Stimulation of Smad1 Transcriptional Activity by Ras-Extracellular Signal-Regulated Kinase Pathway: A Possible Mechanism for Collagen-Dependent Osteoblastic Differentiation†
The authors have no conflict of interest
Abstract
Signals from bone morphogenetic protein receptors (BMPRs) and cell adhesion to type I collagen are both important for osteoblastic differentiation and functions. BMP signals are mediated mostly by Smad and collagen signals are transduced by integrins to activate focal adhesion kinase (FAK) and its downstream molecules. This study was undertaken to clarify how extracellular matrix collagen signals converge with BMP actions. We show that integrin activation by collagen was involved in BMP signals because disruption of either collagen synthesis or collagen-α2β1-integrin binding inhibited the stimulatory effect of BMP-2 on osteoblastic MC3T3-E1 cells. Downstream signals of collagen-integrin might be FAK-Ras-extracellular signal-regulated kinase (ERK) in osteoblastic cells. We further show that Ras-ERK signals enhanced the transcriptional activity of Smad1 in response to BMP in these cells transiently transfected with expression plasmids for a constitutively active mutant RasV12, a dominant negative mutant RasN17, and an ERK phosphatase CL100. Ras-ERK signals did not augment the transcriptional activity of Smad3 in response to transforming growth factor β (TGF-β) receptor activation but that of Smad1 in response to BMPR activation as examined in COS-1 cells. These observations suggest that the Ras-ERK pathway downstream of integrin-FAK is involved in Smad1 signals activated by BMP and provide a possible mechanism for cooperation between intracellular signals activated by integrin and BMPRs in osteoblastic cells.
INTRODUCTION
Osteoblastic differentiation is an essential part of bone formation because active osteoblasts should be recruited at the site of osteoclastic bone resorption to compensate for the continuous loss of bone matrix and maintain structural integrity of the skeletal system. Accumulating evidence indicates that bone morphogenetic proteins (BMPs) are important for osteoblastic differentiation.1-4 BMPs are members of the transforming growth factor β (TGF-β) superfamily and bind to and activate their receptor serine/threonine kinases. BMP receptor (BMPR) complexes are heterooligomer of type I (BMPR-I) and type II (BMPR-II) receptors and phosphorylate Smad1 and its related molecules on ligand binding.5, 6 Phosphorylated Smads together with Smad4 are translocated into the nucleus where they may regulate transcriptional activity of genes involved in osteoblastic differentiation.7
Cell-matrix interactions through integrins also generate essential signals for osteoblastic differentiation.8-10 In osteoblastic cells, type I collagen-α2β1-integrin interaction induces activation of focal adhesion kinase (FAK) and extracellular signal-regulated kinase (ERK), a member of mitogen-activated protein (MAP) kinases.8 Our previous studies indicate that BMP and integrin-FAK-ERK signals activated by collagen are both required for the osteoblastic differentiation.8, 11 BMP actions are involved specifically in the expression of osteoblastic phenotypes, whereas integrin-FAK signals generate diverse actions depending on cellular context.12 Thus, it is possible that some of integrin-FAK signals merge with BMP actions within cells in an osteoblast lineage to stimulate their differentiation. However, it is yet uncertain where both signals merge and how they cross-talk each other in osteoblastic cells.
Recent studies for signal transduction of BMP have revealed its communication with Ras-MAP kinase-activator protein 1 (AP-1) pathways at various steps. Ras-Raf (MAP kinase kinase kinase)-AP-1 is involved in BMP-4 signals for the dorsal-ventral patterning of Xenopus embryos.13 TAK-1, a member of MAP kinase kinase kinase, which is activated by BMPR-IA,14 cooperates with Smad1 or -5 in the BMP signal transduction pathway in Xenopus embryonic development.15 ERK is involved in BMP-2-induced osteoblastic differentiation of mesenchymal cell line C3H10T1/2.16 In contrast, although tyrosine kinase domains of cell-surface receptors for fibroblast growth factor (FGF) and epidermal growth factor (EGF) activate the Ras-ERK pathway, these growth factors oppose BMP signaling pathways in limb bud outgrowth17 and in mammalian cells,18 respectively. BMP-2 and BMP-4 can both counteract the induction by FGF of genes essential for tooth development.19 Then, although integrin-mediated signals resulting in ERK activation are essential for BMP-induced osteoblastic differentiation, it is uncertain how downstream signals of integrin stimulate BMP actions.
