The Cyclin D1 Gene Is Transcriptionally Repressed by Caveolin-1*

The cyclin D1 gene encodes the regulatory subunit of the holoenzyme that phosphorylates and inactivates the retinoblastoma pRB protein. Cyclin D1 protein levels are elevated by mitogenic and oncogenic signaling pathways, and antisense mRNA to cyclin D1 inhibits transformation by the ras , neu , and src oncogenes, thus link-ing cyclin D1 regulation to cellular transformation. Caveolins are the principal protein components of caveolae, vesicular plasma membrane invaginations that also function in signal transduction. We show here that caveolin-1 expression levels inversely correlate with cyclin D1 abundance levels in transformed cells. caveolin-1 cyclin D1 levels, whereas caveolin-1 overexpression inhibited expression of the cyclin D1 gene. Cyclin D1 promoter activity was selectively repressed by caveolin-1, but not by caveolin-3, and this required the caveolin-1 N terminus. Maximal inhibition of the cyclin D1 gene promoter by caveolin-1 was dependent on the cyclin D1 promoter T-cell factor/lymphoid

Cellular growth induced by mitogenic stimuli is coordinated by an orderly progression through sequential and distinct phases of the cell cycle (1,2). The progression of quiescent cells from the G 0 through the G 1 phase of the cell cycle is orchestrated by interactions between components of the cell cycle regulatory apparatus (1,3). The genetic program induced by serum addition includes the activation of immediate-early gene expression, which peaks within 30 -60 min after serum stim-ulation (4,5). The induction of immediate-early genes (for example c-fos and c-jun), is under tight control of counterregulatory mechanisms that lead to transcriptional repression and/or rapid degradation of the target gene product. The c-fos gene is under autoregulatory trans-repression (6), and the JunB protein inhibits the activity and function of c-Jun (7,8). A second phase of serum-induced gene expression occurs 3-6 h after serum stimulation and is dependent on new protein synthesis (9). The induction of the G 1 phase regulatory cyclins, the cyclin-dependent kinase phosphatases, and the E2F-responsive genes contributes to the continued passage of the cell through G 1 and into the S phase (10,11).
Both the cyclin D1 and cdc25A genes are induced with characteristic delayed-early gene kinetics and contribute to the induction of DNA synthesis (3,12,13). The cyclin D1 gene product encodes a regulatory subunit of a holoenzyme that phosphorylates and inactivates pRB. Immunoneutralizing antibody and antisense expression studies demonstrated that the abundance of cyclin D1 is rate-limiting in growth factor-and mitogen-induced progression through the G 1 phase (14 -17). Mouse embryo fibroblasts derived from mice in which the cyclin D1 gene was homozygously deleted (cyclin D1 Ϫ/Ϫ ) displayed reduced cell proliferation (18).
Several lines of evidence suggest that c-Fos and c-Jun may induce the cyclin D1 gene and thereby enhance S phase entry. Thus, c-Fos was shown to induce the cyclin D1 gene (19), and the low levels of cyclin D1 in mouse embryo fibroblasts derived from c-fos/FOSB Ϫ/Ϫ mice were rescued by c-Fos overexpression (19). In addition, c-jun Ϫ/Ϫ mouse embryo fibroblasts display a proliferative defect in response to serum and a reduction in cyclin D1 abundance (20).
Serum and growth factor signaling to discrete transcription factor targets is coordinated by evolutionarily conserved modular intracellular signaling kinase cascades (21). Mitogen-activated protein kinases (MAPKs), 1 which relay these signals, are proline-directed serine/threonine kinases and include extracellular signal-regulated kinases (ERKs), c-Jun N-terminal kinase, and p38 MAPKs (22). The modularity and specificity in these signal transduction cascades are coordinated by several mechanisms, including selective phosphorylation of downstream kinases (23), targeting by specific MAPK phosphatases, subcellular localization of the kinases (24), MAPK isoformselective targeting of specific transcription factors (25), and the interaction with scaffolding proteins that mediate the interactions between components of the MAPK module (26,27). The ERK/MAPK cascade is also regulated by the relative abundance of the caveolin-1 protein (28,29).
