphorylation correlates with, and might be functionally linked to, the observed DEX-mediated synergistic induction of osteoblastic differentiation.
Mutation of Ser125 decreases Runx2 serine phosphorylation
Several serine phosphorylation sites have been identified on the Runx2/Cbfa1 type I isoform; however, Ser104 (corresponding to Ser125 on Runx2 type II) was the only residue that exhibited changes in phosphorylation during bone morphogenetic protein-2 (BMP-2)-induced osteogenesis in C2C12 cells (Wee et al., 2002). Furthermore, phosphorylation of this serine negatively regulated the transcriptional activity of Runx2. In order to assess whether Ser125 is involved in the observed net decrease in Runx2 serine phosphorylation following DEX treatment, site-directed mutagenesis was performed on this residue within the full-length Runx2 construct (Fig. 4A). Mutation of Ser125 to glutamic acid (125Glu) mimics constitutive phosphorylation of this residue by placing a bulky, negatively charged group at the site in the same manner as the presence of phosphoserine (Whalen et al., 1997). Mutation of Ser125 to glycine (125Gly) leads to constitutive dephosphorylation by preventing post-translational modification of this residue. Primary dermal fibroblasts were transduced with wild-type Runx2 (Runx2-WT), Runx2-125Gly, or Runx2-125Glu retrovirus and cultured in osteogenic media with or without DEX. Retroviral transduction efficiency was approximately 65% for all retroviral stocks (Fig. 4B). Runx2 mRNA expression was upregulated by two orders of magnitude at 3 days post-transduction with Runx2-WT, Runx2-125Gly and Runx2-125Glu retroviral vectors compared with control cells (Fig. 4C). Moreover, equivalent levels of Runx2-WT, Runx2-125Gly and Runx2-125Glu protein were detected at 7 days (Fig. 4D). Overall, there were no differences in Runx2 mRNA or protein levels between transduced samples, excluding the possibility that experimental results obtained with these retroviral vectors were skewed by differences in Runx2 protein levels and/or transduction efficiencies.
The effect of Ser125 on Runx2 phosphoserine levels was investigated at 7 days post-transduction by immunoprecipitation and western blotting (Fig. 5). Mutation of Ser125 to glycine significantly reduced total Runx2 serine phosphorylation in the absence of DEX, suggesting that this serine residue is a major phosphorylation site on exogenously expressed Runx2 in primary dermal fibroblasts. Furthermore, no apparent net change in Runx2-125Gly serine phosphorylation was observed upon addition of DEX, whereas the hormone significantly reduced phosphoserine levels in Runx2-WT. Mutation of Ser125 to glutamic acid had similar effects on Runx2 phosphoserine levels as those observed upon mutation of this residue to glycine (J.E.P. and A.J.G., unpublished data). These results indicate that Runx2 is phosphorylated at a basal level in untreated cultures, particularly on Ser125, and suggest that DEX modulates the phosphorylation state of Runx2-Ser125 in parallel with the stimulation of osteogenesis.
Runx2-Ser125 phosphorylation regulates DEX-induced osteoblastic differentiation
Phosphorylation of Ser125 has been reported to negatively regulate Runx2 transactivation in NIH-3T3 cells transfected with an OCN promoter-driven reporter gene (Wee et al., 2002). However, the direct effects of wild-type Runx2 and its mutated derivatives on osteoblastic differentiation have not been examined. We hypothesized that the phosphorylation state of Runx2-Ser125 plays a crucial role in DEX-induced osteogenesis. Primary dermal fibroblasts were transduced with Runx2-WT, Runx2-125Gly or Runx2-125Glu retrovirus and cultured in osteogenic media supplemented with or without 10 nM DEX. Mutation of Ser125 to glycine, mimicking constitutive dephosphorylation, significantly upregulated OCN, BSP and OPN mRNA transcript expression compared with Runx2-WT in the absence of DEX (Fig. 6). DEX treatment of cultures expressing Runx2-125Gly had no significant effect on expression of the gene encoding OCN, but enhanced expression of the genes encoding BSP and OPN. By contrast, mutation of Serine125 to glutamic acid, mimicking constitutive phosphorylation, inhibited Runx2 transactivation of all three osteoblastic genes, whereas DEX partially recovered the effect of Runx2-125Glu on OCN and BSP expression only. These results corroborate the observations of Wee et al. that Ser125 phosphorylation inhibits Runx2 transactivation of an OCN-driven reporter gene (Wee et al., 2002).