This study was undertaken to clarify how extracellular matrix collagen signals converge with BMP actions. First, we describe evidence that collagen-integrin interactions are involved in BMP actions via FAK-ERK signals in osteoblastic cells. Then, we show that Ras-ERK signals preferentially enhance the transcriptional activity of Smad1 in response to BMP, and this may be a possible mechanism for cooperation between intracellular signals activated by integrin and BMPRs in osteoblastic cells.
MATERIALS AND METHODS
Materials
Mouse monoclonal antibody against phospho-ERK1/2 was obtained from New England Biolabs, Inc. (Beverly, MA, USA); hamster monoclonal antibody against mouse α2β1-integrin (HMα2) was obtained from Sumitomo Denko Co. (Osaka, Japan); hamster nonimmune immunoglobulin G (IgG) was from Cedarlane Laboratories, Ltd. (Ontario, Canada); l-azetidine-2-carboxylic acid was from Wako Pure Chemicals (Osaka, Japan); L-proline, PD98059, and wortmannin were from Sigma (St. Louis, MO, USA). Recombinant human (rh) BMP-2 was a gift from Yamanouchi Pharmaceutical Co. (Tokyo, Japan). MC3T3-E1 cells were kindly provided by Dr. H. Kodama (Ohu Dental College, Japan). HOS-TE85 cells were obtained from RIKEN Cell Bank (Tsukuba, Japan).
Cell culture
Osteoblastic MC3T3-E1 cells established from neonatal mouse calvariae and human osteosarcoma HOS-TE85 cells were cultured in α-modified minimum essential medium supplemented with 10% fetal bovine serum, penicillin, and streptomycin including 50 mg/liter of ascorbic acid. The medium was changed twice a week. Cells were treated with indicated reagents or antibodies after being confluent in the presence of 10% fetal bovine serum. COS-1 cells were cultured in DulbeccO's modified Eagle's medium with 10% fetal bovine serum and antibiotics.
Alkaline phosphatase activity in osteoblastic cell
Osteoblastic MC3T3-E1 cells were cultured until confluence. Cells were treated with either 0.3 mM of l-azetidine-2-carboxylic acid (a proline analogue that inhibits collagen synthesis) or 10 mg/liter of anti-α2β1-integrin antibody that blocks the integrin collagen receptor for 3 days. Nonimmune IgG was included in control cultures. Cells were washed twice with ice-cold phosphate-buffered saline (PBS) and scraped in 10 mM of Tris-HCl containing 2 mM of MgCl2 and 0.05% Triton X-100, pH 8.2. The cell suspension was homogenized using Pellet Pestle (Kontes, Vineland, NJ, USA) on ice after two cycles of freeze-thawing. Aliquots of supernatants were assayed for protein concentration with a Bio-Rad Kit (Bio-Rad, Hercules, CA, USA) according to Bradford's method and for alkaline phosphatase (ALP) activity.20 In brief, the assay mixture contained 10 mM of p-nitrophenyl phosphate in 0.1 M of sodium carbonate buffer, pH 10, supplemented with 1 mM MgCl2, followed by an incubation at 37°C for 30 minutes. After adding 0.1 M of NaOH, the amount of p-nitrophenol liberated was measured by spectrophotometer.
Plasmid construction and complementary DNA transfection
To facilitate the analysis of Smad1 as a transcriptional activator, we generated the construct Gal4-Smad1 by fusing complementary DNA (cDNA) for full-length Smad1 C-terminally to the Gal4 DNA binding domain.21 The construct containing full-length Smad2 or Smad3 instead of Smad1 was generated also. The pGL-Gal4-UAS (17 mer X 2-β-globin promoter-luciferase) containing Gal4 binding site22 was constructed as a reporter plasmid. Constructs of a constitutively active mutant RasV12 and a dominant negative mutant RasN17 cDNA were as described.23 Constructions of a constitutively active form of BMPR-IA and TGF-β type I receptor (TGF-βR-I) were reported.24 The construct of a phosphatase for ERK (CL100) that inactivates ERK was as described.25 For transient transfection, cells at 50% confluence in 12-well plates were transfected using LipofectAMINE PLUS (Gibco-BRL, Rockville, MD, USA) with indicated plasmids. The total amounts of DNA were adjusted by supplementing up to 1.0 μg of an empty vector. Indicated plasmids (0.25 μg each) with various amounts (0, 12.5, 25, and 50 ng) of RasV12 cDNA and/or 0, 25, and 50 ng of RasN17 cDNA were added to cultures. When indicated, cells were treated with 100 μg/liter of rhBMP-2, 18 nM of EGF, 100 μM of PD98059 (an inhibitor for MEK1 that activates ERKs), or 0.1 μM of wortmannin (an inhibitor for phosphatidylinositol 3-kinase [PI3-K]).
