The role of cadherin, beta-catenin, and AP-1 in retinoid-regulated carcinoma cell differentiation and proliferation.

Vitamin A derivatives (retinoids) are potent regulators of cell proliferation and differentiation. Retinoids inhibit the function of the oncogenic AP-1 and beta-catenin/TCF pathways and also stabilize components of the adherens junction, a tumor suppressor complex. When treated with retinoic acid (RA), the breast cancer cell line, SKBR3, undergoes differentiation and reduction in cell proliferation. The present work demonstrates that in SKBR3 cells, which exhibit high AP-1 activity, RA-regulation of cadherin expression and function, but not changes in AP-1 (or beta-catenin/TCF) signaling, is responsible for the epithelial differentiation. However, cadherin function and recruitment of beta-catenin to the membrane is not required for RA to regulate DNA synthesis in these cells. RA also reduces the activity of an AP-1 and TCF-sensitive cyclin D1 reporter in SKBR3 cells in a manner that is independent of the TCF site. In contrast, in SW480 cells, which have high levels of beta-catenin/TCF signaling, the activity and retinoid responsiveness of the cyclin D1 promoter was markedly inhibited by mutation of the TCF site. These data indicate that the remarkably broad effects of RA on the growth and differentiation of many different epithelial cancers may well be explained by the ability of RA to differentially regulate the activity of RAR/RXR, AP-1, and beta-catenin/TCF pathways.

inhibit the growth of a number of breast cancer cell lines (10 -12) and can reduce tumor incidence, average number of tumors, and average tumor burden in a rat breast cancer model (10). In some instances the effects of RA on cell growth have been attributed to the ability of RA to down-regulate AP-1 activity (1,13,14). However, it is not clear if the effects of retinoic acid on cell differentiation can be separated from its effects on cell growth. We and others have previously demonstrated that the effects of retinoids may involve modulation of adherens junction structure and function (11,(15)(16)(17).
The adherens junction is a molecular complex that is essential for initiating and maintaining strong cell-cell adhesion in epithelial cells (18). The basic components of the adherens junction include a trans-membrane cadherin molecule, the cytoplasmic catenins, and the actin cytoskeleton (19). Cadherins are calciumdependent cell adhesion molecules that are involved in the organization of the developing embryo, and are essential for the maintenance of tissue integrity in the adult (20). In epithelial cells, loss of function or expression of E-cadherin is correlated with the progression of tumors to a more invasive phenotype (21). Moreover, expression of other cadherins can inhibit invasion and thus compensate for a loss of E-cadherin (22)(23)(24). Loss of cadherin function may also be mediated in other ways. For example, loss of ␣or ␤-catenin protein expression can disrupt normal cell-cell adhesion (25,26). Alternatively, cadherin function can be modulated by tyrosine kinase activity (27). For example, the v-src oncoprotein phosphorylates tyrosine residues on ␤-catenin and cadherins and disrupts the adherens junction, resulting in a shift to a fibroblastoid phenotype that exhibits increased invasiveness (28,29). Finally, loss of cadherin function may be mediated by activity of the transcription factor complex, AP-1. AP-1 is made up of the proto-oncogenes jun and fos, and its activity is associated with cell proliferation and neoplastic transformation (30). Activation of c-jun in mammary epithelial cells resulted in a loss of epithelial polarity, a disruption of intercellular junctions and normal barrier function, and the formation of irregular multilayers. These morphological changes were accompanied by a reduction in the association between E-cadherin and ␤-catenin (17).
In the breast cancer cell line, SKBR3, RA reduces cell proliferation and induces a cell differentiation (15). We now show that modulation of AP-1 or ␤-catenin/TCF signaling is not involved in RA-induced cell differentiation. However, cadherin expression and function are necessary and sufficient to mediate the effects of RA on adhesion and differentiation but are not required for RA-mediated inhibition of cell proliferation.

EXPERIMENTAL PROCEDURES
Cell Lines and Treatments-SKBR3 and SW480 cells were obtained from ATCC and grown in Dulbecco's modified Eagle's medium (Invitrogen) plus 10% fetal bovine serum as described previously (31). Cells were treated with 1 M 9-cis-RA or ethanol for 48 h. Several experi-ments were repeated using all trans-retinoic acid with similar results. TPA (Sigma) was used at 100 nM overnight.
