Ectopic expression of E-cadherin blocks cell cycle progression in epithelial and fibroblastoid cells
To test the relationship between E-cadherin expression and β-catenin/LEF signaling during cell cycle progression, we used the FosER cell system (Reichmann et al., 1992
) for two reasons. First, the cellular phenotype is controlled by estradiol-dependent activity of a constitutively expressed cFos–estrogen receptor chimeric protooncogene. Whereas cells form highly polarized epithelial cell sheets in the absence of estradiol, estradiol-induced activation of FosER induces EMT, leading to fibroblastoid cells that lack E-cadherin and exhibit increased β-catenin/LEF-1 transcriptional activity (Eger et al., 2000
). This allowed us to analyze the effects of ectopic E-cadherin expression on cell proliferation in two different cellular phenotypes of a single cell clone, expressing or lacking endogenous E-cadherin. Second, we have shown previously that transient expression of E-cadherin in fibroblastoid and epithelial FosER cells significantly reduced β-catenin/LEF transcriptional activity (Eger et al., 2000
Upon transient expression of myc- or GFP-tagged full-length E-cadherin in epithelial FosER cells, we observed two major phenotypes. Cells expressing exogenous E-cadherin at a very low level revealed a nearly exclusive localization of the tagged protein at the lateral plasma membrane (
A), similar to the localization of the endogenous E-cadherin. Higher expression levels led to an additional cytoplasmic localization of the ectopic E-cadherin, presumably representing E-cadherin localized in the ER, the Golgi, or in vesicles on their way to the plasma membrane ( B; Adams et al., 1998
). Ectopically expressed E-cadherin colocalized with endogenous β-catenin at the plasma membrane, suggesting that it is incorporated into functional cell–cell adhesion complexes ( A′), but high levels of ectopic E-cadherin led to cytoplasmic accumulation of endogenous β-catenin most likely due to the formation of cytoplasmic E-cadherin–β-catenin complexes ( B′). In fibroblastoid FosER cells, expression of exogenous E-cadherin at moderate levels resulted in its predominantly peripheral localization at the plasma membrane and a partial recruitment of β-catenin to the cellular periphery ( C). Higher expression levels resulted in increased cytoplasmic E-cadherin and a strong upregulation of endogenous β-catenin levels in the cytoplasm ( D). In agreement with previous studies (Gottardi et al., 2001
) showing that E-cadherin expression in colon cancer cells did not change β-catenin's nuclear localization, although it interfered with β-catenin transcriptional activity, we did not observe a clear depletion of β-catenin from the nucleus in E-cadherin–expressing cells.
Figure 1. Transient expression of ectopic E-cadherin. Epithelial and fibroblastoid FosER cells were transfected with an expression plasmid encoding full-length myc-tagged E-cadherin and processed for double immunofluorescence microscopy 24 h after transfection (more ...)
To analyze the effect of ectopic E-cadherin expression on cell cycle progression, we determined the percentage of transfected cells in S phase by testing for the incorporation of BrdU into cellular DNA. 48 h after transfection of cells with E-cadherin constructs or with GFP vectors alone, nuclei of most cells expressing high levels of E-cadherin did not stain for BrdU, whereas a large number of nuclei in nontransfected cells in the same preparation (
A, top) or in GFP-expressing control cells (unpublished data) were stained brightly with the BrdU antibody. Statistical analysis of these data revealed that ~70% of epithelial and fibroblastoid cells transfected with the GFP construct alone incorporated BrdU independent of GFP expression levels ( B). In contrast, upon moderate expression of E-cadherin in epithelial cells (defined by mostly peripheral localization of E-cadherin; A) only 20% of cells exhibited BrdU-specific staining ( B, Epithelial FosER, white bar), and almost no BrdU incorporation could be detected in epithelial cells expressing high levels of the protein in the cytoplasm ( B, Epithelial FosER, black bar).
