Transforming Growth Factor-β (TGF-β1) Down-regulates Notch3 in Fibroblasts to Promote Smooth Muscle Gene Expression*

Select signaling pathways have emerged as key players in regulating smooth muscle gene expression during myofibroblast and smooth muscle differentiation, an event that is important for wound healing and vascular remodeling. These include the transforming growth factor-β (TGF-β1) signaling cascade, which has been assigned multiple roles in these cells, and the Notch pathway. Notch family members have been implicated in governing cell fate in a variety of cells; however, the mechanisms are not well understood. We sought to explore how these prominent signaling mediators regulate differentiation, and in particular, how they might converge to control the transcription of smooth muscle genes. Using TGF-β1 to induce the differentiation of 10T1/2 fibroblasts, we investigated the specific function of Notch3. Overexpression of activated Notch3 caused repression of TGF-β1-induced smooth muscle-specific genes, whereas knockdown of Notch3 by small interfering RNA did not convincingly alter their expression. Surprisingly, the addition of TGF-β1 caused a significant decrease in Notch3 RNA and protein and a reciprocal increase in Hes1 gene transcription. The repression of Notch3 was mediated by SMAD activity and p38 mitogen-activated protein (MAP) kinase, whereas analysis of the Hes1 promoter revealed direct activation by Smad2 but not Smad3. Furthermore, the Hes1 repressor protein augmented Smad3 transactivation of the SM22α promoter. These results offer a novel mechanism by which TGF-β1 promotes the expression of smooth muscle differentiation genes through the inhibition of Notch3 and activation of Hes1.

Select signaling pathways have emerged as key players in regulating smooth muscle gene expression during myofibroblast and smooth muscle differentiation, an event that is important for wound healing and vascular remodeling. These include the transforming growth factor-␤ (TGF-␤1) signaling cascade, which has been assigned multiple roles in these cells, and the Notch pathway. Notch family members have been implicated in governing cell fate in a variety of cells; however, the mechanisms are not well understood. We sought to explore how these prominent signaling mediators regulate differentiation, and in particular, how they might converge to control the transcription of smooth muscle genes. Using TGF-␤1 to induce the differentiation of 10T1/2 fibroblasts, we investigated the specific function of Notch3. Overexpression of activated Notch3 caused repression of TGF-␤1-induced smooth muscle-specific genes, whereas knockdown of Notch3 by small interfering RNA did not convincingly alter their expression. Surprisingly, the addition of TGF-␤1 caused a significant decrease in Notch3 RNA and protein and a reciprocal increase in Hes1 gene transcription. The repression of Notch3 was mediated by SMAD activity and p38 mitogen-activated protein (MAP) kinase, whereas analysis of the Hes1 promoter revealed direct activation by Smad2 but not Smad3. Furthermore, the Hes1 repressor protein augmented Smad3 transactivation of the SM22␣ promoter. These results offer a novel mechanism by which TGF-␤1 promotes the expression of smooth muscle differentiation genes through the inhibition of Notch3 and activation of Hes1.
The differentiation of smooth muscle cells and myofibroblasts is characterized by the coordinate up-regulation of smooth muscle genes that include smooth muscle ␣-actin, SM22␣, and h1-calponin (1)(2)(3). These proteins serve as indicators of cell function as they are associated with the contractile properties of the cell and signify a state of maturation. A well established modulator of myofibroblast and smooth muscle cells is transforming growth factor-␤1 (TGF-␤1), 2 which has been shown to have multiple roles in vascular remodeling and wound healing (4 -6). In the vasculature, this cytokine can promote smooth muscle differentiation and inhibit proliferation and migration, all of which are compatible with a stable vessel (1,7). Conversely, TGF-␤1 has also been shown to be robustly expressed in experimental balloon injury models and can cause neointimal hyperplasia (1,7). During wound healing, TGF-␤1 activates fibroblasts to elicit contraction and the production of critical extracellular matrix components, but it is also associated with fibrosis, leading to an abundance of myofibroblasts that cause scarring (5). These data imply that the activity of TGF-␤1 is largely context-dependent and that its effect on a cell is determined by additional regulators that together establish its mode of function. TGF-␤1 has been shown to directly activate smooth muscle gene expression primarily, but not exclusively, through SMAD proteins that bind to regulatory regions of smooth muscle genes (8 -15). An important aspect in understanding TGF-␤1 function is deciphering how its downstream mediators intersect with other signaling pathways, particularly those that might oppose or enhance its actions under certain circumstances.