Vectors-Vectors encoding either wild type or a degradation-resistant mutant of ␤-catenin (both hemagglutinin-and FLAG-tagged) have been described previously (32). A wild type human E-cadherin vector was kindly provided by Barry Gumbiner, (Memorial Sloan-Kettering Cancer Center, New York). The E-cadherin reporter construct was kindly provided by Antonio Herreros, Barcelona, Spain. Dominantnegative c-jun, wild type, and mutant AP-1 luciferase constructs were described previously (33,34). Cyclin D1 wild type and mutant reporter constructs were provided by Richard Pestell, New York. A FLAG-tagged dominant-negative mutant of c-jun was constructed using PCR. Based on the previously characterized TAM-67 mutant of c-jun, we synthesized a c-jun deletion mutant that had the same amino acid deletion as TAM-67. c-Jun (amino acids 3-122 deleted), missing the transactivation domain along with the FLAG tag, was amplified by PCR and cloned into a pcDNA3.1 expression vector. This construct, pcDNA3.1-FLAG-TAM-67, was sequenced to confirm the correct nucleotide sequence of the FLAG-TAM67 gene. Expression of this protein in MCF7 cells inhibits AP-1 activity as well as TAM-67 (without the FLAG tag). A human c-jun expression vector was kindly provided by Bart van der Burg, Hubrecht Laboratory, The Netherlands. A GFP expression vector was used (pEGFP, CLONTECH). pTOPFLASH and pFOPFLASH were provided by Marc Van de Wettering (35).
Subcellular Fractionation-Cells from confluent 10-cm dishes were isolated and subjected to sequential Dounce homogenization in a hypotonic solution (10 mM Tris, 0.2 mM MgCl 2 , pH 7.5) (32). This procedure results in cell lysis and the isolation of sheets of plasma membrane rather than vesicles (32). The homogenate was spun first for 10 min at 3000 ϫ g to remove nuclei. The supernatant was then ultracentrifuged at 150,000 ϫ g for 1 h. The supernatant, defined as the cytoplasmic fraction, was added to 4 volumes of ethanol, and the proteins were precipitated overnight. The proteins were then reisolated by ultracentrifugation and solubilized in sample buffer (2% SDS, 60 mM Tris, pH 6.8, 10% glycerol). The pellet from the original ultracentrifugation was solubilized in a 1% Nonidet P-40 buffer (1% Nonidet P-40, 150 mM NaCl, 50 mM Tris, pH 8.0) for 30 min, then reclarified in a microcentrifuge for 15 min to remove insoluble material and cell debris. The resulting supernatant, the Nonidet P-40-soluble fraction, was then added to sample buffer.
Western Blotting-Bio-Rad DC Protein Assay kit was used to measure protein content in the samples. 25 g of protein were separated on an 8% reducing polyacrylamide mini-gel (Novex), transferred onto nitrocellulose (Protran), and blocked overnight in 5% skim milk. The blot was then probed with an appropriate antibody followed by a secondary peroxidase-labeled antibody (Kirkegard and Perry), and the bands were visualized by enhanced chemiluminescence (Amersham Biosciences). The blots were then stripped at 50°C for 30 min (stripping solution: 62.5 mM Tris, pH 7.5, 2% SDS, 1.7% (v/v) ␤-mercaptoethanol), washed two times in phosphate-buffered saline, and blocked in 5% milk prior to reprobing.
Immunocytochemistry-Cells were grown on 12-mm coverslips. Most coverslips were fixed in 100% ice-cold methanol for 3 min at Ϫ20°C. To retain green fluorescent protein staining, cells were fixed in 2% paraformaldehyde for 30 min followed by 0.5% Triton X-100 for 5 min. Coverslips were then blocked in 3% ovalbumin for 30 min. For double staining, the first primary antibody was bound overnight at 4°C, followed by a fluorescein isothiocyanate-conjugated secondary antibody for 1 h at room temperature. Next, the second primary antibody was bound at room temperature for 1 h followed by a Texas Red-conjugated pre-absorbed secondary antibody for 1 h at room temperature. Cells were mounted (Vectashield) and visualized on a Zeiss microscope. There was no cross-over between fluorescein and Texas Red channels.
