Calreticulin Represses E-cadherin Gene Expression in Madin-Darby Canine Kidney Cells via Slug*

Calreticulin (CRT) is a multifunctional Ca2+-binding molecular chaperone in the endoplasmic reticulum. In mammals, the expression level of CRT differs markedly in a variety of organs and tissues, suggesting that CRT plays a specific role in each cell type. In the present study, we focused on CRT functions in the kidney, where overall expression of CRT is quite low, and established CRT-overexpressing kidney epithelial cell-derived Madin-Darby canine kidney cells by gene transfection. We demonstrated that, in CRT-overexpressing cells, the morphology was apparently changed, and the original polarized epithelial cell phenotype was destroyed. Furthermore, CRT-overexpressing cells showed enhanced migration through Matrigel®-coated Boyden chamber wells, compared with controls. E-cadherin expression was significantly suppressed at the protein and transcriptional levels in CRT-overexpressing cells compared with controls. On the other hand, the expression of mesenchymal protein markers, such as N-cadherin and fibronectin, was up-regulated. We also found that the expression of Slug, a repressor of the E-cadherin promoter, was up-regulated by overexpression of CRT through altered Ca2+ homeostasis, and this led to enhanced binding of Slug to the E-box element in the E-cadherin promoter. Thus, we conclude that CRT regulates the epithelial-mesenchymal transition-like change of cellular phenotype by modulating the Slug/E-cadherin pathway through altered Ca2+ homeostasis in cells, suggesting a novel function of CRT in cell-cell interaction of epithelial cells.

ulation of Ca 2ϩ homeostasis (2), intercellular or intracellular signaling (3,4), gene expression (5), glycoprotein folding (6), and nuclear transport (7). The biological significance of CRT was revealed by the finding that CRT-deficient mice die in the embryonic stage due to impaired development of cardiac and neural tissues (8,9). CRT is expressed in rat embryos, especially in the heart, but its expression is significantly suppressed after birth (10). On the other hand, CRT-overexpressing transgenic mice are born alive, but suffer a complete heart block and sudden death after birth (11). We also found that overexpression of CRT enhanced sensitivity to apoptosis in myocardial H9c2 cells undergoing differentiation in response to retinoic acid (12) or in cells exposed to stress caused by hydrogen peroxide (13), suggesting the importance of CRT in the pathophysiology of myocardial cells. These findings indicate that CRT expression plays a vital role in the development and physiology of cardiac cells.
Despite its general importance in cell physiology, CRT is differentially expressed in various organs and tissues in mammals, showing a characteristic expression pattern. For example, CRT levels are low in the kidney and heart, compared with the pancreas and liver, in both bovine and rat tissues (14,15). This characteristic distribution of CRT suggests specific functions in each organ or tissue.
In this study, we focused on the function of CRT in kidney epithelial cells, because CRT levels are quite low in these cells compared with other cell types such as liver cells (14). To investigate the functional effects of CRT overexpression in kidney epithelial cells, we chose Madin-Darby canine kidney (MDCK) cells to establish stable CRT-overexpressing cell lines by gene transfection. MDCK cells are derived from canine kidney and have a well polarized epithelial cell phenotype, maintaining the normal characteristics and functions of renal efferent duct epithelial cells (16). The results showed apparent changes in the morphology of CRT-overexpressing MDCK cells and destruction of the polarized epithelial cell phenotype. Furthermore, overexpression of CRT repressed E-cadherin gene expression through up-regulation of its repressor, Slug, via altered Ca 2ϩ homeostasis in MDCK cells. The results suggest a novel function of CRT related to an epithelial-mesenchymal transitionlike change of cellular phenotype. mary antibody in TBS containing 0.05% Tween 20. The blots were coupled with peroxidase-conjugated secondary antibodies, washed, and then developed using the ECL detection kit (Amersham Biosciences) according to the instructions recommended by the manufacturer.
Immunofluorescence Microscopy-Cells (5 ϫ 10 5 per ml) were grown on Lab Tek chamber slides (Nalgen Nunc International, Naperville, IL) for 24 h. They were fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS, pH 7.2) and permeabilized for 10 min with PBS containing 1% Triton X-100. The cells were then blocked with 1% bovine serum albumin in PBS, incubated with the antibody for 1 h, and washed with PBS containing 1% bovine serum albumin. The immunoreactive primary antibodies were visualized with fluorescein isothiocyanate-conjugated anti-rabbit immunoglobulins (Cappel), anti-mouse immunoglobulins (Dako), or rhodamine-conjugated anti-rabbit immunoglobulins (Cappel). After being washed, stained cells were mounted in the Vectashield medium, visualized under a Carl Zeiss LSM5 microscope (Carl Zeiss, Jena, Germany), and analyzed using PASCAL analytic software.