In fibroblastoid cells, the decrease in BrdU incorporation was less efficient than in epithelial cells, yielding a reduction of BrdU-positive cells to 40% upon low level expression of E-cadherin and to ~5% in highly expressing cells ( B, Fibroblastoid FosER, white and black bars). Thus, E-cadherin expression led to a significant reduction in DNA synthesis, suggesting that increased cellular E-cadherin levels might cause a growth arrest in the G1 phase of the cell cycle. To directly prove G1 phase arrest, we stained transfected cells with an antibody to the cyclin-dependent kinase inhibitor p27KIP1
, which has been found to be upregulated upon E-cadherin–induced growth suppression (St. Croix et al., 1998
). As shown in A (bottom), p27KIP1
was significantly upregulated exclusively in nuclei of cells expressing high levels of E-cadherin.
Figure 2. Expression of E-cadherin leads to G1-S phase arrest. Epithelial and fibroblastoid FosER cells were transfected with expression plasmids encoding full-length GFP- or myc-tagged E-cadherin and analyzed for BrdU incorporation and/or double immunofluorescence (more ...)
Prolonged expression of ectopic E-cadherin leads to apoptosis
To test whether ectopic E-cadherin–mediated G1 phase arrest is a transient or constitutive effect and to analyze the fate of transfected cells upon prolonged cultivation, we transiently transfected epithelial and fibroblastoid FosER cells with E-cadherin–GFP fusion constructs or GFP alone and followed the number of transfected GFP-positive cells over a period of 4 d after transfection by FACS® analysis. When compared with GFP-expressing cells (set to 100%), the number of GFP–E-cadherin–positive epithelial cells dropped dramatically to a value close to zero after 4 d (
A). The number of fibroblastoid cells expressing E-cadherin–GFP was also strongly reduced within 4 d, but as in the BrdU incorporation experiments the reduction was less efficient than in epithelial cells. The decrease in E-cadherin–GFP–expressing cells during prolonged cultivation was consistent with a G1 phase arrest of transfected cells, but the dramatic downregulation within the short time period could not be explained by growth arrest alone. Since the stability of GFP and E-cadherin–GFP fusion proteins were similar (unpublished data), we reasoned that G1 phase–arrested cells might enter apoptosis. Indeed, >50% of cells highly overexpressing E-cadherin showed plasma membrane blebbing and nuclear fragmentation 24–48 h after transfection and were shown by TUNEL staining to undergo apoptosis ( B). After 4 d, only cells expressing low levels of E-cadherin were detected, and <10% of those were apoptotic. Thus, cell cycle arrest by ectopic E-cadherin apparently resulted in cell death upon prolonged cultivation.
Figure 3. Prolonged expression of ectopic E-cadherin caused apoptosis. Epithelial and fibroblastoid FosER cells were transfected with expression plasmids encoding full-length GFP- or myc-tagged E-cadherin, GFP-cadherin and β-catenin, or GFP as control. (more ...)
The growth-suppressive effect of E-cadherin is mediated by domains affecting β-catenin transcriptional activity and is cell adhesion independent
To determine the domains of E-cadherin responsible for growth suppression and apoptosis, we generated E-cadherin deletion mutants, missing either the cytoplasmic juxtamembrane p120ctn binding domain (Ecad-βcat) or the COOH-terminal β-catenin binding domain (Ecad-p120), but both constructs contained the complete extracellular NH2-terminal ectodomains and the transmembrane region (
A). In addition, we generated nonmembrane-anchored E-cadherin fragments, representing either the p120ctn (BD-p120) or the β-catenin (BD-βcat) binding domains. Upon transient transfection, tagged E-cadherin fragments were expressed at significant levels and demonstrated the expected molecular weights as shown by immunoblotting of total cell lysates using anti-myc or anti-GFP antibodies ( B, asterisks). Unlike the membrane-anchored E-cadherin fragment missing the β-catenin binding site ( C, c, Ecad-p120), the β-catenin–binding E-cadherin fragment (Ecad-βcat) localized to the plasma membrane and led to increased β-catenin levels in both epithelial and fibroblastoid cells and caused a partial translocation of β-catenin from the cytoplasm to the membrane in fibroblastoid cells ( C, a). The overexpressed cytoplasmic E-cadherin domains, BD-βcat and BD-p120, were distributed more diffusely throughout the cytoplasm and nucleus ( C, b and d), and as expected the β-catenin fragment ( C, b, BD-βcat) stabilized endogenous β-catenin.