For pan-cadherin staining antigen retrieval was performed. Following fixation, coverslips were placed in 1 liter of 0.01 M citrate buffer, pH 6.0 (18 ml of 0.1 M citric acid, 82 ml of 0.1 M sodium citrate, pH 6.0, 900 ml of distilled H 2 O). The coverslips were microwaved for 30 min on high and then allowed to cool slowly before blocking.
GFP Sorting and Immunocytochemistry of Transfected Cells-1 ϫ 10 6 cells were plated in a 10-cm dish. The cells were transfected with 10 g of a green fluorescent protein expression vector, pEGFP (CLON-TECH) and 50 g of the plasmid of interest. Cells were transfected for 6 h by the calcium phosphate method (36) then shocked with media containing 20% glycerol for 4 min. Cells were then washed 3ϫ in phosphate-buffered saline and incubated for 24 h. Cells were then FIG. 1. Transient overexpression of ␤-catenin is not sufficient to induce a morphologic change. SKBR3 cells were transfected with vector alone (A and B) or wild type ␤-catenin (C). In B, the cells were also treated with RA for 48 h prior to fixation. The cells were fixed and stained for ␤-catenin. RA dramatically increased the expression of ␤-catenin at sites of cell-cell contact in vector-treated cells (compare A to B). However, despite the high level of ␤-catenin expression there was no change in cell morphology, and the exogenous ␤-catenin did not concentrate at cell-cell contact sites (C). SKBR3 cells were transfected with vector (pCDNA3) or with wild type ␤-catenin, and Western blotting was performed using whole cell lysates (D). trypsinized and sorted by FACS, isolating cells expressing high levels of GFP. Cells were replated on 12-mm coverslips, then treated with RA or vehicle for 48 h, fixed, and stained as described above.
Luciferase Reporter Assays-Cells were seeded in 12-well plates at 1 ϫ 10 5 cells per well. Cells were transiently transfected using the calcium phosphate method (36). For luciferase assays, cells were transfected with 1 g of either the wild type or mutant E-cadherin-luciferase construct, AP-1 luciferase construct, CD1-luciferase construct or with the LEF-reporter pTOPFLASH/pFOPFLASH (35) along with 0.02 g of pCMV-Renilla luciferase (Promega) (32). RA treatment was initiated 24 h posttransfection. Luciferase activity was monitored using the DUAL-luciferase Assay System (Promega). The experimental reporter activity was controlled for transfection efficiency by comparison with the constitutively expressed Renilla luciferase.

␤-Catenin
Overexpression Is Not Sufficient to Mimic the Effects of RA on Cell Morphology and Differentiation-RA has a profound effect on cell-cell adhesion and cell differentiation in a number of breast cancer cell lines, effects that are accompanied by an increase in membrane-associated ␤-catenin (10,15). We also showed that RA treatment stabilizes ␤-catenin protein levels without affecting the steady state levels of its mRNA. To determine whether overexpression of ␤-catenin is sufficient to mimic the effects of RA on cell morphology and differentiation, SKBR3 cells were transfected with wild type ␤-catenin and its S37A-stable mutant form. Immunocytochemistry was performed to detect ␤-catenin protein. Fig. 1 shows that the morphology of the cells transfected with ␤-catenin was similar to that of untreated control cells despite the high expression of ␤-catenin, most of which was detected in the cytoplasm. Similar results were obtained in cells transfected with S37A ␤-catenin (not shown). However, upon RA treatment a dramatic change in cell morphology was observed in untransfected and ␤-catenin-transfected cells. Following RA treatment most of ␤-catenin was translocated to the cell membrane. These results showed that increased levels of ␤-catenin alone are not responsible for RA-induced changes in cell morphology and differentiation. However, it is possible that translocation of ␤-catenin to the membrane is mandatory for RA-mediated effects on cell morphology. To test this we examined if cadherin-mediated translocation of ␤-catenin to the membrane could mimic the effects of RA on cell morphology.