Consistent with alterations in osteoblastic gene expression, mutation of Ser125 to glycine stimulated ALP activity to levels significantly higher than Runx2-WT cultures and comparable with Runx2-WT cultures treated with DEX (Fig. 7A). DEX had no additional effect on ALP activity in cultures expressing Runx2-125Gly. Mutation of Ser125 to glutamic acid diminished ALP activity to levels comparable with unmodified control cultures and DEX treatment partially recovered this activity. Furthermore, mutation of Ser125 to glycine also significantly enhanced matrix mineralization compared with cultures expressing Runx2-WT in the absence of DEX (Fig. 7B,C). DEX treatment synergistically enhanced mineralization in cells expressing Runx2-WT and cells expressing Runx2-125Gly. Conversely, mutation of Ser125 to glutamic acid completely blocked mineralization of these cultures in the absence and presence of DEX. The mineral phase in Runx2-WT and Runx2-125Gly cultures displayed FTIR spectrograms similar to those shown in Fig. 2D with the characteristic bands of a carbonate-containing, poorly crystalline hydroxyapatite (J.E.P. and A.J.G., unpublished data).
Finally, these experiments were repeated in primary bone marrow stromal cells (BMSCs) in order to ensure that results were not an artifact of the non-osteogenic cell source used in this study (Fig. 8A,B). DEX treatment alone stimulated osteoblastic differentiation in BMSCs, including ALP activity and matrix mineralization, compared with untransduced controls. Moreover, co-treatment with Runx2-WT and DEX enhanced osteogenic differentiation in BMSCs compared with Runx2-WT overexpression or DEX treatment alone. Mutation of Ser125 to glycine stimulated ALP activity and mineralized nodule formation to levels significantly higher than Runx2-WT cultures and equivalent to Runx2-WT cultures treated with DEX. DEX treatment showed no additional effect on osteogenesis in cultures expressing Runx2-125Gly. Mutation of Ser125 to glutamic acid antagonized ALP activity and mineralization to similar levels as untransduced BMSC controls with or without DEX. Overall, these results demonstrate equivalent functional effects of Runx2-Ser125 during DEX-induced osteogenesis in primary BMSCs and Runx2-engineered primary dermal fibroblasts. We speculate that the low levels of ALP activity and mineralization observed in untransduced BMSCs and BMSCs expressing Runx2-125Glu may be a result of interactions between DEX and endogenously expressed Runx2 or additional Runx2-independent pathways.