Firefly and Renilla luciferase activities were assayed with the dual luciferase assay system (Promega, Madison, WI, USA). Firefly luciferase activity was normalized with respect to the Renilla luciferase activity.
Immunoblot analysis
Cells were transiently transfected with RasV12 or empty control vector along with or without CL100 and then serum-starved for 12 h before treatment with 18 nM of EGF for 30 minutes. Cells were washed twice with ice-cold PBS and lysed in 10 mM of Tris-HCl and 1% sodium dodecyl sulfate (SDS), pH 7.4, with boiling for 5 minutes. Equal amounts of samples were electrophoresed on 10% SDS-polyacrylamide gel electrophoresis (PAGE) gels (Daiichi Pure Chemicals Co., Tokyo, Japan). After electrophoresis, proteins were transferred to Immobilon-P (polyvinylidene difluoride [PVDF] membrane from Millipore Co., Bedford, MA, USA). Activated ERKs were detected by incubation with the mouse monoclonal antibody against phospho-ERK1/2 followed by enhanced chemiluminescence detection using horseradish peroxidase-conjugated anti-mouse IgG second antibody (Amersham Pharmacia Biotech, Tokyo, Japan).
Statistical analysis
Data were expressed as means ± SE and analyzed by one-way analysis of variance (ANOVA) followed by Bonferroni's method. A value of p < 0.05 was considered to be significant.
RESULTS
Interaction with collagen enhances effect of BMP-2 on osteoblastic cells
BMP signals and collagen production are both important for osteoblastic differentiation.11, 26 Treatment with 100 μg/liter of rhBMP-2 induced up to a 12-fold increase in ALP activity, an early differentiation marker of osteoblasts, in osteoblastic MC3T3-E1 cells (Table 1). When the cells were treated with 0.3 mM of l-azetidine-2-carboxylic acid to inhibit collagen production, ALP activity decreased to approximately 50% of the basal level in the absence of rhBMP-2 as reported previously.26 The increase in ALP activity by rhBMP-2 also was significantly inhibited in the presence of L-azetidine-2-carboxylic acid (Table 1). L-Azetidine-2-carboxylic acid is a proline analogue and its inhibitory effect was restored by an addition of 10 times higher concentration of L-proline in the same experiment, indicating that synthesized collagen per se was required for the BMP-2 action. Treatment of cells with the blocking antibody against α2β1-integrin, a collagen receptor, also inhibited the BMP-2 action, although the inhibitory effect was partial (Table 1). Thus, BMP-2 action in osteoblastic cells required collagen production and was sustained by collagen-α2β1-integrin interactions at least in part.
|
|
Aggregation of the β1-integrin by collagen and some other matrix proteins activates FAK and its downstream pathways.12 We have shown that FAK-ERK is involved in the osteoblastic differentiation using MC3T3-E1 cells stably expressing antisense messenger RNA (mRNA) for FAK.8 FAK activates diverse signaling pathways via Ras27 and PI3-K.28, 29 Our previous report shows that inactivation of PI3-K by wortmannin has no effect on ALP activity in MC3T3-E1 cells, whereas ERK inhibition suppresses it.8 Hence, Ras-ERK pathway downstream of FAK may be involved in signals activated by BMP-2 in osteoblastic cells.