Caveolin-1 is an important component of caveolar membranes, invaginations of the plasma membrane thought to participate in vesicular trafficking and signal transduction events (30). Caveolins are most abundant in differentiated cells, and caveolin-1 levels have been shown to be reduced in fibroblasts transformed by oncogenic Ha-ras (G12V) or v-abl (31) and in mammary adenocarcinoma cells induced by overexpression of ErbB2 (32). Furthermore, caveolin-1 was identified as one of 26 genes whose mRNA was down-regulated in human breast cancer cell lines (33). Overexpression of caveolin-1 in v-abl-and Ha-ras-transformed NIH-3T3 cells abrogated their anchorageindependent growth (34), and transfection with antisense caveolin-1 was sufficient to induce cellular transformation and ERK activity (28). Despite these studies, the molecular mechanisms by which caveolin-1 regulates cellular transformation are largely unknown.
In this study, we assessed whether caveolin-1 can directly regulate cyclin D1 expression. We show that the cyclin D1 gene is inhibited during overexpression of caveolin-1 as a result of repression of the cyclin D1 promoter and that the DNA sequences required contain the T-cell factor (TCF)/lymphoid enhancer factor-1 (LEF-1)-binding site. We conclude that repression of the cyclin D1 gene by caveolin-1 may contribute to the inhibition of cellular transformation.
Construction of Reporter and Expression Vectors-The human cyclin D1 promoter-reporter constructions (19,36,39,40) and the c-jun promoter-luciferase reporter from Ϫ225 to ϩ150 (41) were previously described. The reporter c-fosLUC (35) contains the human c-fos promoter from Ϫ361 to ϩ157 in the pA 3 LUC reporter (42). The junB promoter was cloned by polymerase chain reaction using oligonucleotides to the published sequences (5Ј-GGT ACC CGC GAG CCG CCT CCT CCC and 3Ј-AAG CTT CCG GGC GGC CCA GGC GGT) and was subcloned into the pA 3 LUC reporter to create the junBLUC reporter. The pALUC reporter, which contains 7 kilobases of the human cyclin A promoter (19,36,39) and the cdc25ALUC reporter (43) were previously described. The serum response element from the c-fos promoter from Ϫ332 to Ϫ277 was linked to the minimal TATA region of the E4 promoter and cloned into the reporter pA 3 LUC.
In transient expression studies, cells were transfected using calcium phosphate precipitation; the medium was changed after 6 h; and luciferase activity was determined after another 24 h. The effect of an expression vector was compared with that of an equal amount of empty FIG. 1. Cyclin D1 protein levels are down-regulated by caveolin-1. A, Western blot analysis of mammary tumors derived from murine mammary tumor virus (MMTV) ras transgenic mice for cyclin D1 and caveolin-1 levels. The membranes were reprobed using Rho-GTPase guanine-nucleotide dissociation inhibitor (GDI) as an internal loading control. Note that when compared with normal mammary gland tissue, cyclin D1 levels were increased in each of the tumor samples, whereas the levels of caveolin-1 were barely detectable. B, NIH-3T3 cells were serum-starved (1% serum) and treated with 10% serum, 1% serum, 5 ng/ml FGF, 50 ng/ml PDGF, or both FGF and PDGF for 16 h. Western blotting was performed for cyclin D1, caveolin-1, and the internal control Rho-GTPase guanine-nucleotide dissociation inhibitor. C, cyclin D1 protein levels were assessed by Western blotting using lysates of NIH-3T3 cell lines stably overexpressing antisense caveolin-1 and a revertant cell line in which the expression of the antisense plasmid was lost. The same membranes were reprobed for caveolin-1, Rho-GTPase guanine-nucleotide dissociation inhibitor, and JunB.
vector. Luciferase content was measured during the initial 10 s of the reaction using an AutoLumat LB953 (EG&G Berthold), and the values are expressed in arbitrary light units (35). Statistical analyses were performed using the Mann-Whitney U test with significant differences established as p Ͻ 0.05.