Invasion Assays-A cell invasion assay was carried out using Boyden chambers (Transwell chambers) as described previously (18) with a slight modification. In brief, Transwell chambers, equipped with 8-m Matrigel-coated filters (24-well format, BD Biosciences), were rehydrated, and suspensions of 1 ϫ 10 5 cells in 200 l of Dulbecco's modified Eagle's medium containing 0.1% fetal calf serum were plated in the upper compartment of the chamber. Serum-free medium (800 l) was placed in the lower compartment. After 24 h at 37°C, noninvasive cells on the upper surface of the filters were removed completely by wiping with a cotton swab. The filters were then fixed with 4% paraformaldehyde in PBS and stained with 0.01% crystal violet. Cells on the lower surface were photographed under a microscope (magnification ϫ 100), and enumerated. The data were expressed as the mean Ϯ S.D. of assays performed in triplicate for each filter.
Cell Proliferation Assays-Cell proliferation assays were performed as described previously (12). The proliferation of cultured cells was evaluated by measuring attached live cells photometrically after staining with crystal violet. Cells were seeded onto 96-well plates at a density of 3000 cells per well in 100 l of medium. After culturing for the times indicated in the text, cells were fixed with 4% paraformaldehyde in PBS, washed three times, and stained with 0.01% crystal violet at room temperature for 20 min. After an extensive wash with water, each well was dried. The stained cells were dissolved in 100 l of 10% SDS and 0.1 M HCl, and cell numbers were estimated by measuring the absorbance at 570 nm using a microplate reader.
Electrophoretic Mobility Shift Assays-EMSA for the E-boxes were performed as described previously (19) with a slight modification. Oligonucleotide probes were labeled with [␥-32 P]ATP using T4 polynucleotide kinase, and then annealed to double-stranded oligonucleotides. Specific oligonucleotides for E-boxes were prepared according to the nucleotide sequence of the canine E-cadherin gene promoter (GenBank TM accession numbers AF330162 and AF330163). Oligonucleotides used were as follows: E-box A probe, 5Ј-CCG CCC GCC GCA GGT GCA GCC GCA G-3Ј; E-box B probe, 5Ј-CTC GCG GCT CAC CTG GCG GCC GGA C-3Ј; and E-box C probe, 5Ј-GGC GCT GCG GGC ACC TGT GAT TC-3Ј (bold letters indicate consensus sequences for the E-box). Binding reactions were carried out in 15 l of reaction mixture (25 mM Tris-HCl (pH 7.0), 6.25 mM MgCl 2 , 0.5 mM dithiothreitol, and 10% glycerol) containing 10 g of nuclear extract and 25 ng of labeled oligonucleotides. For the supershift assay, anti-Slug and anti-SIP1 antibodies were added to the reaction mixture during the 30-min binding reaction.
Promoter Reporter Assays-The promoter of human E-cadherin (Ϫ178 to ϩ66 bp: GenBank TM accession number L34545) was isolated from the genomic DNA of A549 cells and amplified by PCR using Pfu turbo DNA polymerase (Stratagene). The primer sequences used were as follows (20): forward primer (5Ј-ACT CCA GGC TAG AGG GTC A-3Ј) and reverse primer (5Ј-TGG AGC GGG CTG GAG TCT-3Ј). The PCR product was subcloned into pGL3-Basic vector (KpnI-PstI site, Promega, Madison, WI). PCR-based site-directed mutagenesis was used for the generation of reporter gene constructs with E-box mutations by using a QuikChange site-directed mutagenesis kit (Stratagene), resulting in a mutation in the E-box element from 5Ј-CANNTG-3Ј to 5Ј-AANNTA-3Ј (sense strand). Each vector was transiently transfected into MDCK cells using Lipofectamine 2000 (Invitrogen) as described above. Twenty-four hours after the transfection, luciferase activities were assayed with cellular extracts using a Dual Luciferase Reporter Assay System (Promega) and were normalized to pRL activity.