Notch proteins and their downstream targets also have prominent roles in vascular biology (16 -18). The Notch family is made up of four transmembrane receptor proteins. Upon ligand binding, the receptors undergo cleavages that release an intracellular domain, which translocates to the nucleus and complexes with the CSL (CBF1/Su(H)/Lag-1) transcription factor (19). The Notch/CSL complex activates transcription of target genes containing CSL binding elements, most notably members of the Hes (hairy/enhancer of split) and Hrt (hairyrelated, also referred to as Hey, CHF, HESR) family of transcriptional repressors (20). Mutations in several components of the Notch signaling cascade result in severe vascular defects in mice and zebrafish that include abnormalities in remodeling and branching and defects in arterial specification (16 -18). Notch3 is expressed in vascular smooth muscle cells and is the causative gene for cerebral autosomal dominant arteriopathy * This work was supported by National Institutes of Health Grant NHLBI00086 (to B. L.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. □ S The on-line version of this article (available at http://www.jbc.org) contains a supplemental figure. 1  with subcortical infarcts and leukoencephalopathy, an inherited disorder characterized by the progressive degeneration of smooth muscle (CADASIL) (21). Phenotypic studies of Notch3deficient mice revealed smooth muscle maturation defects, suggesting a role in differentiation (22). Consistent with this, Notch3 is down-regulated in arterial injury models (23,24). However, Notch proteins are associated with both the promotion of differentiation as well as the maintenance of an undifferentiated state (25). In smooth muscle cells, molecular studies have shown that Notch1 and Notch3 can each inhibit smooth muscle-specific gene expression (26). Furthermore, the Notch target, Hrt2, can abrogate expression of smooth muscle genes by Myocardin, a potent cardiovascular transcription factor (27)(28)(29). By contrast, others have reported activation of smooth muscle genes by Notch proteins (30,31). Taken together, these results indicate that, like TGF-␤1, Notch function is contextdependent and its actions and subsequent effect on a cell are likely based on specific interactions with other signaling pathways.
In the present study, we investigated the mechanisms associated with TGF-␤1-induced smooth muscle gene expression in relation to Notch signaling. Our data show that Notch3 blocks TGF-␤1-dependent differentiation, whereas TGF-␤1 causes a decrease in Notch3 expression through SMAD and p38 MAP kinase. Unexpectedly, a known Notch target gene, Hes1, is up-regulated by TGF-␤1, and contrary to its recognized ability to repress gene expression, Hes1 cooperates with Smad3 to activate the SM22␣ promoter. The results indicate that aside from directly regulating smooth muscle differentiation genes, TGF-␤1 acts to repress the expression of a critical differentiation inhibitor, Notch3.

Notch3 Inhibits TGF-␤1-induced Smooth Muscle Gene
Expression-Notch signaling components have been shown to both activate and repress smooth muscle genes (26 -28, 30, 31, 41, 42), and loss-of-function studies in mice show that Notch3 is required for smooth muscle maturation (22). The implications Gapdh was used as control for sample integrity. C, reporter assays to assess Notch3 activity. Luciferase reporter plasmids harboring the SM22␣-1344 promoter or CSL binding elements were cotransfected with control GFP or NICD3 plasmid, and transcriptional activity was determined by luciferase expression. RLU, relative light units. #, p Ͻ 0.0001 when compared with control minus TGF-␤1, *, p Ͻ 0.0001 when compared with control plus TGF-␤1.