Transcriptional activity of Smad1 in osteoblastic cells
BMP-2 actions are mediated by Smads (Smad1 as a prototype) that are phosphorylated by BMPR-I on its ligand binding,6 and Smad-dependent transcriptional activity is necessary for osteoblastic differentiation.7, 30 Thus, it is possible that collagen-integrin-mediated activation of Ras-ERK signals converge with BMP-2 actions at the step of Smad1-dependent transcriptional activity. To address this issue, we first examined an effect of l-azetidine-2-carboxylic acid on Smad1-dependent transcriptional activity in MC3T3-E1 cells transiently expressing a fusion protein of the Gal4 DNA binding domain and Smad1 (Gal4-Smad1; Fig. 1A). These cells express BMPR-IA and BMPR-II, and respond well to 100-500 μg/liter of rhBMP-2.11 rhBMP-2 at 500 μg/liter gave a 2.5-fold induction of luciferase activity. Treatment with 0.3 mM of L-azetidine-2-carboxylic acid inhibited the luciferase activity in the presence of rhBMP-2, whereas its inhibitory effect was restored by an addition of a 10 times higher concentration of L-proline in the same experiment, indicating that synthesized collagen per se was important for Smad1 transactivation induced by BMP-2. We next examined effects of Ras and its downstream signals activated by collagen-integrin interactions on Smad1-dependent transcriptional activity in the presence and absence of rhBMP-2 in MC3T3-E1 cells (Fig. 1B). Expressing RasV12, a constitutively active mutant Ras, in the presence of rhBMP-2 further stimulated the luciferase activity to 6.5-fold over the control level, and rhBMP-2 alone gave a 2.5-fold induction (Fig. 1B). Coexpressing RasN17, a dominant negative mutant, or treatment with 100 μM of PD98059, an MEK1 inhibitor, almost completely suppressed luciferase activity stimulated by RasV12. These observations indicate that Ras activation further enhances Smad1-dependent transcriptional activity up-regulated by BMP-2 in osteoblastic cells, and that ERK may mediate the stimulatory effect of Ras.
Regulation of transcriptional activity of Smad1 by Ras-ERK pathways in osteoblastic cells. (A) MC3T3-E1 cells in 12-well plates were transfected transiently with an expression plasmid for a fusion protein of the Gal4 DNA binding domain and Smad1 (Gal4-Smad1) along with a Gal4-dependent luciferase reporter construct in the presence and absence of 500 μg/liter of rhBMP-2. After transfection, some cultures were treated with 0.3 mM of l-azetidine-2-carboxylic acid (AZC) with or without 3 mM of L-proline. The luciferase activities were determined 48 h after transfection. Data are means ± SE; n = 4. *Significantly higher than values obtained from control cultures; #significantly lower than values obtained from cells treated with BMP-2 alone. (B) MC3T3-E1 cells were transfected transiently with plasmids as mentioned previously in the presence and absence of 500 μg/liter of rhBMP-2. Cells also were transfected with a plasmid for constitutively active mutant RasV12 or dominant negative mutant RasN17. Indicated plasmids (0.25 μg each) with 50 ng of RasV12 and/or 50 ng of RasN17 were added to cultures. After transfection, some cultures were treated with 100 μM of PD98059. Data are means ± SE; n = 4. *Significantly higher than values obtained from control cultures; #significantly lower than values obtained from cells transfected with RasV12 in the presence of BMP-2. (C) HOS-TE85 cells in 12-well plates were transfected transiently with plasmids as mentioned previously. Cells also were transfected with a plasmid for constitutively active mutant RasV12 or dominant negative mutant RasN17. Indicated plasmids (0.25 μg each) with various amounts (0, 25, and 50 ng) of RasN17 and/or 50 ng of RasV12 were added to cultures. Data are means ± SE; n = 4. *Significantly higher than values obtained from cells transfected with control DNA alone; #significantly lower than values obtained from cells transfected with RasV12 but not RasN17.
The effect of Ras was examined also in human osteosarcoma HOS-TE85 cells. HOS-TE85 cells constitutively expressed a high level of BMP-2 and ALP so that rhBMP-2 added exogenously did not further increase ALP activity (data not shown). rhBMP-2 of 500 μg/liter did not stimulate the luciferase activity (data not shown), suggesting that endogenous BMP activity maintained the basal level of Smad1-dependent transcriptional activity in HOS-TE85 cells. Expressing RasV12 markedly stimulated the luciferase activity in the absence of rhBMP-2. Expression of RasN17 inhibited the luciferase activity either in the presence or absence of RasV12 expression (Fig. 1C). This result corroborates that Ras is involved in Smad1-dependent transcriptional activity in osteoblastic cells.