DEX upregulates MKP-1 through a GC-receptor-mediated transcriptional mechanism
As a first step towards elucidating the DEX-mediated mechanism(s) involved in the regulation of Runx2 phosphorylation, we examined the ability of DEX to activate components of the MAPK signaling pathway. In particular, MAPK phosphatase-1 (MKP-1) is a dual-specificity phosphatase that dephosphorylates and inactivates MAPKs such as extracellular signal-regulated kinase (ERK1/2), c-Jun N-terminal kinase (JNK) and p38 protein kinase (Imasato et al., 2002; Lasa et al., 2002; Liu et al., 1995; Sun et al., 1993). Pharmacological doses of DEX (≥100 nM) have been shown to upregulate MKP-1 in a variety of cell types (Engelbrecht et al., 2003; Kassel et al., 2001; Lasa et al., 2002; Wu et al., 2005). We postulated that MKP-1 may be stimulated by physiological concentrations of DEX during osteoblastic differentiation in our experimental model. Primary fibroblasts transduced with Runx2 retrovirus or left unmodified as controls were cultured in osteogenic media supplemented with or without DEX. MKP-1 mRNA and protein levels were evaluated at 1, 3 and 7 days post-transduction by quantitative RT-PCR and western blot analysis, respectively. DEX treatment stimulated expression of the gene encoding MKP-1 (Fig. 9A) and its protein (Fig. 9B) in unmodified and Runx2-expressing cultures relative to untreated controls. Notably, MKP-1 protein levels were upregulated by DEX after 3 days and remained elevated through 7 days in culture, which correlates with the observed decrease in Runx2 serine phosphorylation after treatment with DEX for 7 days. Overexpression of Runx2 significantly inhibited MKP-1 mRNA and protein levels at 3 and 7 days post-transduction compared to unmodified controls and addition of DEX to Runx2-transduced cultures restored MKP-1 to basal expression levels. Moreover, the induction of MKP-1 mRNA by DEX was abrogated by treatment with the partial GC receptor agonist/antagonist RU486 (100 nM) for 72 hours (Fig. 9C). 18S gene expression remained unchanged for all treatment groups in this experiment. Finally, no differences in p38 MAPK, protein phosphatase 5 and protein tyrosine phosphatase type D expression were observed among experimental groups, suggesting that the observed shifts in expression were specific for MKP-1 (J.E.P. and A.J.G., unpublished data). Taken together, these data demonstrate that DEX induces MKP-1 in Runx2-expressing fibroblasts through a GC-receptor-mediated mechanism.
Inhibition of MKP-1 attenuates the DEX-mediated decrease in Runx2 serine phosphorylation
Sanguinarine has recently been identified as a potent and selective inhibitor of MKP-1 activity, exhibiting at least a threefold selectivity for MKP-1 over dual-specificity phosphatases such as MKP-3, VH-1-related phosphatase, Cdc25B and protein-tyrosine phosphatase 1B (Vogt et al., 2005). We utilized this inhibitor to assess the role of MKP-1 during the DEX-mediated modulation of Runx2 serine phosphorylation. Primary dermal fibroblasts were transduced with Runx2 retrovirus or left unmodified for controls and cultured in osteogenic media with and without 10 nM DEX. After 7 days in culture, cells were treated with vehicle (ethanol), vehicle plus DEX (10 nM), sanguinarine (50 μM), or sanguinarine (50 μM) plus DEX (10 nM) for 30 minutes. Protein expression for MKP-1, MKP-3, phospho-ERK, ERK and GAPDH was assessed by western blotting (Fig. 10A). MKP-1 protein levels were markedly decreased by sanguinarine in unmodified and Runx2-expressing fibroblasts. Consequently, the decrease in ERK phosphorylation caused by DEX induction of MKP-1 was reversed upon treatment with sanguinarine. ERK and GAPDH total protein levels remained unchanged for all experimental conditions. Notably, MKP-3 protein levels were not significantly altered by sanguinarine, suggesting that the inhibitor was selective for MKP-1 over this closely related dual-specificity phosphatase. Runx2 phosphoserine levels were then examined by immunoprecipitation and western blot analysis (Fig. 10B). Sanguinarine treatment blocked the DEX-mediated decrease in Runx2 serine phosphorylation, suggesting that DEX modulates the inhibitory phosphorylation of Runx2 through MKP-1.