The Cyclin D1 Gene Is Repressed by Caveolin-1-Previous
studies have demonstrated that cyclin D1 levels are increased (19) and that caveolin-1 levels are decreased in Ha-ras (G12V)transformed fibroblast cells (31). We have therefore examined the abundance of cyclin D1 and caveolin-1 in tumors derived from murine mammary tumor virus ras-transformed cells. Cyclin D1 levels were increased in each of the tumors examined (Fig. 1A) and were associated with reduced or undetectable caveolin-1 levels. In previous studies, we showed that cyclin D1 levels were increased in mammary tumors from murine mammary tumor virus src transgenic mice (36), whereas caveolin-1 levels were undetectable in these tumors (32). To examine the effect of serum and growth factors on caveolin-1 and cyclin D1 levels, NIH-3T3 cells were serum-starved or treated with serum, FGF, or PDGF for 16 h. Serum starvation was associated with a reduction in cyclin D1 levels and an increase in caveolin-1 abundance. The addition of serum, FGF (5 ng/ml), or PDGF (50 ng/ml) induced cyclin D1 levels and reduced caveolin-1 abundance (Fig. 1B).
To determine whether caveolin-1 overexpression can directly regulate cyclin D1 levels, cell lines stably overexpressing antisense caveolin-1 (28) were examined for the abundance of cyclin D1. Revertants of 3T3 cells that have lost antisense caveolin expression (28), similar to the parental NIH-3T3 cell line (data not shown), showed a 4-fold increase in caveolin-1 protein levels ( Fig. 1C) compared with the antisense caveolin-1-expressing clone. Cyclin D1 levels were increased by 60% in the antisense caveolin-1 stable cell line compared with the revertant (Fig. 1C). Caveolin-1 and cyclin D1 expression levels therefore appear to be inversely related in 3T3 cells and mammary tumor tissue. In contrast with the reduction in cyclin D1 protein levels by caveolin-1, the JunB protein was increased in association with increased caveolin-1 levels (Fig. 1C).
Caveolin-1 Inhibits Transcription of the Cyclin D1 Gene-To determine whether caveolin-1 overexpression can regulate the activity of the cyclin D1 gene promoter, transient expression studies were performed using a caveolin-1 expression plasmid and the empty expression vector (pCB7). The results summarized in Fig. 2B show that overexpression of caveolin-1 repressed the activity of the cyclin D1 promoter in a dose-dependent manner. In contrast, overexpression of caveolin-3 ( Fig. 2A, Cav-3) did not inhibit the activity of the cyclin D1 promoter (Fig. 2B). Cyclin D1 promoter-luciferase construct was repressed by 70% using a reporter/expression vector ratio of 4:1 (Fig. 2B).
Previous studies have shown that caveolin-1 can inhibit the function of the serum-responsive transcription factor Elk-1 in a heterologous luciferase reporter assay (32). We therefore examined the effect of caveolin-1 overexpression on the native c-fos gene promoter. Comparison was made with the effect of caveolin-1 on the cyclin D1 promoter. The data are shown as mean luciferase activity in Fig. 3A. In contrast with the Ϫ1745CD1LUC reporter, which was repressed by caveolin-1, the c-fos promoter was not significantly repressed using a reporter/expression vector ratio of 4:1. As recent studies identified a serum response element in the cdc25A gene that is distinct from the one in the c-fos gene (13), the activities of the cdc25A promoter and those of the immediate-early genes c-jun and junB were determined in cells transfected with caveolin-1. The results shown in Fig. 3C demonstrate that caveolin-1 inhibited the activity of the cdc25A promoter by 80% and that of the c-jun gene reporter by 70%. The effect of caveolin-1 on the cyclin D1 and c-fos promoters was not significantly changed by serum concentrations (data not shown). In contrast, the junB promoter was induced by caveolin-1 by 15-30-fold (Fig. 3D). Together, these studies suggest that caveolin-1 repression of cyclin D1 promoter activity is inhibited by a serum-independent mechanism.