Figure 4. Expression of E-cadherin deletion mutants in epithelial and fibroblastoid FosER cells. (A) Schematic drawing of E-cadherin deletion mutants. p120 and β-cat denote interaction domains of E-cadherin with p120ctn and β-catenin, respectively. (more ...)
To test the effects of transiently expressed E-cadherin mutants on cell cycle progression, we determined BrdU incorporation 48 h after transfection. In fibroblastoid cells lacking endogenous E-cadherin, both constructs containing the β-catenin binding site with (Ecad-βcat) or without the transmembrane domain (BD-βcat) clearly reduced the number of BrdU-positive cells compared with GFP-expressing cells in a dose-dependent manner (
A, right), whereas E-cadherin fragments missing the β-catenin binding site (Ecad-p120; BD-p120) had no significant effect. Furthermore, analysis of β-catenin transcriptional activity using the TOPFLASH luciferase reporter construct (Korinek et al., 1997
) revealed that exclusively the β-catenin–binding E-cadherin fragments causing cell cycle arrest also significantly reduced β-catenin/LEF-1 activity ( B). Thus, the negative effect of E-cadherin fragments on cell proliferation is likely mediated by their interference with the signaling function of endogenous β-catenin and is independent of their function in cell adhesion. To support this hypothesis, we expressed a chimeric E-cadherin–α-catenin protein missing the β-catenin binding site, which is active in mediating cell adhesion but does not interfere with β-catenin signaling in colon carcinoma cells ( A; Gottardi et al., 2001
). As expected, the chimeric protein reduced neither BrdU incorporation nor β-catenin activity in fibroblastoid cells.
Figure 5. Growth-suppressive effect of E-cadherin fragments is strictly correlated with downregulation of β-catenin/LEF-1 transcriptional activity. Epithelial and fibroblastoid FosER cells were transfected with expression plasmids for GFP-tagged full-length (more ...)
In epithelial cells containing endogenous E-cadherin–β-catenin complexes, the effects of E-cadherin fragments seemed to be more complex, since all E-cadherin fragments, including those missing the β-catenin binding site, showed a significant reduction of BrdU incorporation and caused reduction in β-catenin transcriptional activity. This can be explained by the fact that these fragments may interfere with endogenous adhesion complexes and indirectly affect transcriptional activity of endogenous β-catenin (see Discussion).
Taken together, these experiments showed that the ability of E-cadherin fragments to arrest cell cycle progression correlated strictly with their effects on β-catenin signaling activity. Furthermore, the ability of E-cadherin fragments to interfere with β-catenin activity correlated nicely with the induction of apoptosis ( C). This raised the interesting possibility that β-catenin transcriptional activity may be important for cell cycle progression.
E-cadherin's growth-inhibiting activity is counteracted by increased β-catenin activity
If β-catenin's transcriptional activity is essential for cell proliferation, an artificial increase of β-catenin activity in E-cadherin–overexpressing cells should release them from cell cycle arrest. As shown in
A, E-cadherin overexpression in epithelial FosER cells caused a significant (≤50%) reduction of endogenous β-catenin/LEF-1 activity, but coexpression of ectopic β-catenin or LEF-1 or TCF-3 rescued or even increased β-catenin activity compared with the control.
Figure 6. Upregulation of β-catenin activity rescues E-cadherin–mediated growth arrest. Epithelial FosER cells were transfected with reporter constructs and with constructs encoding E-cadherin alone or together with constructs expressing β-catenin (more ...)
To test whether rescue of β-catenin/LEF activity can also rescue cell cycle progression, we used a reporter construct containing multiple E2F binding sites in front of the luciferase gene, which was shown to be activated during G1-S phase transition in proliferating cells (Krek et al., 1993
). The ectopic E-cadherin–mediated cell cycle arrest is reflected by a reduction of E2F-dependent reporter activity to ~60% of the control activity in mock-transfected cells ( B). Coexpression of β-catenin or LEF-1 or TCF-3 together with E-cadherin rescued E2F-dependent transcription to values close to the control or approximately twofold higher levels than the control. Thus, an increase in β-catenin–dependent transcriptional activity can overcome E-cadherin–mediated cell cycle arrest. This conclusion was further supported by the observation that coexpression of β-catenin with E-cadherin–GFP partially rescued the steep decrease in the number of GFP-positive cells seen in E-cadherin–GFP–expressing cells within 4 d of cultivation after transfection ( A, broken line), indicating reduced apoptosis in these cells.
