Smad2 Overexpression Enhances Smad4 Gene Expression and Suppresses CBFA1 Gene Expression in Osteoblastic Osteosarcoma ROS17/2.8 Cells and Primary Rat Calvaria Cells*

Mothers againstdecapentaplegic-related proteins (Smads) are essential intracellular components for the signal transduction of transforming growth factor-β (TGF-β) family members. Smad1 mediates bone morphogenetic protein (BMP) signals, whereas Smad2 functions downstream of TGF-β. TGF-β is expressed in osteoblastic cells and acts as an autocrine and/or paracrine factor in regulation of osteoblastic functions. In this study, we examined the levels and functions of Smad2 in osteoblastic cells. Smad2 mRNA expression was hardly detectable by Northern blot analysis in an osteoblast-like cell line, ROS17/2.8, as well as in primary rat calvaria (PRC) cells. Overexpression of Smad2 gene enhanced endogenous Smad4 gene expression in both ROS17/2.8 and PRC cells, while Smad3 levels were not altered. Smad2 overexpression suppressed osteocalcin mRNA expression in ROS17/2.8 cells. Furthermore, Smad2 overexpression also suppressed transcriptional activity of the 1-kilobase pair osteocalcin gene promoter, which was linked to chloramphenicol acetyltransferase reporter gene in both ROS and PRC cells. Since core binding factor A1 (CBFA1) is involved in osteocalcin gene expression, we further examined CBFA1 expression in the Smad2-overexpressing ROS17/2.8 and PRC cells. The levels of CBFA1 mRNA were suppressed by the overexpression of Smad2 by about 50% in both ROS17/2.8 and PRC cells. TGF-β treatment enhanced Smad4 expression in PRC cells, and this TGF-β effect was blocked by the cotreatment with BMP, indicating that TGF-β signaling pathway is interfered by BMP. These data indicate that Smad2 regulates Smad4 specifically and that CBFA1 gene is one of the downstream targets of Smad2.

The TGF-␤ 1 superfamily is a large family of multifunctional ligands that regulate cellular growth and differentiation. Among them, TGF-␤ signals through distinct heteromeric receptor complexes including type I and type II serine/threonine kinase type receptors. Activation of the receptor complex initiates upon the binding of the ligand to type II receptor, which then recruits and phosphorylates the GS domain of type I receptor to activate it (1)(2)(3). The activated type I receptor then propagates the signal to downstream targets including Smads (4,5).
Smads act as well conserved components in TGF-␤ family signal transduction pathway and have been identified in a variety of species including fruit fly and humans. Smad1 and Smad2 are rapidly and specifically phosphorylated by BMP2 and TGF-␤, respectively (4 -7), and translocated to nuclei to be involved in regulation of gene expression. Smads are highly conserved across species and share conserved amino-and carboxyl-terminal regions termed MH1 and MH2 domains, respectively (8,9). The main active domains of the Smad proteins appear to be located in the carboxyl-terminal MH2 region. The activities of the MH2 domain are masked by the presence of amino-terminal MH1 domain, whereas they are unmasked upon the removal of the inhibition by MH1 domain upon activation, possibly by phosphorylation (9).
The gene for Smad2 has been mapped to the site closely linked to Smad4 (also called deleted in pancreatic carcinoma 4, DPC4) on chromosome 18q21 (10), a region deleted in some of the human cancers, for instance pancreatic carcinoma, colorectal carcinoma, and ovary or lung carcinoma (10 -14). Missense mutations of Smad2 or Smad4/DPC4 gene lead to either loss of protein and/or loss of TGF-␤-regulated responses (10,15).
TGF-␤ is most abundantly stored in bone matrix in the body. It is produced by osteoblasts and appears to regulate bone metabolism in various ways, including skeletal development and bone remodeling (16). It modulates the expression of several markers of the osteoblastic phenotype. Although TGF-␤ promotes extracellular matrix production, it inhibits some features of fully differentiated osteoblastic phenotype, such as osteocalcin expression (17). Osteocalcin, a bone-specific calcium-binding protein, is a major non-collagenous component of the bone matrix and acts as a suppressor of bone formation as shown in knock-out mice (18). It is expressed during differentiation of normal rat osteoblasts (19 -21) and is produced constitutively in a rat osteosarcoma cell line, ROS17/2.8 (22). Osteocalcin expression is down-regulated by TGF-␤ 1 treatment in both normal osteoblasts and osteosarcoma cells (17,23).