The TCF Site in the Cyclin D1 Promoter Is Required for Full Repression by Caveolin-1-To examine the DNA sequences in the cyclin D1 promoter required for repression by caveolin-1, the promoter activities in a series of cyclin D1 promoter constructs containing truncations and point mutations were assayed. Repression of the cyclin D1 promoter by caveolin-1 was maintained when the sequences between Ϫ1745 and Ϫ163 base pairs of the promoter were deleted (Fig. 4A). Since the cAMP response element site of the cyclin D1 promoter was previously FIG. 2. Caveolin-1 repression of the cyclin D1 promoter. A, schematic representation of the expression vectors encoding caveolin-1 and caveolin-3. B, the Ϫ1745CD1LUC reporter was transfected with the indicated amounts of the caveolin-1 or caveolin-3 expression vector into CHO cells. The luciferase (LUC) activity (relative light units) compared with the activity induced by equal amounts of control vector cassette was set as 100%. The data are means Ϯ S.E. of nine separate experiments. Note the inhibition of the cyclin D1 promoter activity by caveolin-1, but not by caveolin-3.vector. Luciferase content was measured during the initial 10 s of the reaction using an AutoLumat LB953 (EG&G Berthold), and the values are expressed in arbitrary light units (35). Statistical analyses were performed using the Mann-Whitney U test with significant differences established as p Ͻ 0.05.
shown to convey serum responsiveness in fibroblasts (18) and the AP-1 site is involved in mitogenic responses to angiotensin II and Ras (19,35), we used point mutants in these sites and found that the inhibitory effect of caveolin-1 on the cyclin D1 promoter was not affected in these mutants (Fig. 4, B and C). Within the proximal Ϫ163 base pairs that are still responsive to caveolin-1 overexpression, a binding site for the ␤-catenin⅐ TCF complex was recently identified (40) (Fig. 4A). Mutation of the ␤-catenin/TCF element (Ϫ163FOP) reduced the ability of caveolin-1 to inhibit the cyclin D1 promoter from 80 to Ͻ50% (Fig. 4D). Additional experiments were conducted comparing the effect of caveolin-1 with equal amounts of empty expression vector cassette (pCB7) at a reporter/expression vector ratio of 1:4 (n ϭ 12). When normalized as paired experiments with the effect of the expression vector normalized to 100%, further experiments confirmed the trend of reduced repression by mutation of the TCF site (Fig. 4D). These findings suggest that the TCF site is required for full repression of the cyclin D1 promoter by caveolin-1 and that additional elements may contribute to full repression. The TCF/LEF sequence in the cyclin D1 promoter is identical to the consensus TCF/LEF-1 site (40) and was sufficient for repression by caveolin-1 when it was linked to a minimal promoter (Fig. 4E).
The Caveolin-1 N Terminus Is Required for Repression of the Cyclin D1 Gene-To determine the domains in caveolin-1 that are required for inhibition of the cyclin D1 promoter, various caveolin-1 mutants were assayed for their ability to inhibit a full-length cyclin D1 promoter-luciferase construct. These mu-tants have previously been shown to be expressed at equivalent levels to the wild-type caveolin-1 in cultured cells (37,45,48,49). Cav-1␤-(32-178) repressed the cyclin D1 promoter to a similar extent as the full-length ␣-isoform (residues 1-178) (Fig. 5A). Deletion of the caveolin-1 carboxyl terminus did not affect repression. In contrast, deletion of the N-terminal 95 residues (positions 96 -178) not only abolished repression, but caused a modest induction of the cyclin D1 promoter. The abundance of the transfected ⌬C and ⌬N mutants was identical by Western blotting of cultured cells (48), suggesting that loss of expression is not responsible for the failure of the ⌬N mutant to repress the cyclin D1 promoter. Since deletion of the Nterminal residues 61-101 prevents caveolin-1 oligomerization in vivo (48) and this domain is sufficient as a glutathione S-transferase fusion for multimerization in vitro (44), we therefore determined whether oligomerization of caveolin-1 was required for repression of cyclin D1 promoter activity. Using the caveolin-1 N-terminal mutant Cav-1-(⌬61-101), we found that this mutant was capable