Cadherin Expression and RA-induced Cell Differentiation-To test if exogenous expression of E-cadherin is sufficient FIG. 2. SKBR3 and MCF-7 cells were transfected with human E-cadherin (؊173 to ؉93)-luc promoter construct, and luciferase reporter activity was measured. In both cell lines, RA increased the reporter activity significantly (A). E-cadherin expression mimics the effects of RA on cell morphology. SKBR3 cells were transfected with vector (B and C), human E-cadherin construct, hecD pCDNA3 (D), or with wild type ␤-catenin (E). Cells were also co-transfected with a green fluorescent protein expression vector and sorted by FACS to isolate the transfected cells. Cells were replated and grown in the absence (B) and in the presence (C) of RA for 48 h and fixed. Cell morphology was determined by phase contrast microscopy. RA had a profound effect on cell morphology compared with untreated cells (B). E-cadherin expression mimics the RA-induced changes in cell morphology (D), however ␤-catenin expression does not (E). to mimic RA-induced effects, SKBR3 cells were transfected with an E-cadherin construct, hecD pCDNA3, along with a green fluorescent protein expression vector (pEGFP), and green cells were separated by FACS, replated, and grown with or without RA for 48 h. As expected RA markedly affected cell morphology (Fig. 2C, compare with untreated control in Fig.  2B). Similar effects were observed in the cells transfected with wild type E-cadherin expression vector (Fig. 2D) but not in the cells overexpressing ␤-catenin (Fig. 2E). E-cadherin expression also increases the translocation of ␤-catenin protein to the membrane as revealed by immunocytochemistry (Fig. 3B). Western blot analysis of whole cell lysates of cells overexpressing full-length E-cadherin revealed an increased level of ␤-catenin protein (Fig. 3C). Note that the increase in ␤-catenin expression is less than that induced by RA because these experiments were performed as transient transfection assays.
To determine whether RA can directly regulate cadherin ex-pression, SKBR3 and MCF-7 cells were transfected with an E-cadherin-luciferase promoter construct (37) and Renilla luciferase as a transfection efficiency control. After transfection, cells were grown in the presence or absence of RA and harvested after 48 h to measure reporter activity. A significant increase in the E-cadherin reporter activity was detected in the presence of RA (Fig. 2A). These results indicate that RA can increase the activity of an E-cadherin promoter and that exogenous expression of E-cadherin is sufficient to mimic RA-induced changes in cell morphology and recruitment of ␤-catenin to the cell membrane. We next examined if RA influenced endogenous cadherin expression.
The rounded and poorly adhesive phenotype of untreated SKBR3 cells is consistent with diminished cadherin expression or function and we showed several years ago that these cells do not express E-cadherin (31). Consistent with this Pierceall et al. showed that these cells have a homozygous deletion of the E-cadherin gene (38). Nevertheless the calcium dependence of RA-induced morphological changes, the recruitment of ␤-catenin to the membrane, as well as the effects of RA on cadherin promoter activity strongly point to a role for an endogenous cadherin in mediating the effects of RA on SKBR3 cells.
To test RA effects on endogenous cadherin expression in SKBR3 cells, we used a polyclonal pan-cadherin antibody raised against a region of the highly conserved N-cadherin C-terminal, to detect cadherin protein. This antibody recognizes most cadherins except notably E-cadherin (39,40). As presented in Fig. 4, A and B, RA treatment dramatically increased cadherin levels at cell-cell contact sites. Western blot analysis showed a dramatic RA-induced increase in an immunoreactive band at ϳ120 kDa, consistent with the molecular mass of most type I and II cadherins (Fig. 4C). A variety of approaches have shown that the RA-induced cadherin in SKBR3 cells is not E-, N-, P-, or LI-cadherin, cadherin-6 or-11 (results not shown). However, our attempts to identify this cadherin by classical methods were unsuccessful. This suggests that the unidentified cadherin may be a novel cadherin that is regulated directly or indirectly by RA. Taken together our results indicate that RA increases endogenous cadherin expression.