β-Catenin transcriptional activity is upregulated in proliferating versus arrested cells
If β-catenin activity is required for cell cycle progression, the endogenous activity should undergo cell cycle–dependent changes. First, we analyzed endogenous β-catenin transcriptional activity in subconfluent, proliferating versus dense, contact-inhibited cultures. When epithelial cells were grown at low densities (≤40% confluency), they exhibited an increased β-catenin–LEF-1/TCF activity, which was five- to sevenfold reduced when cells reached a confluency of >80% (
A). A similar dependence of β-catenin–LEF-1/TCF–dependent reporter activity on cell confluency was detected in fibroblastoid cells except that the overall values of β-catenin transcriptional activity were up to tenfold higher than in epithelial FosER cells. To account for any unspecific changes in basic transcriptional activity during different cell cycle stages, all β-catenin/LEF-1 activities were related to respective reporter activities obtained with a control reporter containing mutated LEF-1 binding sites unable to bind LEF-1 (FOPFLASH).
Figure 7. β-Catenin/LEF-1 activity depends on cell proliferation. (A) Epithelial and fibroblastoid FosER cells were transfected with TOPFLASH or FOPFLASH luciferase and β-galactosidase reporter constructs at different cell densities and analyzed (more ...)
In a second set of experiments, we arrested cell proliferation by serum starvation of subconfluent epithelial and fibroblastoid FosER cells for 48 h and released them from the cell cycle block for up to 24 h by addition of serum. β-Catenin/LEF-1 activity increased more than twofold upon serum stimulation ( B), confirming a strict correlation of cell proliferation and β-catenin signaling.
Stable expression of ectopic E-cadherin reduced cell proliferation in fibroblastoid cells
Our data suggest that the relative expression levels of E-cadherin versus β-catenin control β-catenin transcriptional activity and cell cycle progression. To address this hypothesis in more detail, we generated fibroblastoid FosER cell clones stably expressing ectopic E-cadherin. Immunofluorescence microscopy revealed that stable E-cadherin–expressing cells reverted to a typical epithelial morphology with a peripheral localization of ectopic E-cadherin at cell contacts (
A). Furthermore, endogenous β-catenin was translocated from the cytoplasm and nucleus in fibroblastoid cells to E-cadherin containing cell–cell contacts. Although the expression levels of ectopic E-cadherin in three independent stable clones were found to be <50% of the amount of endogenous E-cadherin in epithelial cells ( B), β-catenin activity was reduced up to fivefold in E-cadherin–expressing clones versus fibroblastoid cells ( C). The analysis of cell proliferation rates ( D) indicated that the three independent E-cadherin–expressing fibroblastoid cell clones grew significantly slower than the E-cadherin–deficient parental fibroblastoid cells, exhibiting generation times of 20.5, 21.2, and 22.9 h versus 16.2 h. Growth retardation may be mediated by a decrease in protein ( E) and mRNA (unpublished data) levels of the positive cell cycle regulator cyclin D1, whose transcription may be directly controlled by β-catenin (Tetsu and McCormick, 1999
), and the upregulation of a negative cell cycle regulator, the cdk inhibitor p27KIP1
(St. Croix et al., 1998
) ( E). E-cadherin did not increase apoptosis in these clones (unpublished data). The reduction in growth rates upon ectopic expression of E-cadherin in fibroblastoid cells is consistent with the role of β-catenin signaling in cell proliferation and its negative control by E-cadherin.
Figure 8. Stable expression of E-cadherin in fibroblastoid FosER cells reduces cell proliferation rate. Fibroblastoid FosER cells were stably transfected with a construct expressing myc-tagged E-cadherin, and three clones were selected by hygromycin resistance. (more ...)