PEBP2/CBF (core binding factor) was originally identified as a polyomavirus enhancer-binding protein 2 (24) and later as a core binding factor (25). The two names refer to an identical molecule. CBF is a complex of two different subunits, A and B. CBFA directly binds to DNA, while CBFB does not interact directly with DNA, but it associates with A subunit to increase DNA binding affinity of the A subunit (25,26). CBFA contains a 128-residue domain, homologous to the Drosophila pair-rule gene, runt. CBFA1 is involved in regulation of T cell gene expression (25), and CBFA2 is homologous to human AML-1 gene found in acute myeloid leukemia (27,28). They specifically recognize a consensus DNA binding sequence, PuAC-CPuCA. The similar response elements exist in the promoter regions of osteoblastic phenotype-related genes including the one encoding osteocalcin (29 -32). Furthermore, null mutation of CBFA1 gene in mice resulted in complete lack of ossification of their bones, indicating that CBFA1 plays a critical role in the regulation of osteoblastic differentiation (33,34).
In the present work, the expression and functions of Smad2 were examined in osteoblast-like ROS17/2.8 cells and primary rat calvaria (PRC) cells. Our data indicate that Smad2 regulates expression of Smad4 and that Smad2 also controls the expression of CBFA1 and osteocalcin genes in these osteoblastlike cells and PRC cells.

MATERIALS AND METHODS
Cell Culture-Rat osteoblastic osteosarcoma ROS17/2.8 cells were kindly provided by Dr. G. Rodan (Merck Research Laboratories, West Point, PA) and were maintained in modified F-12 medium supplemented with 5% fetal bovine serum (FBS) (17). Cells were cultured in a humidified atmosphere of 95% air and 5% CO 2 at 37°C. PRC cells were prepared by five sequential enzymatic digestions using collagenase (0.2%) and Dispase (2.4 units/ml) of fetal (18 days post coitum) calvariae, and the cells in the last three fractions were pooled and were cultured in modified F-12 medium supplemented with 5% FBS. Human recombinant TGF-␤1 was purchased from R&D Systems (Minneapolis, MN). Human recombinant BMP2 was a kind gift from Genetics Insti- Transient DNA Transfection-Cells were plated at 6 ϫ 10 4 cells/cm 2 24 h before transfection. Transfection of plasmid DNA into ROS17/2.8 cells was performed by using DNA-lipid complexes (LipofectAMINE, Life Technologies, Inc.). pBluescript SK (ϩ) (pBS) plasmid was used as a control. Smad2-MH2 expression vector is described elsewhere (10). The cells were exposed to a complex of DNA (0.2 g/cm 2 ) and Lipo-fectAMINE for 8 h in serum-free medium. The cells were then cultured in fresh medium supplemented with 5% FBS and were harvested after 72 h of transfection.
cDNAs-The cDNAs encoding human Smad2 and Smad1 linker regions, Smad2SA and Smad1S, were described elsewhere (4,10). A 264-base pair fragment of the Smad2 linker region and a 375-base pair fragment of the Smad1 linker region were excised with SacII/AccI and were used as specific probes for Smad2 and Smad1, respectively. Smad4 plasmid was provided by Dr. Kern (11), and BamHI/EcoRI fragment was used as a probe. A fragment of about 1 kb excised by EcoRI/SalI from Smad3 plasmid was used (35). Osteocalcin (OC) and alkaline phosphatase (AP) cDNAs were a gift from Dr. G. Rodan, and EcoRI fragments were used as probes. CBFA1(PEBP2␣A) and CBFB-(PEBP2␤) cDNAs were described elsewhere (24,36), and fragments were used as probes.
RNA Isolation and Northern Blot-Total cellular RNA was prepared according to the acid guanidium thiocyanate-phenol-chloroform method (37). Aliquots of 10 -15 g of the total RNA per lane were electrophoresed in 1.0% agarose gels containing 0.66 M formaldehyde and was transferred to nylon filters (Hybond-N, Amersham Pharmacia Biotech) by electroblotting. Filters were prehybridized overnight at room temperature. Each cDNA was labeled by random primer method using Klenow fragment (Amersham Pharmacia Biotech) and [␣-32 P]dCTP (NEN Life Science Products). Hybridization was performed at 42°C for 18 h. Filters were washed in 1ϫ SSC, 0.1% SDS for 15 min at room temperature, and 0.2ϫ SSC, 0.1% SDS for 20 min at 65°C. Washed filters were exposed to x-ray film using intensifying screens at Ϫ80°C for several days.