Assays for Release and Uptake of Ca 2ϩ in the Cell-For the 45 Ca 2ϩ release assay, cells were cultured for 48 h with medium containing 45 Ca 2ϩ (1 Ci/ml). After a wash with Ca 2ϩ -free Earle's balanced salt solution (EBSS, Invitrogen) containing 3 mM EGTA, cells were detached from the culture plates with trypsinization buffer (0.25% trypsin and 0.02% EDTA in EBSS), and the cell suspensions were preincubated in Ca 2ϩ -free EBSS at 37°C for 3 min, and sequentially stimulated with thapsigargin (0.1 M), ionomycin (2 M), and monensin (2 M). The cell suspensions were collected 5 min after the addition of each reagent and centrifuged. The radioactivity released from the cells was measured in the supernatant. Cell pellets were lysed, and protein amounts were determined using a BCA assay kit (Pierce). 45 Ca 2ϩ release was expressed as the cpm subtracted from those recovered in the preceding collection, and normalized to the protein in the corresponding cell pellets. The uptake of Ca 2ϩ was measured radiometrically using the Millipore filtration technique as described previously (13) with a slight modification. The cells were washed with 45 Ca 2ϩ uptake buffer, which consisted of EBSS supplemented with 0.1 mM CaCl 2 , and cultured for specific periods in 45 Ca 2ϩ uptake buffer containing 45 Ca 2ϩ (5 Ci/ml). Cells were detached from the culture plates by trypsinization buffer, and the cell suspension was filtered through a 0.45-m nitrocellulose filter (Bio-Rad) under vacuum. The filters were rinsed twice with 0.5 ml of washing buffer (10 mM Hepes (pH 7.4), 150 mM KCl, 2 mM EGTA, and 2.5 mM MgCl 2 ). 45 Ca 2ϩ uptake was calculated by measuring the radioactivity and standardized using protein concentrations.
Measurement of Cytoplasmic-free Ca 2ϩ -The cytoplasmic free Ca 2ϩ concentration, [Ca 2ϩ ] i , was measured with a dualexcitation wavelength fluorescence microscope using Fura-2. Cultured cells on quartz-bottom dishes were loaded with 5 M Fura-2 tetra(acetoxymethyl)ester (Fura-2-AM, Dojindo, Kumamoto, Japan) for 20 min in EBSS with 2 mM CaCl 2 in the pres-ence of 0.01% pluronic acid F-127. After four washes with EBSS, Fura-2 fluorescence was determined at 37°C using an IX71 inverted research microscope (Olympus, Tokyo, Japan) and a FURA ratiometric imaging system operating at an emission wavelength of 505 nm with an excitation wavelength of 340 and 380 nm. The FURA ratiometric imaging system includes filters (Chroma Technology Corp., Rockingham, VT) switched by filter wheels (Sutter Instrument Company, Novato, CA) and a MicroMax camera (Roper Scientific, Tucson, AZ) controlled by SlideBook software. To measure the change in [Ca 2ϩ ] i during store-operated Ca 2ϩ influx, Fura-2-labeled cells were washed with Ca 2ϩ -free EBSS, then stimulated with thapsigargin (5 M) followed by re-addition of Ca 2ϩ (2 mM). Stimulated calcium release was calculated as the change in the excitation ratio from baseline integrated over 800 s of stimulation with thapsigargin or Ca 2ϩ . The maximal signal (R max ) was obtained by adding ionomycin at a final concentration of 4 M. The minimal signal (R min ) was then obtained by adding EGTA at a final concentration of 10 mM, followed by Tris-free base to a final concentration of 30 mM, to increase the pH to 8.3. R is the ratio (F 1 /F 2 ) of the fluorescence of Ex 340 nm, Em 505 nm (F 1 ) to that of Ex 380 nm, Em 505 nm (F 2 ). The actual calcium concentration was calculated as K d ϫ (R Ϫ R min )/(R max Ϫ R) ϫ Sf2/Sb2 with the K d equal to 224 nM (21). Sf2/Sb2 is the ratio of Fura-2 fluorescence at 380 nm in Ca 2ϩ -free and Ca 2ϩ -replete medium, respectively.