TGF-␤1 Modulates Notch3 Expression
of these studies led us to examine the role of Notch3 in an established model of myofibroblast/smooth muscle differentiation. 10T1/2 fibroblasts have been useful for defining the mechanisms involved in smooth muscle gene expression as numerous studies have demonstrated their ability to differentiate toward a smooth muscle fate in response to TGF-␤1 (27,(43)(44)(45)(46)(47). Notch3 is expressed in 10T1/2 cells and in primary cultures of adult mouse aortic smooth muscle cells but not in the endothelial cell line bEND.3 (Fig. 1). This is consistent with in vivo data, demonstrating the expression of Notch1 in endothelial cells and of Notch3 in smooth muscle cells (16,48). Additionally, the Notch ligand Jagged1 is also expressed in 10T1/2 fibroblasts; thus, these cells are capable of activating Notch signaling through receptor/ligand interactions. We first examined the effect of overexpressing an activated form of Notch3 on TGF-␤1-induced differentiation. 10T1/2 fibroblasts were transfected at 80% confluency with control plasmid or a plasmid expressing the Notch3 intracellular domain (NICD3) (35) and treated with vehicle or TGF-␤1 for 24 h. Western blot and RT-PCR were performed to examine expression of smooth muscle marker genes (Fig.  2, A and B). In these experiments, NICD3 blocked the expression of smooth muscle ␣-actin, SM22␣, and calponin in the presence of TGF-␤1, suggesting that in fibroblasts, Notch3 acts to inhibit differentiation. We further verified these results using the SM22␣-promoterluciferase reporter plasmid (37) (Fig. 2C). Transfection of the SM22␣ reporter showed NICD3 strongly inhibited promoter activity and prevented its induction by TGF-␤1. In contrast, NICD3 activated a luciferase plasmid containing CSL binding elements (35), a defined Notch target, demonstrating that these repressive abilities are promoter-specific. These experiments were performed with cells cultured at high density (80% confluency at the start of transfection). Given that TGF-␤1 has been shown to have density-dependent effects on smooth muscle cells and that Notch proteins are modulated by cell-cell contact, we tested whether cell density affected the inhibitory abilities of activated Notch3. Cells were transfected at ϳ30% confluency, treated with TGF-␤1 for 24 h, and collected for analysis prior to reaching 50% confluency. Although TGF-␤1 did not activate the expression of smooth muscle genes as strongly in subconfluent conditions, overexpression of Notch3 still inhibited the expression of smooth muscle ␣-actin, SM22␣, and calponin (Supplemental Fig. 1). Overall, these results are similar to previous reports, which showed an inhibitory effect of Notch family members on smooth muscle gene expression (26,42). However, they differ from others that described Notch activating these same genes (30,31). Our findings are unique in that they demonstrate that Notch3 blocks a specific differentiation cue by preventing TGF-␤1-induction of smooth muscle genes.  . Knockdown of Notch3 using gene-specific siRNA. Control (C) or Notch3 (N3) siRNA was transfected into 10T1/2 cells using indicated concentrations, and TGF-␤1 (10 ng/ml) was added 24 h later to selected samples. Cells were harvested for Western analysis (A) and RT-PCR (B) to detect smooth muscle (sm)-expressed genes. Tubulin was used as a control to verify total amount of protein, and Gapdh was used to assess RNA integrity. C, quantitation of smooth muscle gene expression in the absence (left graph) and presence of 1, 3.3, and 10 ng/ml TGF-␤1 (right graph) using 50 nM siRNA. Pixel density for each band was determined and relative band intensity was calculated by normalizing to sample control (Gapdh or tubulin). *, p Ͻ 0.05 when compared with respective control (n ϭ 4). JANUARY 18, 2008 • VOLUME 283 • NUMBER 3

JOURNAL OF BIOLOGICAL CHEMISTRY 1327