CAT Assay-The cells were cotransfected with a CAT reporter plasmid (38) containing a fragment of rat osteocalcin promoter (Ϫ 1094 to ϩ 147) and Smad2-MH2 expression vector or pBS as a negative control. pSV 2 CAT plasmid was used as a positive control. Cell extracts were prepared and used for the analysis of CAT activity. Protein concentrations in the lysates were determined by the Coomassie Brilliant Blue G method (39). Equivalent amounts of the cellular proteins were incubated in a reaction buffer (0.25 M Tris-HCl, 40 mM acetyl CoA, [ 14 C]chloramphenicol) overnight at 37°C. The levels of acetylation were examined by TLC followed by autoradiography of the TLC plates. Quantitation of the acetylation levels was performed by using a laser densitometer. Experiments were repeated three times in triplicate with independent preparations of cell extracts.
Statistical Analysis-Statistical evaluations of the data were conducted by using Student's t test for per-comparison analysis. The data were based on three independent experiments and are presented as mean Ϯ standard deviation (S.D.). Statistical significance (p Ͻ 0.01 or p Ͻ 0.05) is indicated by an asterisk (*).

Overexpression of Smad2 Enhances Expression of Smad4 mRNA Levels in ROS17/2.8 Cells and Primary Rat Calvaria
Cells-We previously showed that TGF-␤ regulates expression of osteoblastic phenotype-related genes in ROS17/2.8 cells (17). To elucidate whether Smad2 is expressed in these cells, we first examined its mRNA level. Smad2 mRNA expression was hardly detectable in ROS17/2.8 cells by Northern blot analysis (Fig. 1a, lane 2). As even a low level of Smad2 may be still functional, we further examined Smad2 function by overexpressing an active domain of Smad2 (Smad2-MH2) in ROS17/ 2.8 cells. ROS17/2.8 cells where Smad2 was overexpressed showed exogenous Smad2 mRNA expression (Fig. 1a, lane 1; exogenous Smad2 is indicated by an asterisk (*)). Overexpression resulted in one major 3.6-kb and one minor 1.1-kb transcript in Smad2-overexpressed ROS17/2.8 cells (Fig. 1a, lane  1). Smad2 untransfected ROS17/2.8 cells express two Smad4 mRNA (a major 3.6-kb and a minor 7.8-kb) species at moderate levels (Fig. 1b, lane 2). Smad2 overexpression enhanced the levels of the major 3.6-kb Smad4 mRNA expression; the abun-dance of the minor 7.8-kb band was slightly reduced (Fig. 1b,  lane 1). This effect was specific to Smad4, as Smad3 and Smad1 mRNA levels were not altered (Fig. 1, c and d).
To examine whether the results obtained in ROS17/2.8 cells could be observed in primary culture osteoblasts, we overexpressed Smad2 in PRC cells. PRC cells also expressed only low levels of Smad2 (Fig. 1e, lane 2); exogenous Smad2 transfection resulted in the appearance of four different Smad2 transcripts (two major 8.0-and 3.6-kb bands and two minor 10.0-and 1.1-kb bands) (Fig. 1e, lane 1). In the case of PRC cells, major Smad4 mRNA band size was 3.6 kb (Fig. 1f, lane 2). As in ROS17/2.8 cells, overexpression of Smad2 enhanced Smad4 mRNA expression in PRC cells (Fig. 1f, lane 1). In addition, expression of an extra higher molecular weight Smad4 mRNA species (10.0 kb) was also induced (Fig. 1f, lane 1).
Overexpression of Smad2 Suppresses Expression of Osteocalcin mRNA Level-We then examined whether Smad2 overexpression could affect expression of osteoblastic phenotypic markers in these cells. Smad2 overexpression suppressed osteocalcin mRNA level (Fig. 2, lane 1 versus lane 2) in ROS17/2.8 cells. Whereas AP mRNA level (2.5 kb) was not altered, Smad2 overexpression enhanced expression of a faint 3.8-kb band detected by the AP probe, although the nature of this band is not known (Fig. 2, lane 1). Smad1 overexpression did not affect the expression of these genes (data not shown). In PRC cells, since osteocalcin mRNA level was very low at least in the early period of culture (7 days) that we used in this experiment, we were not able to detect the basal osteocalcin mRNA levels as well as suppression by Smad2 overexpression (data not shown).