of repressing cyclin D1 promoter activity to a similar extent as the ␣-isoform (Fig. 5A). The C-terminal half of the oligomerization domain of caveolin binds to and regulates the activity of several signaling molecules in the Ras/ERK pathway (29,32,50). To determine further the possibility of whether ERK signaling is involved in the repression of the cyclin D1 promoter, we used a mutant caveolin-1 that is completely defective in inhibiting p42/p44 MAPK signaling (Cav-1-(1-81) (45)) and found that repression of the cyclin D1 promoter activity by this mutant was minimally affected (Fig. 5A). These results suggest that caveolin-1 oligomerization and inhibition of the ERK signaling pathway are not required for repression of the cyclin D1 promoter activity and that the region responsible for this activity resides between amino acids 32 and 60; thus, a unique domain is required for full repression of cyclin D1. Interestingly, this region is not conserved among the various caveolins (Fig. 5B). DISCUSSION The cyclin D1 gene encodes the regulatory subunit of the holoenzyme that phosphorylates and inactivates the pRB protein, thereby promoting entry into the DNA synthetic phase of the cell cycle (2). Antisense cyclin D1 inhibits S phase entry induced by serum, growth factors, or steroids and inhibits transformation by Ha-ras, src, and neu (3,51,52). Caveolin-1 levels are reduced in a variety of tumor types (32), whereas increasing the level of caveolin-1 levels can suppress the transformed phenotype (28). In the present study, we found that cyclin D1 protein abundance and promoter activity were inhibited by overexpression of caveolin-1 protein. Caveolin-1 also inhibited the activity of the cdc25A promoter. The Cdc25A phosphatase dephosphorylates inhibitory phosphorylation sites on cyclin-dependent kinases (12,53), and overexpression of Cdc25A enhances transformation by oncogenic ras (53). Taken together, these studies demonstrate that the transcriptional activity of two major components of the cell cycle regulatory apparatus that governs DNA synthesis and cell transformation may be regulated by caveolin-1.
Caveolin-interacting proteins include G-protein ␣-subunits, Ha-Ras, Src family tyrosine kinases, endothelial nitric-oxide synthase, epidermal growth factor receptor, and other related tyrosine kinases and protein kinase C isoforms (54). The caveolin-1 mutants used in the present study suggest that repression of the cyclin D1 promoter activity most probably involves different domains of the caveolin-1 molecule than those required Ϫ1745CREmt) or the AP-1 site (Ϫ963AP-1mt) of the cyclin D1 promoter were examined for inhibition of their activity by caveolin-1. D, the effect of caveolin-1 on a TCF/LEF site mutant in the Ϫ163 construct of the cyclin D1 promoter (Ϫ163FOP) was also analyzed. Note the reduced effect of caveolin-1 on the TCF/ LEF mutant (n ϭ 6). In the inset, a comparison is made with the effect of the pCB7 empty expression vector cassette established as 100% for n ϭ 12. E, the trimeric TCF/LEF site linked to a minimal promoter was analyzed for the effect of caveolin-1 on its activity. The amount of cotransfected caveolin-1 expression vector (150 or 300 ng) used with the TCF/ LEF luciferase reporter (2.4 g) is indicated. The effect of the empty expression vector cassette pCB7 was normalized to 100%. Note that this minimal heterologous construct ((TCF/LEF) 3 LUC) is sufficient for repression by caveolin-1. The data are means Ϯ S.E. of seven separate transfections.
for regulation of epidermal growth factor receptor signaling and formation of caveola structures. Three distinct caveolin genes have so far been identified (caveolin-1, -2, and -3), which can form homo-or hetero-oligomers (54). Interestingly, the loss of only caveolin-1, but not the other family members, was observed in tumors; and selective reduction of caveolin-1 levels, without affecting caveolin-2, was sufficient to drive transformation of NIH-3T3 cells (28,30). The structural conservation is high among the three caveolar proteins, but divergence is displayed at the N terminus of these molecules (Fig. 5B) (30). In agreement with this observation, caveolin-1, but not caveolin-3, was found to repress the cyclin D1 promoter.