Regulation of transcriptional activity of Smad1by Ras-ERK
Regulation by Ras of transcriptional activity of Smad1 was estimated more precisely using COS-1 cells that were cotransfected with the Gal4-Smad1, the luciferase reporter and a constitutively active BMPR-IA constructs (Fig. 2A). Expressing active BMPR-IA gave a 6-fold induction of luciferase activity. Expressing RasV12 stimulated luciferase activity up to 6-fold in the absence of active BMPR-IA. Simultaneous expression of active BMPR-IA and RasV12 showed up to a 30-fold increase in luciferase activity, and the activation was dose-dependent on amounts of transfected RasV12 construct (Fig. 2A). When COS-1 cells were cotransfected with RasV12 and an ERK phosphatase CL100 construct, the stimulatory effect of RasV12 was significantly inhibited (Fig. 2B). Treatment with the MEK1 inhibitor PD98059 instead of CL100 transfection gave a similar result. In contrast, wortmannin did not affect Smad1-dependent transcriptional activity stimulated by the expression of RasV12, being consistent with no influence of wortmannin for osteoblastic differentiation. These results indicate that Smad1-dependent transcriptional activity is further potentiated by a Ras-ERK pathway in the presence of signals triggered by BMPR-IA.
Regulation of transcriptional activity of Smads by Ras-ERK pathways in COS-1 cells. (A) COS-1 cells in 12-well plates were transfected transiently with an expression plasmid for a fusion protein of the Gal4 DNA binding domain and Smad1 (Gal4-Smad1) and a Gal4-dependent luciferase reporter construct. Cells were cotransfected with a plasmid for constitutively active mutant RasV12 along with or without that for constitutively active BMPR-IA. Indicated plasmids (0.25 μg each) with increasing amounts (0, 12.5, 25, and 50 ng) of RasV12 were added to cultures. The luciferase activities were determined 48 h after transfection. Data are means ± SE; n = 4. *, Significantly higher than values obtained from control cultures; #, significantly different from values obtained from cells transfected with active BMP-IA but not RasV12. (B) COS-1 cells were transfected transiently as mentioned previously. Some cultures were cotransfected with a plasmid for CL100, an ERK phosphatase. After transfection, some cultures were treated with 100 μM of PD98059 or 0.1 μM of wortmannin. The luciferase activities were determined 48 h after transfection. Data are means ± SE; n = 4. *Significantly lower than values obtained from cells transfected with RasV12 alone; #significantly lower than values obtained from cells transfected with RasV12 and active BMPR-IA. (C) COS-1 cells were transfected transiently with an expression plasmid for a fusion protein of the Gal4 DNA binding domain and Smad3 (Gal4-Smad3) and a Gal4-dependent luciferase reporter construct. Cells were cotransfected with increasing amounts (0, 12.5, 25, and 50 ng) of a plasmid for constitutively active mutant RasV12 along with or without that for constitutively active TGF-βR-I. The luciferase activities were determined 48 h after transfection. Data are means ± SE; n = 4. *Significantly higher than values obtained from control cultures.
Ras does not enhance TGF-β-Smads signal
Expressing RasV12 in COS-1 cells transfected with a Gal4-Smad3 construct stimulated luciferase activity up to 4-fold in the absence of active TGF-βR-I (Fig. 2C). Expressing active TGF-βR-I without RasV12 showed a 5-fold increase in luciferase activity. Simultaneous expression of RasV12 along with active TGF-βR-I did not give a significant increase in luciferase activity over that obtained with active TGF-βR-I expression alone (Fig. 2C). Experiments using a Gal4-Smad2 construct showed essentially the same results (data not shown). This observation suggests that the expression of RasV12 preferentially stimulates BMP-Smad1 signals with no effect on TGF-β-Smads pathways.
EGF stimulates ERK but not Smad1-dependent transcriptional activity
Although EGF transiently activates its receptor tyrosine kinase followed by ERK activation via Ras31 and stimulates proliferation of osteoblastic cells,32 this growth factor inhibits osteoblastic differentiation.33 When COS-1 cells were treated with 18 nM of EGF for 18 h, Gal4-Smad1 transcriptional activity was not stimulated but rather was inhibited in the cells cotransfected with active BMPR-IA construct (Fig. 3A). Cotransfection of CL100 slightly but further inhibited the luciferase activity in the presence of EGF. Treatment with EGF increased the active form of ERK detected by monoclonal antibody against phospho-ERK, as did overexpression of RasV12 (Fig. 3B). In both cases, the activation of ERK was suppressed by cotransfection of CL100 plasmid. Thus, EGF did not stimulate Smad1-dependent transcriptional activity although it activated ERK, and this observation may be one of the reasons why EGF inhibits osteoblastic differentiation.