Overexpression of Smad2 Suppresses Osteocalcin Gene Promoter Activity in ROS17/2.8 Cells and Primary Rat Calvaria Cells-To determine whether Smad2 is involved in regulation of the activity of osteocalcin promoter in ROS17/2.8 cells, we investigated the Smad2 effect on the activity of a 1-kb fragment of rat osteocalcin gene promoter. Overexpression of Smad2-MH2 suppressed moderately (by about 30%) but reproducibly the transcriptional activity of a 1-kb osteocalcin gene promoter, which was linked to CAT (OC-CAT) (Fig. 3, a, lanes 4 -6, and  b). This suppression was specific to the osteocalcin promoter since Smad2-MH2 cotransfection did not affect transcriptional activity of pSV 2 CAT (Fig. 3, a, lanes 7-12, and b).
Although osteocalcin mRNA level was very low in the PRC cells, activity of the transfected 1-kb osteocalcin promoter was detectable. The basal level was relatively low; however, overexpression of Smad2 still suppressed the osteocalcin promoter activity in this in vitro system by about 50% (Fig. 3c, lanes 4 -6  compared with lanes 1-3; Fig. 3d), similarly to the observation in ROS17/2.8 cells.
Overexpression of Smad2 Suppresses Expression of CBFA1 mRNA but Enhances CBFB mRNA Levels in ROS17/2.8 Cells and Primary Rat Calvaria Cells-In order to explore whether expression of CBFs is regulated by molecules involved in TGF-␤ signaling pathway, we examined CBFA1 and CBFB expression levels in Smad2-overexpressing cells. We found that CBFA1 mRNA was expressed as a 5.5-kb band in ROS17/2.8 cells (Fig. 4a, lane 2) and overexpression of Smad2 suppressed its expression by about 50% (Fig. 4, a, lane 1, and b). CBFB was also expressed as a 3.5-kb band in these ROS17/2.8 cells, although at a low level (Fig. 5a, lane 2). In contrast to the suppression on CBFA1 mRNA level, Smad2 overexpression enhanced CBFB expression over 20-fold for the 3.5-kb mRNA species (Fig. 5, a, lane 1, and b). In the Smad2-overexpressed cells, a faint high molecular weight mRNA species (8.5 kb) was also observed (Fig. 5a, lane 1).
In PRC cells, CBFA1 mRNA was expressed as a major 5.5-kb band and a minor 9.5-kb band (Fig. 4c, lane 2). Overexpression of Smad2 in PRC cells suppressed 5.5-kb mRNA level, but enhanced 9.5-kb mRNA level (Fig. 4c, lane 1). Comparison of the sum of quantified values of these two bands in lane 1 with the band in the control (lane 2) indicated suppression by Smad2 overexpression in PRC cells (Fig. 4d). CBFB mRNA was barely expressed in PRC cells (Fig. 5c, lane 2), while Smad2 overexpression enhanced 3.5-kb mRNA expression (Fig. 5, c, lane 1,  and d). In addition, expression of two higher molecular weight species, 8.5-and 12.0-kb mRNA bands, were induced by Smad2 overexpression (Fig. 5c, lane 1).

TGF-␤ Enhances Smad4 Expression while Co-treatment with BMP Results in Loss of the TGF-␤ Effect-To examine whether
Smad4 expression is also regulated by ligand-dependent signals, effects of TGF-␤ was examined. As shown in Fig. 6, TGF-␤ treatment enhanced Smad4 expression in PRC cells (Fig. 6,  lane 1 versus lane 4). Interestingly, BMP treatment also enhanced Smad4 expression in these cells (Fig. 6, lane 2 versus lane 4). Correspondingly, Smad1 overexpression also enhanced Smad4 expression in PRC cells (data not shown) as well as Smad2 overexpression (Fig. 1f, lane 1). In order to examine possible interplay between TGF-␤ signals and BMP signals, cotreatment with TGF-␤ and BMP was conducted. Enhancement of Smad4 expression by either of the two cytokines alone was blocked by the cotreatment with TGF-␤ and BMP (Fig. 6,  lane 3). On the other hand, neither TGF-␤ nor BMP affected Smad4 mRNA levels in ROS17/2.8 cells (data not shown).