We next asked whether eliminating Notch3 from fibroblasts would promote differentiation in the absence of TGF-␤1, and further, would exacerbate the TGF-␤1 response. Transfection of siRNA revealed that in the absence of TGF-␤1, smooth muscle ␣-actin, SM22␣, and calponin were significantly increased when Notch3 was knocked down (Fig. 3). However, Notch3 knockdown in the presence of TGF-␤1 did not considerably alter smooth muscle marker gene expression. We more closely examined this by using lower doses of TGF-␤1 in an attempt to reveal potential differences. Nevertheless, no significant changes were observed in smooth muscle marker genes in the presence or absence of Notch3 between 1 and 10 ng/ml TGF-␤1 (Fig. 3C). These data indicate that the absence of Notch3 alone is sufficient to promote differentiation and that an inducing signal, such as TGF-␤1, ablates the inhibitory effects of Notch3. In verifying the efficacy of our Notch3 siRNA, we noticed that TGF-␤1 decreased the level of Notch3 protein in control siRNA samples (Fig. 3A). We further examined this down-regulation in untransfected 10T1/2 cells over a time course of TGF-␤1 stimulation. Both Notch3 protein and transcript levels significantly decreased after 12 h of TGF-␤1 treatment and were difficult to detect at the 48-h time point (Fig. 4A). Comparable down-regulation of Notch3 was observed using TGF-␤1 concentrations between 1 and 10 ng/ml (data not shown). The reduction in both RNA and protein suggested that this activity was regulated at the level of transcription. The addition of the transcription inhibitor actinomycin D reduced the level of Notch3 in both TGF-␤1-treated and TGF-␤1-untreated cells but abolished the TGF-␤1-specific decrease of Notch3 (Fig. 4B). These data confirm that this regulation is under transcriptional control, although it could be indirect and could be dependent upon de novo transcription of a specific inhibitor that modulates other aspects of Notch3 synthesis. To determine whether this inhibition occurred in other cells, we assessed Notch3 levels in primary cultures of mouse embryonic fibroblasts and adult aortic smooth muscle cells. Both cell types showed a decrease in Notch3 expression in the presence of TGF-␤1 (Fig. 4C), signifying that this was not unique to 10T1/2 fibroblasts and likely represents a general mechanism of TGF-␤1 function.
Repression of Notch3 Requires Smad4 and p38 MAP Kinase-Our findings imply that TGF-␤1 promotes differentiation by down-regulating Notch3 and effectively removing an inhibitor of smooth muscle gene expression. To investigate the mechanism of TGF-␤1 suppression, we targeted TGF-␤1 mediators and analyzed Notch3 protein expression. SMAD proteins are the best described activators downstream of TGF-␤1 and have

TGF-␤1 Modulates Notch3 Expression
been shown to activate smooth muscle-specific gene transcription (8 -14). To determine whether Notch3 was inhibited through a SMAD-dependent pathway, we utilized siRNA directed against Smad4, a common partner of all regulatory SMADs (49). Inhibition of Smad4 abrogated Notch3 down-regulation (Fig. 5A), demonstrating its requirement to repress Notch3 expression. We additionally tested other signaling mediators reported to be downstream of TGF-␤1 (50) by using pharmacological inhibitors to block individual pathways. As shown in Fig. 5B, inhibitors directed against TGF-␤ receptor I (TGF-␤RI), MAP kinases ERK, p38, and JNK, protein kinase A, protein kinase C, and HDAC were added prior to TGF-␤1 treatment. As expected, the TGF-␤ receptor I inhibitor prevented Notch3 repression, and additionally, inhibition of p38 MAP kinase and HDAC similarly prevented down-regulation. Interestingly, although Notch3 levels remained high in the presence of the TGF-␤1 receptor and p38 inhibitors, smooth muscle ␣-actin, and SM22␣ expression remained low, underscoring the opposing relationship of these factors in the conversion of smooth muscle cells. In fact, p38 MAP kinase has been shown to be required for TGF-␤1-dependent activation of smooth muscle genes (51). The elevated expression of Notch3 and SM22␣ in the presence of the HDAC inhibitor probably reflects direct derepression by histone acetylation of the two genes that likely overrides specific repressive activities.