We demonstrate that a combination of DEX supplementation and constitutive Runx2 overexpression synergistically induced osteoblastic differentiation in primary dermal fibroblasts, as characterized by enhanced OCN and BSP gene expression, ALP activity, and biological mineral deposition. DEX treatment decreased Runx2 phosphoserine levels, particularly on Ser125. Mutation of Ser125 to glutamic acid, mimicking constitutive phosphorylation, inhibited Runx2-induced osteogenic differentiation, which was not rescued by DEX treatment. Conversely, mutation of Ser125 to glycine, mimicking constitutive dephosphorylation, markedly increased osteogenic differentiation, which was enhanced by but did not require additional DEX supplementation. The DEX-induced decrease in Runx2 phosphorylation correlated with upregulation of MKP-1 through a GC-receptor-dependent mechanism. Furthermore, inhibition of MKP-1 abrogated the effect of DEX on Runx2 phosphoserine levels. To our knowledge, this is the first time that DEX-activated MKP-1 expression has been implicated in the regulation of Runx2 phosphorylation. Collectively, these results suggest that DEX induces osteogenesis by modulating the phosphorylation state of a negative-regulatory serine residue (Ser125) on Runx2/Cbfa1 via MKP-1. This work offers important insights into the role of Runx2 during hormone-regulated skeletal development and maintenance.
Primary dermal fibroblasts engineered to express elevated and sustained levels of Runx2 were utilized for the investigation of the effects of DEX on osteoblastic differentiation. A major advantage of this model system is that it allows for direct analysis of the Runx2 type II isoform and its mutants during GC-induced osteogenesis in the absence of endogenous Runx2 and DEX-responsive, osteoblast-specific pathways. Notably, DEX stimulation of Runx2-expressing fibroblasts induced several important components of the osteoblastic differentiation program, including OCN and BSP gene expression, ALP activity, and matrix mineralization, whereas DEX treatment alone did not significantly influence any of these osteoblastic markers. This enhancement in osteogenesis is consistent with the effects of DEX on several osteoblastic systems, including rat calvarial cells, rat and human BMSCs, and chick periosteal cells (Bellows et al., 1987; Byers and Garcia, 2004; Chen et al., 1983; Cheng et al., 1994; Rickard et al., 1994; Tenenbaum and Heersche, 1985). Although these model systems have been instrumental in the elucidation of numerous DEX-responsive signaling cascades, they are limited because they prevent the isolation of Runx2-dependent from Runx2-independent pathways. Finally, in order to ensure that these results were not an artifact of the non-osteogenic cell source, we analyzed the effects of Runx2-WT and Runx2-Ser125 mutants during DEX-induced osteoblastic differentiation in primary BMSCs. This osteoblastic model exhibited equivalent functional responses during DEX-induced osteoblastic differentiation compared with Runx2-engineered fibroblasts. On the basis of these results, we hypothesized that the anabolic effects of GCs in vitro occur through a Runx2-dependent mechanism involving the post-translational modification of Runx2.
We demonstrate that DEX decreases Runx2 phosphoserine levels, particularly on Serine125, in parallel with osteoblastic differentiation. By contrast, Shui et al. reported that phosphorylation of Runx2 on tyrosine, threonine and serine residues increases during DEX-induced osteoblastic differentiation in human BMSCs (Shui et al., 2003). However, this report did not include a `No DEX' condition to isolate the effects of DEX from alternative pathways activated during the onset of osteogenesis. Thus, beyond this correlative evidence, a direct link between DEX and Runx2 phosphorylation has not been established prior to this work. However, it is important to note that our results do not rule out the possibility that additional Runx2 phosphorylation sites are altered during DEX-induced osteogenesis. Previous analyses have also shown that collagen, FGF-2 and mechanical loading enhance Runx2 transcriptional activity through the MAPK pathway (Wang et al., 2002; Xiao et al., 2002; Xiao et al., 1998). Moreover, protein kinase A (PKA) has been shown to phosphorylate Runx2, and parathyroid hormone enhances Runx2 transactivation of the collagenase-3 promoter through a PKA-dependent pathway (Selvamurugan et al., 1998; Selvamurugan et al., 2000). These pathways stimulate Runx2 phosphorylation on putative residues within the C-terminal PST domain (Selvamurugan et al., 2000; Xiao et al., 2000), but the specific residues targeted have not been identified. Interestingly, whereas these stimulatory phosphorylation sites were found in the PST domain, Ser125 is located within the N-terminal runt domain, suggesting that phosphorylation at different regions within the Runx2 protein may play different functional roles in osteoblastic differentiation.