Data are presented as mean Ϯ S.D. Statistically significant difference existed between Smad2/OC-CAT cotransfection group and pBS/OC-CAT control (p Ͻ 0.01; indicated by an asterisk). There was no difference between the two positive control groups. c, CAT assay were carried out by using the lysates prepared from PRC cells as described under "Materials and Methods." d, quantification of the data presented in c. The results were obtained in triplicate from three independent cell extract preparations. Data are presented as mean Ϯ S.D. Statistically significant difference existed between Smad2/OC-CAT cotransfection group and pBS/OC-CAT control (p Ͻ 0.01; indicated by an asterisk). There was no difference between the two positive control groups.  (a and b) and PRC cells (c and d) were maintained in modified F-12 medium supplemented with 5% FBS and were cotransfected with Smad2-MH2 expression vector and 1-kb OC-CAT reporter gene plasmid (lanes 4 -6). pBS plasmid was used as a negative control (lanes 1-3). pSV 2 CAT plasmid was used as a positive control (lanes 7-12 in a and lanes 7-10 in c). Transient DNA transfections were performed as described under "Materials and Methods." a, CAT assays were carried out by using the lysates prepared from ROS17/2.8 cells as described under "Materials and Methods." b, quantification of the data presented in a. The results were obtained in triplicate from three independent cell extract preparations.

DISCUSSION
In the present report, we showed that Smad2 overexpression enhanced Smad4 mRNA expression and suppressed CBFA1 expression in ROS17/2.8 cells as well as PRC cells. These observations suggest that overexpressed Smads could be triggering a positive feedback system since Smad4 is their partner to form heteromers to be fully active. Although Smad molecules mediate signals for diverse members of TGF-␤ superfamily, Smad4 is the only common partner for the other Smads (Smad1, 2, 3, and 5). It appears that Smad4 may play an important role in adjusting different Smad pathways. A recent report in which Smad4 acts as TGF-␤-inducible DNA-binding protein further indicates its key role in signal transduction (41).
Smad3 is a close homologue of Smad2 (42)(43)(44), which has been reported to play a role similar to Smad2 in mediating TGF-␤ signal transduction. Another Smad family member, Smad1, is considered to mediate BMP signals. We observed that Smad3 and Smad1 were expressed constitutively in ROS17/2.8 cells; however, the levels of Smad3 and Smad1 were not affected by the overexpression of Smad2-MH2, indicating the specificity of the effects of Smad2 overexpression on Smad4 levels. These observations also indicate that Smad2 and Smad3 might have independent signaling pathways, which could mediate different aspects of TGF-␤ actions.
Our results indicate not only that cross-talks among the different Smads family members are present but also that CBFA1 gene is the downstream target of these Smads. TGF-␤ suppresses osteocalcin production (17), and osteocalcin promoter activity is under the control of CBFA1 as reported previously (33,34,45). Whether Smad2 suppression of CBFA1 could be involved at least in part in osteocalcin promoter suppression by TGF-␤ is being investigated by using cells derived from CBFA1 knock-out mice. We also observed that TGF-␤ inhibited CBFA1 mRNA expression in the presence of BMP, which enhanced CBFA1 expression in ROS17/2.8 cells and PRC cells (data not shown). Although treatment with TGF-␤ alone did not suppress CBFA1 mRNA levels, it appears that under physiological condition where both BMP and TGF-␤ are likely to be present at the same time, the role of TGF-␤ would be to inhibit BMP-induced enhancement of CBFA1 expression as a part of the cytokine network.
With regard to the CAT assay, ROS17/2.8 cells showed relatively high osteocalcin promoter activity, and it was suppressed by Smad2 overexpression. On the other hand, although PRC cells revealed low activity of osteocalcin promoter, Smad2 overexpression still suppressed the activity. Thus, Smad2 suppression was observed in relatively immature PRC cells as well as in relatively mature ROS17/2.8 cells.
The role of CBFB has been described as a binding partner of CBFA to enhance its binding affinity to DNA. CBFB knock-out mice show similar phenotype to CBFA2 mutant mice, which are embryonic lethal due to the failure in fetal hematopoiesis and to the hemorrhage in central nervous system (46,47). The enhancement of CBFB mRNA level by Smad2 overexpression suggest that CBFB may also be a downstream target of Smad2 and it may contribute to the modulation of the transcriptional mechanism, such as facilitating the association of CBFA1 to other transcription factors. It has been reported that CBFB is mainly located in cytoplasm and its level is increased by the differentiation of skeletal myogenic cells (48). CBFB has also been reported to interact at a high affinity with cytoskeleton (49), suggesting that CBFB could likely function in as yet unidentified aspects besides the function as a subunit of CBF transcription factors. It is also possible that Smad2 activates other CBFA isoform actions via the increase in CBFB; however, this is still a speculation and needs to be elucidated.