Isolation of the Notch3 Promoter-The data thus far show that TGF-␤1 suppresses Notch3 expression, and this is dependent upon Smad4 and p38 MAP kinase. To further define the mechanism of inhibition, we isolated the promoter and tested its response to TGF-␤1. Comparative sequence analysis of mouse and human Notch3 genes showed a conserved region of ϳ170 nucleotides upstream of the 5Ј-untranslated region of the message (Fig. 6A). The region does not contain a recognizable TATA box but harbors an initiator sequence immediately adjacent to the start site and several GC-rich Sp1-like elements consistent with TATA-less promoters (52) (data not shown). A 513-nucleotide fragment (Ϫ430 to ϩ83) of the putative mouse promoter was cloned into the pGL3-basic-luciferase reporter construct and tested for activity in 10T1/2, aortic smooth muscle, and endothelial cells. When compared with the promoterless reporter plasmid, the Notch3 promoter exhibited an activity of ϳ7-fold in fibroblasts and smooth muscle cells and ϳ3.5-fold in endothelial cells (Fig. 6B). The activity parallels the protein expression observed in these cell types (Fig. 1), with comparatively high levels of Notch3 in 10T1/2 and smooth muscle cells, and undetectable levels in endothelial cells. We then tested the promoter for its responsiveness to TGF-␤1. As shown in Fig. 6C, TGF-␤1 treatment caused only a slight decrease in Notch3 promoter activity. Although this decrease was statistically significant (p value ϭ0.022), it does not reflect the robust decrease of endogenous Notch3 RNA and protein (Fig. 4). Thus, although the Notch3 promoter shows cell-specific activity, additional elements outside of the proximal promoter must be required for TGF-␤1 to repress its expression.
The Hes1 Gene Is Activated by TGF-␤1 and Cooperates with Smad3-Hes (hairy/enhancer of split) and Hrt (hairy-related) genes are well established targets of Notch signaling, and in particular, Hrt2 has been implicated in inhibiting smooth muscle gene transcription (20,27,28). We therefore examined the expression of these factors by RT-PCR in the presence of TGF-␤1, with the prediction that as with Notch3, the expression of these factors would also be reduced. Interestingly, however, the Hes1 transcript was increased by TGF-␤1, whereas Hrt2 mRNA  JANUARY 18, 2008 • VOLUME 283 • NUMBER 3 remained unchanged (Fig. 7A). Levels of Hes5, Hrt1, and Hrt3 RNA were present at low or undetectable levels and showed no observable changes in response to TGF-␤1 (data not shown).

TGF-␤1 Modulates Notch3 Expression
We also examined expression of the Hes1 protein in 10T1/2 cells and other cell types exposed to TGF-␤1. Western blot analysis revealed that Hes1 protein was increased by TGF-␤1 in 10T1/2 cells, as well as primary cultures of mouse embryonic fibroblasts and adult aortic smooth muscle cells (Fig. 7B).
The increase of Hes1 by TGF-␤1 is consistent with earlier studies, one of which demonstrated a cooperative interaction between Smad3 and Notch1 on CSL binding elements (53,54). Given that Notch1 is not present in 10T1/2 fibroblasts (Fig. 1) and Notch3 is down-regulated by TGF-␤1, we considered alternative modes of transcriptional activation. The mouse Hes1 promoter is responsive to Notch signaling through three CSL binding sites (35,38). Further examination of the transcriptional unit revealed two consensus SMAD elements that are conserved from mouse to human (Fig. 7C). We tested whether the Hes1 promoter was responsive to TGF-␤1 and SMAD proteins in 10T1/2 cells by cloning it into pGL3-basic-luciferase and measuring activity. Indeed, TGF-␤1 activated transcription of the reporter construct by ϳ3-fold, and similarly, Smad2, but not Smad3, caused a 5-fold increase in transcription (Fig.  7D). Moreover, TGF-␤1 also activated the Hes1 promoter in mouse embryonic fibroblasts and aortic smooth muscle cells (Fig. 7E). Taken together, these results demonstrate that Hes1 is a target of TGF-␤1 in fibroblasts and smooth muscle cells and is activated by Smad2 through elements within its proximal promoter.