The mutagenesis analysis conducted in this study demonstrates that the phosphorylation state of Runx2-Ser125 plays a crucial role during DEX-induced osteoblastic differentiation. These results offer important insights into skeletal pathogenesis, as mutation of this residue to arginine has been documented in one patient with cleidocranial dysplasia (Quack et al., 1999). Ito and colleagues recently identified Ser14, Ser104, Ser451, Ser485 and Ser489 as potential phosphorylation sites on the human Runx2/Cbfa1 type I isoform (Wee et al., 2002). Of these residues, Ser104 and Ser451 were implicated in the negative regulation of Runx2 transcriptional activity. Mutation of Ser104, corresponding to Ser125 on the murine Runx2 type II isoform, to both glycine and glutamic acid inhibited Runx2 transactivation of an OCN-promoter-driven reporter gene (Wee et al., 2002). These results are consistent with our observations for Runx2-mediated differentiation, but contradict reports that mutation of Runx1/AML1c Ser94, analogous to Runx2-Ser125, had no effect on transcriptional activity (Zhang et al., 2004). Wee and colleagues also reported that the phosphorylation state of Ser451, corresponding to Ser472 on the murine Runx2 type II isoform, has a crucial role in the transcriptional activity of Runx2 (Wee et al., 2002). However, the phosphorylation state of Ser472 had no apparent effect on Runx2-induced osteogenic gene expression, ALP activity or mineralization in the present study (J.E.P. and A.J.G., unpublished data). Similarly, mutation of Runx1/AML1c Ser424, corresponding to Runx2 type II Ser472, did not alter transcriptional activity of this runt-domain protein family member (Zhang et al., 2004). Overall, it is evident that disparities exist in the phosphorylation pattern of Runx protein family members, suggesting that the phosphorylation state of Ser125 and Ser472 might be isoform specific, cell-type-specific, or regulated by independent signaling pathways.
The present analysis supports a mechanism by which DEX induces osteoblastic differentiation through modulation of the phosphorylation state of a negative-regulatory serine on Runx2. The ability of DEX to partially recover osteogenic gene expression and ALP activity in cultures expressing Runx2-125Glu suggests that the hormone might also have auxiliary modes of regulating Runx2 beyond the mechanism detailed in this study. Indeed, GCs might also mediate osteogenesis though a Runx2-dependent mechanism involving the physical association of the transcription factor with co-regulatory proteins. Recent evidence suggests that Runx2 serves as a molecular scaffold that facilitates the assembly of co-regulatory proteins and accessory transcription factors into a macromolecular transcriptional regulatory complex (Franceschi, 2003). Runx2 contains specific functional regions that physically interact with several accessory factors (Ducy, 2000). In particular, the runt domain is a conserved region of 128 amino acids that is essential for DNA binding and heterodimerization with transcription factors such as Cbfβ/PEBP2β (Ito, 1999; Thirunavukkarasu et al., 1998), LEF-1 (Kahler and Westendorf, 2003) and c-Fos/c-Jun (D'Alonzo et al., 2002; Hess et al., 2001; Selvamurugan et al., 1998). The C-terminal PST domain contains a nuclear localization signal, a transcriptional activation region and a repressor region, and has been shown to colocalize with SMADs (Hanai et al., 1999; Zhang et al., 2000), CCAAT/enhancer-binding proteins (C/EBPβ and C/EBPδ) (Gutierrez et al., 2002), HES-1 and Groucho/TLE proteins (McLarren et al., 2000). Osteogenic agents, such as PTH and BMP, regulate the association of Runx2 with several of these factors, but the role of DEX in these protein-protein interactions is poorly understood. Interestingly, Wee et al. found that the mutation of Ser104 to glutamic acid, which decreased Runx2 transcriptional activity, also appeared to destabilize the protein and inhibit the heterodimerization of Runx2 with CBF-β (Wee et al., 2002) Thus, it is possible that the DEX-mediated regulation of Runx2 phosphorylation alters the interaction of this transcription factor with accessory proteins, which might have downstream effects on Runx2 transcriptional activity. Finally, these results do not rule out the possibility that GCs might mediate osteogenesis by Runx2-independent signaling pathway(s), which may cooperatively act with Runx2-stimulated gene products to synergistically induce matrix mineralization.