Caveolin oligomers directly bind cholesterol and interact with glycosphingolipids, enhancing the formation of the caveolar structures (55). A central hydrophobic domain (residues 102-134) forms a hairpin-like structure within the membrane, which positions both the N-and C-terminal domains of the molecule in the cytoplasm. Deletion of the C-terminal domain abrogates the interaction of homo-oligomers; this interaction contributes to the formation of the caveolin-rich scaffold (48). Similar to the effect observed by the ␣-isoform of caveolin-1, the C-terminal mutant (Cav-1⌬C) repressed the cyclin D1 promoter, suggesting that interaction of caveolin-1 homo-oligomers is apparently not required for cyclin D1 promoter inhibition. Deletion of the N-terminal 95 residues of the molecule abolished this repression of the cyclin D1 promoter by caveolin-1. The N terminus of caveolin-1 is involved in homo-oligomerization and interaction with the ERK signaling pathway (50), but its deletion (Cav-1-(⌬61-101)) did not affect the magnitude of cyclin D1 promoter inhibition, supporting the view that formation of caveolin-containing structures is not necessary for cyclin D1 promoter repression.
This study links, for the first time, the caveolin-1 protein with inhibition of the cell cycle regulatory apparatus involved in tumorigenesis. Cyclin D1 overexpression is known to induce mammary tumors in transgenic mice (56) and cooperates in oncogenic transformation with several oncogenes, including ras, myc, and E1A (57)(58)(59). Cdc25A also cooperates in cell transformation with ras (53). Since cyclin D1 is frequently overexpressed in a variety of human tumors (3) and caveolin-1 abundance is reduced in many tumors (54), our studies point to the possibility that loss of caveolin-1 expression during tumorigenesis may lead to cellular proliferation through induction of the cyclin D1 gene. The cyclin D1 gene is induced by several signaling pathways implicated in cellular transformation, including the phosphatidylinositol 3-kinase, ␤-catenin/TCF/LEF, ERK, and nuclear factor-B signaling pathways (40,60,61). Examining the effect of caveolin-1 mutants on cyclin D1 promoter activity, we found that repression of the cyclin D1 promoter apparently does not involve the ERK pathway since an N-terminal caveolin-1 mutant incapable of inhibiting signaling by the Ras/ERK pathway (45) could still repress cyclin D1 promoter activity. Thus, although the ERK pathway can activate cyclin D1 (19, 35), a different pathway is most probably affected by caveolin-1.
A mutation in the TCF/LEF site of the cyclin D1 promoter FIG. 5. The N terminus of caveolin-1 is required for repression of the cyclin D1 promoter. A, the caveolin-1 expression plasmids, shown schematically on the left, were cotransfected into CHO cells with the cyclin D1 promoter (Ϫ1745CD1LUC), and the effect on the cyclin D1 promoter activity was determined. The effect of the caveolin-1 constructs was compared with that of the empty expression vector (pCB7). Note that the repression of the activity of the cyclin D1 promoter (4-fold) was not only abolished by deletion of the N terminus (⌬N) of caveolin-1, but resulted in modest induction. The percent repression is presented as the means Ϯ S.E. of eight separate transfections. B, the conserved regions of caveolin-1, 2, and -3 are shown. Residues 31-60 (indicated by an arrow), which are poorly conserved among the caveolins, are implicated in cyclin D1 repression. OD, oligomerization domain; TM, transmembrane domain; ␤-iso., ␤-isoform of caveolin-1.
that abolished binding to TCF proteins in electrophoretic mobility shift assays (40) reduced repression by caveolin-1. Interestingly, we found that caveolin-1 also inhibited the activity of the c-jun promoter, a gene that is also activated by ␤-catenin/ TCF signaling and that contains a TCF site in its promoter (62). In contrast, the c-fos promoter, which is induced by ERK, was not significantly repressed by caveolin-1, further supporting the view that caveolin-1 repression of promoters in mammary epithelial cells involves a pathway that is distinct from the ERK pathway. Future studies will have to address the molecular mechanisms involved in the role of caveolin-1 in the regulation of the ␤-catenin/TCF/LEF signaling pathway.