Previously, two groups independently showed that Hrt/Hes proteins inhibit Myocardin-dependent smooth muscle gene expression (27,28). Given that Hes1 was up-regulated by TGF-␤1, we wanted to explore its direct role on smooth muscle transcription. Using the SM22␣ luciferase reporter, we cotransfected a Hes1 expression plasmid and measured transcriptional activity (Fig. 8). By itself, Hes1 had little effect on SM22␣ expression; surprisingly, however, it cooperated with Smad3 to amplify transcription. The enhancing abilities of Hes1 were concentration-dependent. At low amounts (200 ng), Hes1 exhibited a cooperative effect with Smad3, whereas higher concentrations (600 ng) resulted in inhibition, possibly due to its prevailing repressive function. We saw no significant difference in SM22␣ enhancer activity when Hes1 was transfected alone at high or low concentrations (data not shown), indicating that  (35), and two consensus binding elements for SMAD proteins are shown in boxes. D, transcription reporter assays with the mouse Hes1 promoter (Ϫ199 to ϩ46) cloned into pGL3-basic-luciferase. In 10T1/2 cells, the Hes1 promoter is strongly activated by TGF-␤1 and Smad2 but not Smad3. *, p Ͻ 0.001 when compared with GFP control. RLU, relative light units. E, Hes1 promoter activity in response to TGF-␤1 in mouse embryonic fibroblasts and primary cultures of adult aortic smooth muscle cells, with 10T1/2 cells for comparison. For each cell type, expression without TGF-␤1 is set to 1. *, p Ͻ 0.05 when compared with respective control. these actions were likely through direct modulation of Smad3. These results demonstrate that aside from its well documented inhibitory abilities, Hes1 has the capacity to augment transcription in collaboration with Smad3.

DISCUSSION
Smooth muscle gene expression is important for the characterization of myofibroblast and smooth muscle cell phenotypes. Their regulation is linked to defining atherosclerotic plaques, restenosis lesions, and tissue fibrosis. To date, multiple factors have been shown to have a hand in the regulation of these genes (1). Specifically, TGF-␤1 is known to play a prominent role through activation of SMADs, which directly bind and increase smooth muscle-specific gene transcription. Although SMADs clearly have a central role in TGF-␤1-dependent expression, other factors are known to act cooperatively with SMADs or independently to induce gene transcription (8 -15, 51, 55, 56). Other factors shown to be downstream of TGF-␤1 in smooth muscle cells include RhoA and RGC-32 (15,51,55). Additionally, the transcription factors serum-response factor and its cofactor Myocardin are essential for smooth muscle gene activation, and their functions are also known to intersect with the TGF-␤1 pathway (11,12). In opposition to these positive regulators, platelet-derived growth factor-BB (PDGF-BB), gut-enriched Kruppel-like (GKLF), and Notch signaling components have been shown to repress these genes. A critical question is how these signaling pathways intersect to regulate smooth muscle genes under specific conditions. In this study, we investigated the relationship of TGF-␤1 and Notch3.
Our experiments reveal that Notch3 prevents TGF-␤1 induction of smooth muscle gene transcription (Fig. 2). This is consistent with previous reports, which showed that Notch1/3 down-regulated smooth muscle genes in human vascular smooth muscle (26) and that Notch2 reduced smooth muscle ␣-actin expression in skeletal muscle (42). Our data extend these findings to demonstrate that Notch3 not only reduces expression but also blocks the induction of these genes by a differentiation mediator. However, these data conflict with other studies. Noseda et al. (30) reported that Notch1/CSL directly activated the smooth muscle ␣-actin promoter, and Doi et al. (31) described Notch1/3-dependent expression of several smooth muscle genes in multiple cell types, including 10T1/2 fibroblasts. The discrepancies between these findings are not clear. Potentially, Notch function is extremely sensitive to culture conditions because endogenous Notch is a transmembrane protein that is normally activated by ligand binding. Slight differences in cell manipulations or serum levels might explain these different responses.