In summary, we have demonstrated that DEX induces osteogenesis, at least in part, by modulating the phosphorylation state of a negative-regulatory serine residue (Ser125) on Runx2 through an MKP-1-dependent mechanism. Although this particular mechanism is probably not the sole signaling pathway activated by DEX during osteogenic differentiation, it provides significant insights towards the role of Runx2 phosphorylation during GC-regulated skeletal development.
Materials and Methods
Cell culture and reagents
Primary fibroblasts were harvested from 8–16-week-old male Wistar rats by enzymatic digestion of the dermal skin layer (Weinberg et al., 1993). Primary bone marrow stromal cells (BMSCs) were harvested from the femora of 8–6-week-old Wistar rats as described previously (Byers and Garcia, 2004). Cells were expanded in growth media consisting of DMEM (fibroblasts) or α-MEM (BMSCs), 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin. Antibiotics and cell culture media were obtained from Invitrogen, FBS was purchased from Hyclone, and all other cell culture supplements and reagents were acquired from Sigma.
The Runx2 retroviral vector utilizes the promoter activity of a 5′ long terminal repeat to express a single, bicistronic mRNA encoding the murine cDNA for the type II MASNSLF Runx2 isoform, followed by an internal ribosomal entry site and a Zeocin-resistance enhanced green fluorescent fusion protein (eGFP) (Byers et al., 2002). Plasmid DNA was purified from transformed Escherichia coli using Megaprep kits from Qiagen. Retroviruses were packaged by transient transfection of helper-virus free ΦNX amphotropic producer cells with plasmid DNA as described elsewhere (Byers et al., 2002).
Passage-four primary fibroblasts and passage-two BMSCs were plated on 6-well tissue culture polystyrene plates coated with 0.1% type I collagen (Vitrogen). Cells at 50-70% confluence were transduced with retroviral stocks and maintained in differentiation media consisting of DMEM (fibroblasts) or α-MEM (BMSCs), 10% FBS, 100 U/ml penicillin G sodium, 100 μg/ml streptomycin sulfate, 50 μg/ml L-ascorbic acid, 2.1 mM sodium β-glycerophosphate, and with or without 10 nM DEX. Culture media was changed every two days until end-point assay. No differences were observed between empty vector retrovirus (negative control) and unmodified cells in all experiments. Runx2-transduced cells were analyzed for transduction efficiency by quantification of eGFP expression by flow cytometry. High levels of eGFP were detected in ≥65% of primary dermal fibroblasts and ≥45% BMSC at 72 hours post-transduction.
Single amino acid mutations were performed on the Runx2 plasmid with the QuikChange site-directed mutagenesis kit (Stratagene). The codon AGT, encoding Ser125 of the Runx2 type II isoform, was mutated to glycine and glutamic acid. The codon TCT, encoding Ser472 of the Runx2 type II isoform, was mutated to alanine and a glutamic acid. The forward primer 5′-CCGCACCGACGGTCCCAACTTCCTG-3′ (mutation underlined) and reverse primer 5′-CAGGAAGTTGGGACCGTCGGTGCGG-3′ were used to mutate Ser125 to Gly125, whereas the forward primer 5′-TGGTCCGCACCGACGAGCCCAACTTCCTGTGCT-3′ and reverse primer 5′-AGCACAGGAAGTTGGGCTCGTCGGTGCGGACCA-3′ were used to mutate Ser125 to Glu125. The forward primer 5′-GGGGGAGACCGGGCACCTTCCAGGATGGT-3′ and reverse primer 5′-ACCATCCTGGAAGGTGCCCGGTCTCCCCC-3′ were used to mutate Ser472 to Ala472, whereas the forward primer 5′-CGGGGGAGACCGGGAGCCTTCCAGGATGGTC-3′ and reverse primer 5′-GACCATCCTGGAAGGCTCCCGGTCTCCCCCG-3′ were used to mutate Ser472 to Glu472. The Runx2 gene was sequenced to verify the presence of the desired mutation (Seqwright).