Inhibition of AP-1 Activity Is Neither Necessary nor Sufficient to Mimic the Effects of RA on Cell Morphology-High AP-1 activity can affect differentiation and changes in cadherin and catenin expression (17). As it is well known that RA can inhibit AP-1 activity we next investigated the role of AP-1 in RAmediated cell differentiation. To determine the role of AP-1 in RA-induced changes in SKBR3 cells, reporter assays were performed using an AP-1 luciferase reporter construct. A mutant construct with a three-nucleotide mutation in the AP-1 site was used as a negative control. We confirmed that RA reduced AP-1 reporter activity in SKBR3 cells in a dose-dependent manner (Fig. 5A). As a control dominant-negative c-jun, TAM-67 also significantly reduces AP-1 reporter activity (Fig. 5A). To determine whether AP-1 function can inhibit the effects of RA on cell morphology and ␤-catenin protein levels, SKBR3 cells were transfected with TAM-67 or with a c-jun expression vector (41). Neither TAM-67 nor c-jun altered ␤-catenin proteins levels in cell lysates (Fig. 5, B-D). Similarly TAM-67 could not induce any changes in cell morphology, ␤-catenin levels, or localization as detected by immunocytochemistry (Fig. 6, A and B). Likewise, overexpression of c-jun (to increase AP-1 activity) was also not sufficient to inhibit the function of RA in altering cell morphology or ␤-catenin expression (Fig. 6, C and D). These results indicate that AP-1 activity is not required for RA-induced morphological changes and differentiation in SKBR3 cells.
Physiological Calcium Levels Are Necessary for RA Treatment to Change Subcellular Distribution of ␤-Catenin-The mechanism whereby RA increases the membrane pool of ␤-catenin is most likely the result of increased cadherin expression. This study and others have shown that exogenous expression of cadherins can recruit cytoplasmic ␤-catenin to the membrane pool, an effect that depends upon the ability of the cadherin cytoplasmic tail to bind ␤-catenin (42,43). We next wanted to test if RA treatment could reduce cytoplasmic levels of exogenously expressed ␤-catenin in SKBR3 cells. We performed three-pool subcellular fractionation to isolate cytosolic, an Nonidet P-40-soluble plasma membrane fraction and an Nonidet P-40-insoluble fraction. Fig. 7A shows that cytoplasmic levels of ␤-catenin are markedly elevated following transient transfection. RA-treatment significantly reduces the level of cytoplasmic ␤-catenin with a corresponding increase in membrane fraction. RA also decreased the cytoplasmic level of the degradation-resistant S37A mutant form of ␤-catenin (data not shown). If the ability of RA to reduce cytoplasmic ␤-catenin depended upon cadherin function we would expect that treatment of cells with low calcium medium should reverse these effects of RA. Fig. 7B shows that exposure of cells to low calcium medium prevents the RA-mediated decrease in the level of cytoplasmic ␤-catenin in transiently transfected cells. These results strongly suggest that the ability of RA to reduce cytoplasmic ␤-catenin depends on cadherin expression.
␤-Catenin protein levels and its subcellular distribution are tightly regulated. ␤-Catenin is present in two pools: a membrane pool, required for cell-cell adhesion; and cytoplasmic/ nuclear pool, responsible for ␤-catenin/TCF signaling. Several studies have shown that translocation of cytoplasmic ␤-catenin to the membrane can reduce ␤-catenin/TCF signaling (44 -46). However, our results showed that RA-treatment inhibits ␤-catenin/TCF reporter activity even in the low calcium medium (Fig. 7C), which is consistent with our demonstration that RAR is able to bind directly to ␤-catenin and inhibit ␤-catenin/ TCF signaling in the presence of RA (47). We next investigated whether the RA effects on cell proliferation can be separated from its effects on cell differentiation.