Similar to the Smad2 enhancement of Smad4 expression, treatment of the PRC cells with TGF-␤ enhanced Smad4 expression. Cotreatment with BMP blocked the TGF-␤ effect on Smad4, suggesting a certain interplay between the two cytokines. Intriguingly, BMP alone enhanced Smad4 expression. At this point, the mechanism of the observed inhibitory actions of the BMP against TGF-␤ enhancement of Smad4 expression is not known. Either alterations in phosphorylation status of the Smad members or the interaction among the pathway-restricted Smads and inhibitory Smads (such as Smad6 or Smad7) should be examined to elucidate the interactive phenomenon between TGF-␤ and BMP in regulation of Smad4 expression. On the other hand, treatment with TGF-␤ and/or BMP did not show regulation of Smad4 mRNA levels in ROS17/ 2.8 cells, suggesting the presence of different signaling pathways at the level of ligand/receptor in ROS17/2.8 cells compared with PRC cells (data not shown).
It has been well described that Smad1, 2, 3, and 5 have the carboxyl-terminal phosphorylation sites which are directly phosphorylated by type I receptor. Smad4 does not have this site and cannot be phosphorylated by type I receptor (9). It has also been reported that the mRNA levels of these Smads are not regulated by the treatment with their ligands (40,50). It seems that alteration in phosphorylation status may be more efficient and quicker in response to ligands binding. We also observed that Smad1 and Smad2 mRNA levels were not regulated by TGF-␤ or BMP treatment (data not shown). Interestingly, Smad4 mRNA levels were regulated by TGF-␤ and/or BMP treatment in PRC cells. Combining with the data that Smad4 was specifically enhanced in Smad2-overexpressed cells, Smad4 plays a key role in regulating signal transduction. Recent studies showed that the inhibitory Smads, such as Smad6 and Smad7, are involved in negative feedback of TGF-␤-related signals and their mRNA levels are up-regulated by different ligands (40,51,52).
We also observed that Smad6 expression was dramatically enhanced by BMP treatment in PRC cells (data not shown). Although TGF-␤ treatment did not regulate Smad6 expression in these cells, it abolished the enhancement of Smad6 expression by BMP (data not shown). This feature is similar to that of Smad4, indicating both Smad4 and Smad6 would be involved in the regulation of different Smad pathways. We assume that Smad-dependent TGF-␤ signals might be regulated at least at three different levels: 1) the phosphorylation status of pathway-restricted Smads by transient binding; 2) the up-regulation of the transcription of inhibitory Smads, which can form stable association with type I receptor and block the phosphorylation of pathway-restricted Smads as a negative feedback mechanism; 3) regulation of the levels of the common-mediator Smad4, as our data showed, to be a positive feedback. The enhancement of Smad4 mRNA probably contributes to TGF-␤ signal transduction in a ligand-dependent manner. It remains to be elucidated how Smad4 exerts its balancing function between TGF-␤ and BMP signals.
TGF-␤ is a suppressor of growth in many types of cancer. Disruption of the TGF-␤ pathway in cancer has been demonstrated in several types of cancers. Inactivation mutations in Smad2 and Smad4/DPC4 were reported in colon cancers as well as other cancers (10 -13). In ROS17/2.8 cells, the level of Smad2 expression was not detected by Northern blot analysis, and this may be related to tumorigenic phenotype of these cells. However, Smad2 mRNA level was also undetectable by Northern blot in another type of osteoblast-like cell line MC3T3E1 (data not shown) as well as in PRC cells. Although Smad2 mRNA expression is difficult to detect by Northern blot, this result does not exclude the possibility that only a small number of such molecules may be enough to mediate TGF-␤ actions, such as those that our data on the Smad2 overexpression suggested in this paper.
In summary, we showed that Smad2 regulates the expression of Smad4 as well as CBFA1 in the osteoblastic osteosarcoma ROS17/2.8 and primary rat calvaria cells.