Overexpression of CRT Causes
Morphological Change in MDCK Cells-Canine renal epithelial MDCK cells were transfected with the expression vector for CRT cDNA to obtain cell lines overexpressing CRT (MDCK-CRT1 and -CRT2). The expression level of CRT was examined by immunoblot analysis in the gene-transfected cells using specific antibodies as described under "Experimental Procedures." Fig. 1A shows that the expression of CRT was increased in the overexpressers to ϳ3-fold the levels in the parental (MDCK-WT) and mock-transfected MDCK (MDCK-Control) cells. The transfection had no apparent effect on the expression of molecular chaperones in the ER, such as BiP, ERp57, and cytosolic GAPDH. However, the expression of CNX, another membrane-bound ER homologue of CRT, showed a slight decrease in the CRT-overexpressing cells. Cell morphology was examined in control and CRT gene-transfected cells by phase-contrast microscopy (Fig. 1B). MDCK cells are known to grow in colonies of adherent cells (22,23). Overexpression of CRT (MDCK-CRT1 and CRT2) caused an apparent morphological change with a fibroblastoid-like phenotype and loss of cell-cell contacts, although there was no remarkable morphological change in mock-transfected cells compared with parental cells. Intracellular localization of CRT was characterized by immunofluorescence microscopy in control and CRT-overexpressing cells ( Fig. 2A). Under conditions in which cellular membranes were permeabilized by Triton X-100, strong immunoreactivity for CRT showed a perinuclear localization and a vesicular pattern in CRT-overexpressing cells, although the immunoreactive signal was weak in controls. No significant increase in the cell surface expression of CRT was observed in the CRT-overexpressing cells under conditions without Triton X-100 treatment. To investigate whether the cytosolic localization of CRT was increased in the gene-transfected cells, control and MDCK-CRT1 cells were lysed and fractionated by centrifugation to separate cytosolic and microsomal fractions as described under "Experimental Procedures." As shown in Fig. 2B, overexpressed CRT was present in the microsomal but not cytosolic fraction. CNX and Cu,Zn-superoxide dismutase were detected as marker proteins for microsomes and the cytosol, respectively. Similar results were also obtained with MDCK-CRT2 cells (data not shown). Together, these results indicate that overexpression of CRT did not influence the localization of CRT in the ER of MDCK cells.
Overexpression of CRT Enhances Cellular Migration in MDCK Cells-To investigate whether the altered morphology in CRT-overexpressing cells affected cellular functions, a cell invasion assay was performed using a modified Boyden chamber as described under "Experimental Procedures." Cells were seeded on Transwell filters that had been coated with extracellular matrix components, including laminin, fibronectin, and proteoglycans (i.e. Matrigel-coated polycarbonate membrane). After 24 h, the numbers of cells that had migrated through the filters were estimated by enumerating stained cells. There was a significant increase in cell motility through a Matrigel-coated polycarbonate membrane in CRT-overexpressing cells compared with controls (Fig. 3, A and B). To investigate whether the enhanced migration of CRT-overexpressing cells was due to an increase in cell growth, cell proliferation was examined in control and CRT gene-transfected cells as described under "Experimental Procedures." However, the results showed that cell growth was suppressed in CRToverexpressing cells compared with controls (Fig. 3C). These results indicate that the increase in migration of CRT-overexpressing cells was not simply due to an increase in cell growth, implicating other mechanisms related to cell movement, such as cell adhesion and cell de-attachment.
Overexpression of CRT Suppresses E-cadherin Expression-E-cadherin plays an important role in cell-cell interaction in MDCK cells, and a loss of E-cadherin is closely associated with enhanced migration of epithelial cells via a mechanism known as epithelial-mesenchymal transition (EMT) (24). In addition, mesenchymal protein markers (such as N-cadherin and fibronectin) are expressed in EMT (25). To investigate whether overexpression of CRT affected the expression of E-cadherin and EMT-related proteins, an immunoblot analysis was carried out to examine the expression level of EMT-related proteins, such as E-cadherin, N-cadherin, ␤-catenin, fibronectin, and vinculin, in control and CRT-overexpressing cells (Fig. 4A). The results showed that the expression level of E-cadherin was apparently decreased in CRT-overexpressing cells compared with control cells. The expression of ␤-catenin, a cytoplasmic signaling molecule associated with E-cadherin, was slightly decreased in CRT-overexpressing cells. In contrast, expression of N-cadherin, fibronectin, and vinculin was apparently induced in CRT-overexpressing cells. The intracellular localization of E-cadherin was characterized by immunofluorescence microscopy (Fig. 4B). In controls, immunoreactivity for E-cadherin was strong and mainly located at cell-cell contact regions. However, the signal was apparently diminished in CRT-overexpressing cells. These results were consistent with the data from the immunoblot analysis (Fig. 4A). Taken together, these results indicate that overexpression of CRT caused a decrease in E-cadherin, and an increase in N-cadherin, fibronectin, and vinculin in MDCK cells, resulting in a gain of migratory characteristics in the cells. This also suggests that CRT expression might be involved in the regulatory mechanism of EMT in the conversion of early stage tumors into invasive malignancies.  , ϫ100). B, invasion assay was performed as described in A, and the number of cells that migrated to the lower surface of filters through the Matrigelா was counted after staining with 0.01% crystal violet and estimated in control and CRT-overexpressing cells. Data represent the mean Ϯ S.D. of three independent experiments. C, proliferation of cultured cells was evaluated by measuring attached live cells photometrically after staining with crystal violet as described under "Experimental Procedures." Cells were seeded onto 96-well plates at a density of 3000 cells per well in 100 l of medium. After culturing for the times indicated in the text, cells were fixed with 4% paraformaldehyde, and stained with 0.01% crystal violet. After a wash, the stained cells were dissolved in 10% SDS and 0.1 M HCl, and then the cell number was estimated by measuring the absorbance at 570 nm using a microplate reader.