By siRNA knockdown of Notch3 in fibroblasts, we observe a significant increase in smooth muscle gene expression in the absence of TGF-␤1 but not in its presence (Fig. 3). This implies that Notch3 serves as a brake on transcription, and removal of the Notch3 brake is sufficient for expression, whereas an inducer of differentiation, such as TGF␤1 acts by removing this inhibition. Consistent with this, one of the most interesting findings to come out of these analyses is that TGF-␤1 represses Notch3 expression (Fig. 4). Our data show that Notch3 downregulation is dependent upon Smad4 and p38 MAP kinase (Fig.  5). To our knowledge, these experiments are the first to demonstrate that Notch expression is governed by TGF␤1 through these specific mediators. Regulatory SMADs are transcriptional activators but possess repressive function in collaboration with other factors, such as HDACs (49). In fact, inhibition of HDAC activity by trichostatin A relieved the TGF-␤1-induced repression of Notch3 (Fig. 5B). An important question that we currently do not have an answer for is whether HDAC repression is mediated by SMAD activity. Moreover, p38 MAP kinase has been shown to activate several downstream transcription factors and can modulate SMAD activity (49,57). Of particular interest to our study is that p38 MAP kinase has been shown to be required for TGF-␤1-dependent smooth muscle gene expression by RhoA and PKN (51). Whether RhoA or PKN cause a decrease in Notch3 expression remains to be determined, but if so, this might offer an additional mechanism through which these factors govern smooth muscle transcription. Currently, it is unclear how Smad4 and p38 specifically repress Notch3 expression. Analysis of the Notch3 promoter revealed cell-selective activity (Fig. 6B); however, the region was not sufficiently repressed by TGF-␤1, indicating that additional elements are required (Fig. 6C). Further analysis of the Notch3 regulatory regions will be required to dissect out the direct mechanism of transcriptional repression.
Intriguingly, although Notch3 is repressed, the Hes1 gene is activated by TGF-␤1 (Fig. 7). This was an unexpected finding as we predicted that Hes and Hrt family members would similarly be down-regulated. Although Hes1 was earlier shown to be regulated by TGF-␤1 (53,54), our mechanism appears to be differ- FIGURE 8. Activity of the SM22␣ promoter in response to Hes1. Luciferase reporter plasmid containing the SM22␣-1344 promoter was cotransfected with indicated plasmids in 10T1/2 cells, and transcriptional activity was measured by luciferase assays. Hes1 does not repress basal expression of the SM22␣ reporter but augments Smad3 activation in a concentration-dependent manner. 400 ng of GFP, Hes1 (alone), and Smad3 plasmids were used for transfection. Varied amounts of Hes1 cotransfected with Smad3 are indicated. RLU, relative light units. *, p Ͻ 0.05 when compared with SMAD3 alone.

TGF-␤1 Modulates Notch3 Expression
ent. Blokzijl et al. (54) reported that Hes1 was up-regulated by cooperative interaction of Notch1 and Smad3 on CSL sites within the Hes1 promoter. Notch1 is not present in 10T1/2 cells at significant levels ( Fig. 1), and thus, this appears an unlikely mechanism in these fibroblasts. Scanning of the Hes1 promoter revealed SMAD consensus binding elements (Fig. 7B), and we show that Smad2, but not Smad3, activates the promoter. This is distinct from SM22␣ promoter, which is activated by Smad3, but not Smad2 (9, 12, 58 -60), and likely reflects differences in flanking sequences of the SMAD sites that influence the specificity of Smad2/3 binding.
With respect to Hes1 function, both Hes1 and Hrt2 were recently shown to strongly oppose the activity of Myocardin and serum-response factor on CArG box-dependent expression of smooth muscle genes (27,28). We show that Hes1 has no substantial effect on basal transcription of the SM22␣ promoter (Fig. 8), and unexpectedly, we further demonstrate that it cooperates with Smad3 to enhance expression of SM22␣. The difference in Hes1 activity could be explained by the fact that TGF-␤1 acts in a Myocardin-independent manner in 10T1/2 cells (61). Despite this, a study by Hinson et al. (56) indicated that Myocardin-related factors MRTFA/B, which are both expressed in 10T1/2 cells, are required for the TGF-␤1 response. At present, it is not known whether MRTF activity is also inhibited by Hes1, but it is tempting to speculate that these factors might, instead, act in concert to invoke smooth muscle gene transcription. The concentration of Hes1 appears critical as low levels enhance, whereas increasing amounts exhibit repressive activity. The exact mechanism of the activation of Hes1 is yet to be defined. A potential clue comes from studies that showed that Hes1 switched from a repressor to an activator by direct interaction with CBP (62). CBP has been implicated in the regulation of SM22␣ in collaboration with TGF-␤1 (59), and Hes1 may be a component to this mode of activation. Further studies investigating the interaction of Hes1, Smad3, and the MRTFs will be required to elucidate the exact means of action. In summary, this study provides a new mechanism by which TGF-␤1 promotes the expression of smooth muscle differentiation genes, through the inhibition of Notch3 and the activation of Hes1.