Cadherin Expression and Function Is Not Necessary for RAmediated Effects on Cell Proliferation-RA has a well documented inhibitory effect on cell proliferation. We showed previously that DNA synthesis is markedly inhibited by RA in SKBR3 cells (15). E-Cadherin has been implicated as a tumor/ invasion suppressor, although there is little evidence that it FIG. 6. Inhibition of AP-1 activity is neither necessary nor sufficient to mimic the effects of RA on cell morphology and ␤-catenin expression. A-D, SKBR3 cells were transfected with a FLAG-tagged TAM-67 (A and B) then stained with an anti-FLAG antibody (A), or ␤-catenin antibody (B). Expression of TAM-67 did not influence the expression or distribution of ␤-catenin. C and D, SKBR3 cells were transfected with c-jun and GFP. The jun plasmid is not epitopetagged, and to detect jun-transfected cells we co-transfected GFP. GFP was detected by immunofluorescence (C). ␤-Catenin was detected in the same cells by an anti ␤-catenin monoclonal antibody (D). Exogenous expression of c-jun and GFP or GFP alone (not shown) did not inhibit the effects of RA on cell morphology or ␤-catenin expression. directly affects cell proliferation (48,49). Thus we wanted to determine whether RA effects on DNA synthesis could be mimicked by exogenous expression of E-cadherin or ␤-catenin. Fig.  8A shows that in control cells after 48 h of RA treatment, DNA synthesis, as measured by tritiated thymidine uptake, was reduced by ϳ50%. E-Cadherin transfection alone did not significantly reduce DNA synthesis in this time period but, as expected, did exert a marked morphological transformation (see Fig. 4). Moreover, DNA synthesis in E-cadherin-transfected cells was still reduced by RA. In this experiment, FACSsorted E-cadherin-transfected cells were used. These experiments show that cadherin expression does not mediate the growth inhibitory effects of RA. However, it is possible that calcium-dependent adhesion is still required for RA to inhibit cell proliferation. To test this we grew cells in physiological (2 mM) and low (50 M) calcium in the absence or in the presence of RA (Fig. 8B). RA treatment inhibited tritiated thymidine uptake under both conditions even though the morphological effects of RA were completely reversed in cells growing in 50 M calcium (data not shown but see Ref. 15). These results show that RA-mediated inhibition of cell proliferation is independent of the cadherin/␤-catenin and calcium-dependent adhesion pathways in SKBR3 cells. To further investigate RA-induced inhibition of cell proliferation we tested the effects of RA on cyclin D1 expression.
Regulation of Cyclin D1 Promoter Activity by RA-RA inhibits cell proliferation by influencing progression from G 1 to S-phase, which in turn depends on changes in cyclin D1 expression (CD1). Cyclin D1 expression and the activity of CD1 reporters are under complex regulation and are markedly influenced by AP-1 and ␤-catenin/TCF signaling pathways (50). To test if RA inhibition of ␤-catenin signaling plays a role in the regulation of CD1 expression, SKBR3 cells were transfected either with wild type or TCF site-mutated CD1-luciferase constructs (50). In SKBR3 cells, RA treatment significantly inhibits the activity of both wild type as well as TCF-deleted CD1 reporter activity showing that, in these cells, CD1-reporter activity is not predominantly regulated by ␤-catenin/TCF signaling (Fig. 8C). This is not surprising because SKBR3 cells have high AP-1 and low ␤-catenin/TCF activity (47). Therefore, the effects of RA on cell proliferation and CD1 expression in SKBR3 cells are probably mediated by inhibition of AP-1 not ␤-catenin/TCF activity. Nevertheless, it is possible that, CD1 activity in cells, which have high ␤-catenin/TCF signaling, is predominantly regulated by this pathway. To test this we used APC (adenomatous polyposis coli) mutant SW480 colorectal cancer cells, which have very high ␤-catenin/TCF activity but low AP-1 activity. In SW480 cells RA significantly decreased wild type CD1 reporter activity but had no effect on the ⌬TCF CD1 reporter (Fig. 8C). These results indicate that, unlike SKBR3 cells, RA repression of CD1 activity in SW480 cells depends on ␤-catenin/TCF signaling.
Taken together, our data strongly suggest that RA-mediated decreases in CD1-activity and DNA synthesis in SKBR3 breast cancer cells are due to changes in AP-1 and/or ␤-catenin/TCF activity that directly inhibit cell proliferation (47). In contrast, the effects of RA on cell morphology and differentiation are independent of AP-1 or ␤-catenin/TCF signaling. Therefore, it can be concluded that RA differentially regulates ␤-catenin/ cadherin and ␤-catenin/TCF and AP-1 pathways.