Effect of EGF on (A) transcriptional activity of Smad1 and (B) on ERK activation. (A) COS-1 cells were transfected transiently with a expression plasmid for a fusion protein of the Gal4 DNA binding domain and Smad1 (Gal4-Smad1) and a Gal4-dependent luciferase reporter construct. Cells were cotransfected with a plasmid for constitutively active BMPR-IA along with or without that for CL100. Some cultures were treated with 18 nM of EGF for 18 h until the end of incubation. The luciferase activities were determined 48 h after transfection. Data are means ± SE; n = 4. *Significantly higher than values obtained from control cultures. #Significantly lower than values obtained form cells transfected with BMPR-IA alone. (B) Cells were transfected transiently with RasV12 or empty control vector along with or without CL100 and then serum-starved for 12 h before treatment with 18 nM of EGF for 30 minutes. Cells were washed twice with ice-cold PBS and lysed in 10 mM of Tris-HCl and 1% SDS, pH 7.4, with boiling for 5 minutes. Equal amounts of samples were applied for an immunoblot analysis. Activated ERKs (pERK) were detected by incubation with the mouse monoclonal antibody against phospho-ERK1/2 followed by enhanced chemiluminescence detection using horseradish peroxidase-conjugated anti-mouse IgG second antibody.
DISCUSSION
Signals from BMPRs and cell adhesion to type I collagen are both important for osteoblastic differentiation and functions.9-11, 26 Collagen is a major component of extracellular matrix that serves as a storage for BMPs secreted by osteoblastic cells to efficiently present the ligands to their receptors.11 This study shows that integrin activation by collagen is involved also in BMP actions, because disruption of either collagen synthesis or collagen-α2β1-integrin binding inhibits the stimulatory effect of BMP on osteoblastic cells. This is consistent with an observation reported by Jikko et al. that antibodies against type I collagen and integrins block BMP actions even if the constitutively active BMP type I receptor is expressed in cells that can differentiate into osteoblasts without these antibodies.10 Thus, the next question is how intracellular signals activated by integrins influence BMP actions.
Collagen-integrin interactions might be involved directly in BMP signals, because Smad1 transactivation induced by BMP-2 was suppressed in osteoblastic cells treated with l-azetidine-2-carboxylic acid, an inhibitor for collagen synthesis. Thus, signals activated by the α2β1-integrin could potentiate BMP actions through their direct effects on BMP-dependent Smad1 transcriptional activity. Activation of the β1-integrin family triggers diverse intracellular signal transduction pathways, including an FAK-Ras-ERK cascade.34 Inversely, disruption of collagen-α2β1-integrin interactions by inhibitors for collagen synthesis or by antibodies against α2β1-integrin or that of focal adhesions by cytochalasin D reduces FAK activity in osteoblastic cells as reported previously.8 Inactivation of FAK results in a decrease in ERK activity and inhibits osteoblastic differentiation.8 FAK activates diverse signaling pathways including Ras27 and PI3-K.28, 29 Inactivation of PI3-K by wortmannin has no effect on ALP activity, whereas ERK inhibitors suppress it.8 Furthermore, an overexpression of dominant negative Ras in MC3T3-E1 cells using a recombinant adenovirus technique and that of CL100 inhibited the increase in ALP activity in response to BMP-2 (M. Suzawa and Y. Takeuchi, unpublished observations, 2000). It also has been reported that ERK is involved in BMP-2-induced osteoblastic differentiation of pluripotent mesenchymal C3H10T1/2 cells.16 Taken together, Ras-ERK pathway downstream of FAK may be involved in signals activated by BMP in osteoblastic cells. Present studies corroborate this notion by indicating that Ras-ERK cascade augmented the transcriptional activity of Smad1 in response to BMP (Fig. 4). It also is interesting to note that this effect is somewhat specific to Smad1, because transcriptional activity of Smad2 and Smad3 in response to activation of TGF-βR-I was not enhanced by constitutively active Ras. This observation is relevant to the fact that TGF-β in contrast to BMP inhibits the differentiation of osteoblastic cells.35
A schematic illustration of cross-talks among intracellular signaling pathways that regulate osteoblastic differentiation. We describe three extracellular signals that regulate osteoblastic differentiation in this study. BMP-2 binds to its receptors (BMPR-I and BMPR-II) and activates Smad1 as an essential transcription factor for osteoblastic differentiation. Collagen binds to α2β1-integrin to activate FAK that subsequently stimulates several signaling molecules including Ras and PI3-K. Using Ras mutants and inhibitors (PD98059, CL100, and wortmannin), Ras but not PI3-K potentiates Smad1 transcriptional activity through ERK activation. Ras-ERK pathway is activated also by EGF, whereas signaling molecules other than ERK inhibit Smad1 transactivation. Then, effects of EGF on osteoblastic differentiation are inhibitory.