Expression of Slug, a Repressor of the E-cadherin Gene, Is Upregulated in CRT-overexpressing MDCK Cells-To investigate whether E-cadherin expression is down-regulated in CRToverexpressing cells at the transcriptional level, we examined the E-cadherin mRNA level by Northern blot analysis in control and CRT-overexpressing cells. As shown in Fig. 5, E-cadherin mRNA expression was definitely reduced in CRT-overexpressing cells. The expression of the E-cadherin gene is known to be regulated by several transcriptional repressors, including Snail, Slug, ␦EF1/ZEB1, SIP1/ZEB2, E12/E47, and Twist (25)(26)(27). To investigate whether some of the repressors are involved in reg-ulating the suppression of E-cadherin gene expression, the mRNA levels of the repressors were examined by Northern blot analysis using specific probes as described under "Experimental Procedures." As shown in Fig. 5, the expression of Slug mRNA was apparently up-regulated in CRT-overexpressing cells. The level of SIP1 mRNA was also slightly increased in the gene-transfected cells compared with the controls. On the other hand, there was little deference in the expression levels of Snail and Twist and no expression of ␦EF1 and E12/E47 in control and CRT-overexpressing cells. Therefore, we focused on the functions of Slug and SIP1 in the mechanism of E-cadherin gene suppression in CRT-overexpressing cells.

Slug Binds to the E-box Element in the E-cadherin Gene Promoter to Repress Gene Expression in CRToverexpressing MDCK Cells-Slug
and SIP1 block E-cadherin transcription by binding to specific DNA sequences (5Ј-CANNTG) called E-boxes (27). As shown in Fig.  6A, in the canine E-cadherin gene promoter, there are two E-box sites located from Ϫ78 to Ϫ73 (designated as E-box A) and from Ϫ28 to Ϫ23 (designated as E-box B), and downstream of the transcription start site there is another 5Ј-CANNTG sequence positioned from ϩ22 to ϩ27 (designated as E-box C) (23,28,29). To investigate whether Slug or SIP1 could interact with E-boxes in the E-cadherin gene promoter, EMSA was performed with nuclear extracts from control (MDCK-Control) and CRT-overexpressing (MDCK-CRT1) cells using 32 P-labeled oligonucleotides designed for each E-box (Fig. 6B). In the case of E-box A, a major band appeared with the extracts from MDCK-CRT1 cells but not with extracts from controls. The shifted band was not affected by the presence of antibodies against Slug and SIP1, suggesting that Slug and SIP1 had no correlation with E-cadherin gene repression. In the case of E-box B, a major band also appeared with extracts from MDCK-CRT1 cells but not with control extracts. The shifted band was not affected by the anti-SIP1 antibody but disappeared with addition of the anti-Slug antibody, suggesting that Slug was involved in repression of the E-cadherin gene in MDCK-CRT1 cells. In the case of E-box C, no shifted band appeared in either control or MDCK-CRT1 cells, suggesting that E-box C was not involved in the regulation of E-cadherin. The shifted bands completely disappeared when a 500-fold excess of unlabeled oligonucleotide probes was added. Similar results were also obtained with MDCK-CRT2 cells (data not shown). Taken together, these results indicate that, in CRT-overexpressing cells, the DNAprotein interactions are enhanced in E-box A and B in the E-cadherin promoter, and that Slug specifically interacts with E-box B, implicating Slug in E-cadherin gene repression by overexpression of CRT. Next, to elucidate whether the binding of Slug to E-box B in the E-cadherin promoter was linked to the transcriptional activation of E-cadherin, we examined the E-cadherin promoter activity in control and CRT-overexpressing cells using a luciferase