DISCUSSION
Retinoids are potent regulators of cell proliferation and cell differentiation. The actions of RA are mediated through its receptors (RAR and RXR), which upon ligand binding affect at least three important transcriptional activation pathways. RAactivated RAR/RXR heterodimers bind to the RARE-containing promoter elements of many genes to regulate transcription (51). RA-activated RARs can also inhibit the action of the Fos/Jun AP-1 complex thereby blocking the transcription of AP-1-regulated genes (52). Finally, we have demonstrated that RA-activated RAR can inhibit ␤-catenin/TCF transactivation by directly binding to ␤-catenin (47). The pleiotropic actions of RA are probably a result of the differential regulation of these three pathways. However, it is not clear which of these pathways accounts for the effects of RA on cell differentiation and proliferation and whether the RA effects on cell proliferation can be separated from the effects on cell differentiation. Using the breast cancer cell SKBR3 as a model system we now show that these actions of RA can be separated.
RA Effects on SKBR3 Cell Differentiation Are Not Regulated by AP-1 or ␤-Catenin/TCF Pathways but Do Require Calcium- FIG. 8. Calcium-dependent adhesion is not necessary to mediate the effects of RA on DNA-synthesis. A, SKBR3 cells, transfected with the indicated plasmids and GFP, were sorted by FACS and grown with or without RA for 48 h. Tritiated thymidine uptake was measured as described (15). Treatment of cells with hydroxyurea completely prevented tritiated thymidine uptake (Neg. Control). All values are normalized to vector alone control, which was considered as 100%. B, SKBR3 cells were grown for 48 h in high (2 mM) or low (50 M) calcium levels, and tritiated thymidine uptake was measured. All values are plotted relative to untreated high calcium values. Exogenous expression of E-cadherin did not significantly reduce DNA synthesis, and reduced calcium medium did not reverse the effects of RA. C, SKBR3 and SW480 cells were transfected with wild type (WT) and TCF binding site-deleted cyclin D1 (Ϫ163)-luciferase constructs. After transfection, cells were treated with RA for 48 h then harvested, and luciferase reporter assay was performed. All values are expressed as relative luciferase units. Results from the present study suggest that RA affects SKBR3 cell differentiation by increasing cadherin expression indirectly through activation of the RARE pathway. This pathway is not required for inhibition of cell proliferation. Because SKBR3 cells have very low levels of ␤-catenin and are known to be growth inhibited by AP-1 blockade, we hypothesize that the growth inhibitory actions of RA in SKBR3 cells are mediated via inhibition of AP-1. In other cells, which have low AP-1 and high levels of ␤-catenin signaling RA may regulate cell proliferation by inhibition of the ␤-catenin/TCF pathway. Although increased cadherin expression can decrease cytoplasmic ␤-catenin levels this mechanism is not absolutely required for RA to inhibit ␤-catenin/TCF signaling. dependent Cell Adhesion and Cadherin Function-In earlier studies we showed that RA treatment induces differentiation and stabilizes the levels of ␤-catenin in SKBR3 cells (15). In SKBR3 cells, very little ␤-catenin signaling is detected even though these cells do not express E-cadherin and levels of TCF family members are comparable with other epithelial cells. Overexpression of ␤-catenin can increase ␤-catenin signaling indicating that this pathway can be activated in these cells (47). These observations indicate that SKBR3 cells might have adjusted their degradation machinery to remove free ␤-catenin very efficiently from the cytoplasm (53). In response to RA, ␤-catenin protein levels at the membrane increased significantly although mRNA levels remained constant (15). The translocation of ␤-catenin to the membrane could be due to either RA-mediated inhibition of ␤-catenin degradation or to RA-mediated increase in cadherin/s expression. If RA effects were solely due to ␤-catenin stabilization, we would expect that overexpression of ␤-catenin could mimic the effects of RA on cell morphology and differentiation. On the other hand, if RA effects are due to increased cadherin expression then over expression of E-cadherin should show similar effects to RA. Our experiments show that overexpression of cadherin but not ␤-catenin can mimic RA effects on cell morphology and differentiation in SKBR3 cells. However, it is also possible that both inhibition of degradation machinery as well as increased cadherin expression are involved and are working synergistically in RA-induced changes.
We also showed that RA-induced changes in cell morphology and differentiation are calcium-dependent, which confirms the requirement for cadherin. Because these cells do not express E-cadherin another cadherin must be involved in this process. Several unsuccessful attempts were made to identify the RAresponsive cadherin in SKBR3 cells. It should be noted that vitamin D-activated vitamin D receptor can also increase Ecadherin expression and promoter activity in colorectal cancer cells (54).