EGF is a representative stimulator of an Ras-ERK cascade via its receptor tyrosine kinase.31 Although this is the case for osteoblastic MC3T3-E1 cells, EGF does not stimulate ALP activity in the presence and absence of BMP-2 (Y. Takeuchi, unpublished observations, 2000) but stimulates cell growth.32 Our observation that EGF inhibited Smad1-dependent transcriptional activity suggests a mechanism by which EGF activates ERK but inhibits osteoblastic differentiation. Kretzschmar et al. have reported also that transcriptional activity of Smad1 is inhibited by treatment of COS cells with EGF.18 Our results that overexpression of active Ras enhanced but EGF suppressed Smad1-dependent transcriptional activity may indicate that signaling molecules other than ERK are involved in its transcriptional activity. Coexpression of an ERK phosphatase CL100 inhibited Smad1 transactivation in response to BMP-2 in the presence of EGF. These observations suggest that ERK generally is involved in Smad1 transactivation if Ras is activated by several stimuli including EGF, and that signals other than Ras-ERK activated by EGF inhibit Smad1 transactivation (Fig. 4). This notion is consistent with a previous report that ERK is required for BMP-2-induced osteoblastic differentiation of mesenchymal cell line C3H10T1/2.16 AP-1 activation in addition to ERK may regulate transcriptional activity of Smad1, because Ras-Raf-AP-1 is involved in BMP-4 signals13 and because AP-1 complexes cooperate with Smads.36 Taken together, it is suggested that Smad1 is not only a second messenger of BMP signals but a mediator of other signals activated by extracellular stimuli that promote osteoblastic differentiation.
The linker domain of Smad1 has four consensus sequences for phosphorylation by ERK, and Kretzschmar et al. have reported that Smad1 is eliminated from the nucleus on phosphorylation of the linker domain by EGF, resulting in loss of its transcriptional activity.18 However, Gal4-Smad1, used in both studies by Kretzschmar et al. and by us, could be localized spontaneously in the nucleus where its Smad1-dependent transcriptional activity is regulated, because Gal4 itself has a nuclear translocation motif. Thus, Smad1-dependent transcriptional activity itself should be regulated by EGF and Ras-ERK signals. This notion is particularly intriguing when Smad1 cooperates with other transcription factors that directly bind to their specific DNA sequences. For example, Nakashima et al. have reported synergistic signaling by the STAT3-Smad1 complex bridged by p300 in fetal brain, where BMP-2 stimulates astrocyte differentiation only in the presence of the leukemia inhibitory factor that activates STAT3.37 Hanai et al. have reported similar observations that several Smad proteins bind to transcription factors in a Runx/PEBP2/Cbf family and that Smad3, for example, cooperates with PEBP2αC/Cbfa3 to stimulate transcription of the germ line Ig Cα promoter.38 Thus, regulation of Smad1-dependent transcriptional activity is to be important in diverse aspects. Cooperation of BMP-Smads with Runx2/Cbfa1 among the Runx/PEBP2/Cbf family is particularly important for osteoblastic differentiation, because both of them are critical transcription factors in osteoblastogenesis.7, 39, 40
Generally, it is believed that cell-matrix interactions are critical not only for morphogenesis during development but for maintaining structural integrity after development such as bone remodeling; however, it is not yet well understood how they modulate cell-specific differentiation induced by growth factors, cytokines, and ligands for nuclear transcription factors. Here, we show a cross-talk between Ras-ERK cascades that can be activated by cell-matrix interactions and the BMP-Smad1 pathway, and this may provide an example of a framework for understanding osteoblastic differentiation.
Acknowledgements
We thank Tomoko Osaki-Kikuchi for technical assistance and Eisuke Nishida (Kyoto University) for providing us with the constructs of RasV12, RasN17, and CL100. This work was supported in part by grants-in-aid for Scientific Research from the Ministry of Education, Science and Culture (Y.T. and S.F.) and by grants from The Vehicle Racing Commemorative Foundation (Y.T.).