Osteoblastic differentiation assays
Osteoblastic differentiation assays were performed as described previously (Byers et al., 2002; Gersbach et al., 2004). Gene expression was investigated at 1, 3 and/or 7 days post-transduction by quantitative RT-PCR using rat-specific primers (Byers and Garcia, 2004). Primers used for the analysis of MKP-1 (NM_053769) were 5′-AGTTTCACGTGCCACCGG-3′ (forward) and 5′-GTTATTGCATTGCTCCTCCCA-3′ (reverse). Alkaline phosphatase (ALP) activity was quantified at 7 days post-transduction using 4-methyl-umbelliferyl-phosphate substrate and normalized to total protein. Matrix mineralization was assessed at 14 and 21 days post-transduction by von Kossa histochemical staining for phosphate deposits. The mineralized surface area was quantified by automated image analysis of 24 representative 2′ images. FTIR spectroscopy was performed on ethanol-fixed cultures pressed into KBr pellets using a Nexus 470 FTIR spectrometer (ThermoNicolet).
Immunoprecipitation and western blot analysis
Cells were lysed in 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.5% (v/v) NP-40, 350 μg/ml PMSF, 10 μg/ml leupeptin, 10 μg/ml aprotinin, 1 mM Na3VO4 and 50 mM NaF after 7 days in culture. Whole-cell extracts (150 μg protein) were immunoprecipitated with 5 μl of anti-Runx2 antibody (Santa Cruz Biotechnology) and 20 μl protein A agarose beads. Immune complexes were resolved on 12% SDS-PAGE gels, transferred to nitrocellulose and blotted with anti-AML3 (Oncogene) and anti-phosphoserine (ab9335, Abcam; or 7F12, Biomol) antibodies, followed by sequential incubation in biotin-conjugated anti-IgG and ALP-conjugated anti-biotin antibodies. Immunoreactivity was detected using ECF substrate (Amersham Bioscience) and a Fuji Image Analyzer. Similar trends were observed for both phosphoserine antibodies. Western blot analysis of whole-cell lysates was performed with anti-AML3, anti-ERK (Santa Cruz Biotechnology), anti-phosphoERK (Cell Signaling Technology), anti-MKP-3 (C-20: Santa Cruz Biotechnology), anti-MKP-1 (M-18 or C-19: Santa Cruz Biotechnology) and anti-GAPDH (Chemicon) antibodies. Adobe Photoshop image analysis software was used to quantify the intensity of the western blot bands.
Experiments were performed at least three times in triplicate, each with unique Runx2 retroviral supernatant preparations, and two independent fibroblast isolates. No differences were observed between unmodified and empty-vector-transduced cells in all assays. Data are reported as mean ± s.e.m., and statistical comparisons using SYSTAT 8.0 were based on an analysis of variance and Tukey's test for pairwise comparisons within timepoints, with P<0.05 considered significant. In order to make the variance independent of the mean, statistical analysis of real-time PCR data was performed following logarithmic transformation of the raw data (Byers et al., 2002).
We thank Benjamin A. Byers for helpful discussions and Amanda A. Walls for technical assistance. This research was funded by the NIH (R01-EB003364), the Georgia Tech/Emory Engineering Research Center on the Engineering of Living Tissues (NSF EEC-9731643), the Emory-Georgia Tech Biomedical Technology Research Center, and a National Science Foundation Graduate Research Fellowship to J.E.P.