Promoter analysis indicates that there are no retinoic acid responsive elements in the 1-kb promoter of E-cadherin. Nonetheless our reporter assays showed that a 783-bp E-cadherin promoter responded to RA treatment. This paradox can be explained if we assume that RA effects on cadherin expression may not be direct. In several breast cancer cell lines including SKBR3, AP-1 activity is very high and might negatively regulate cadherin expression. Because RA treatment can inhibit AP-1 activity, it is possible that RA inhibition of AP-1 activity is responsible for increased expression of cadherin/s. However, our results showed that neither AP-1 nor ␤-catenin is involved in RA-mediated up-regulation of cadherin expression and changes in cell morphology and differentiation. Taken together, we conclude that RA can increase cadherin expression but that it is not directly mediated by regulation of ␤-catenin/TCF, RAR/RXR, or AP-1 activity. The most likely explanation for the effects of RA on cadherin expression is that RA regulates the activity of a cadherin-regulating gene that does have RARE elements in its promoter.
An abnormal activation of the wnt pathway occurs in several types of cancer, most notably colorectal cancer that is characterized by high ␤-catenin/TCF activity. In SKBR3 cells, ␤-catenin/TCF activity is very low so it is unlikely that this pathway is involved in the transformed phenotype of SKBR3 cells. However, the ability of RA to increase membrane ␤-catenin levels independently from its effects on ␤-catenin signaling suggests that the RA effects on cell proliferation and differentiation respectively are mediated by a different pathway. These results are consistent with our demonstration that RA-activated RAR is able to bind directly to ␤-catenin and inhibit ␤-catenin signaling.
Cadherin Expression and Function Is Not Necessary to Mediate the Effects of RA on SKBR3 Cell Proliferation-Misregulated cell proliferation is the hallmark of cancer. The rate of DNA synthesis and cyclin D1 promoter activity is a good measure of cell proliferation. We found that exogenous expression of E-cadherin or growing cells in reduced calcium medium did not affect DNA synthesis or the ability of RA to inhibit DNA synthesis. At the molecular level RA is known to inhibit G 1 -S progression and cyclin D1 expression. Because the cyclin D1 promoter has AP-1 and TCF sites, we investigated the effects of RA on cyclin D1 promoter activity in SKBR3 cells (high AP-1 activity) and in SW480 colorectal cancer cells (high ␤-catenin/ TCF activity). In SKBR3 cells mutation of the TCF site had little effect on cyclin D1 reporter activity or its response to RA. In contrast the activity and retinoid responsiveness of the cyclin D1 promoter in SW480 cells was markedly inhibited by mutation of the TCF site.
Taken together these results suggest the following model ( Fig. 9). In SKBR3 cells, RA increases the expression of a cadherin that mediates strong cell-cell adhesion and translocates ␤-catenin to the membrane, thereby mediating the effects of RA on cell morphology and differentiation. Because these effects are not regulated by AP-1 or ␤-catenin/TCF signaling it is likely that they are mediated via activation of the RAR/RXR pathway that directly or, more likely, indirectly regulates cadherin expression. However, this RAR/RXR pathway is not required for RA-mediated growth inhibition in SKBR3 cells. Because ␤-catenin/TCF signaling is very low in SKBR3 cells these data are consistent with a role for RA-RAR inhibition of AP-1 activity in mediating the RA effects on SKBR3 cell proliferation and cyclin D1 expression. In other cells such as SW480 colorectal cancer cells that do have high levels of ␤-catenin/TCF signaling RA could inhibit cell proliferation by directly interfering with this pathway rather than AP-1. Thus, the remarkably broad effects of RA on the growth and differentiation of many different epithelial cancers may well be explained by its ability to differentially regulate the activity of these three important pathways. In addition to co-activating TCF-regulated genes ␤-catenin can augment the ability of ligand to activate RA, androgen, and vitamin D-responsive promoters (54,55). Taken together with the recent discovery that wnt and RA signaling cooperate to regulate the expression of the RAresponsive gene Stra6, these data indicate that cross-regulation of wnt/␤-catenin/TCF and steroid receptor pathways may well be important in a number of embryological and neoplastic settings (56).