reporter plasmid, in which a proxi-mal E-cadherin promoter fragment containing E-boxes A, B, and C is connected upstream of the luciferase gene. To investigate the importance of E-boxes in the transcriptional repression, a mutation was generated in E-boxes A and B (from 5Ј-CANNTG to 5Ј-AANNTA) as described under "Experimental Procedures" (Fig. 6C). The activity of luciferase was assayed with the cells transfected with each vector. As shown in Fig. 6D, in MDCK-CRT1 cells, the promoter activity was suppressed to ϳ50% of the control level. The repression of promoter activity in MDCK-CRT1 cells was apparently restored by mutation of E-box B, but not by mutation of E-box A. Similar results were also obtained with MDCK-CRT2 cells (data not shown). These results suggest that E-box B plays the most significant role in the repression of E-cadherin gene transcription in CRT-overexpressing cells.
Ca 2ϩ Homeostasis Is Altered in CRT-overexpressing MDCK Cells-In previous reports, overexpression of CRT led to an increase in the intracellular store of Ca 2ϩ (1). CRT also appears to modulate store-operated Ca 2ϩ influx (1,30). We also reported that overexpression of CRT influences Ca 2ϩ homeostasis in myocardial H9c2 cells under various stressful conditions (12,13). However, the mechanism by which overexpression of CRT influenced Ca 2ϩ homeostasis was not clear in MDCK cells. To investigate whether the intracellular store of Ca 2ϩ was affected by overexpression of CRT, intracellular Ca 2ϩ pools were characterized in control and CRT-overexpressing cells. After 48 h of loading with 45 Ca 2ϩ , the cells were washed and resuspended in Ca 2ϩ -free buffer. Unidirectional fluxes to the extracellular medium after stimulation with several Ca 2ϩ modulators were then measured as described under "Experimental Procedures." Thapsigargin (an inhibitor of sarcoplasmic/endoplasmic reticulum Ca 2ϩ -ATPase), ionomycin (a Ca 2ϩ ionophore), and monensin (another ionophore affecting acidic stores) were used to stimulate the cellular Ca 2ϩ pools (Fig. 7A). The results showed that cellular Ca 2ϩ levels were apparently increased mainly in the thapsigarginsensitive Ca 2ϩ pools of CRT-overexpressing cells, compared with controls, suggesting that Ca 2ϩ stores in the ER were increased in CRT-overexpressing cells.
Next, to investigate whether the intracellular Ca 2ϩ homeostasis was affected by overexpression of CRT, the cytoplasmic free Ca 2ϩ concentration ([Ca 2ϩ ] i ) and response of [Ca 2ϩ ] i to thapsigargin were examined in control and CRT-overexpressing cells using Fura2-AM as described under "Experimental Procedures." The release of Ca 2ϩ from ER stores by thapsigargin was measured in the absence of extracellular Ca 2ϩ , and the influx of Ca 2ϩ from the extracellular space was measured by re-adding Ca 2ϩ to the external medium. As shown in Fig. 7 (B  and C), the basal level of [Ca 2ϩ ] i was slightly increased in CRToverexpressing cells, compared with controls. Thapsigargin induced a greater increase of [Ca 2ϩ ] i in CRT-overexpressing cells than in the controls, indicating that substantially more Ca 2ϩ was released from Ca 2ϩ stores in CRT-overexpressing cells. Furthermore, the Ca 2ϩ influx-induced increase of [Ca 2ϩ ] i when Ca 2ϩ was re-added was also more extensive in CRT-overexpressing cells than the controls. These results suggest that the level of [Ca 2ϩ ] i was relatively high in CRT-overexpressing cells compared with controls. However, it was reported that store-

FIGURE 5. Northern blot analysis was carried out for E-cadherin and transcriptional repressors of the E-cadherin gene in control and CRT-overexpressing MDCK cells. Northern blot analysis was carried out in control (MDCK-WT and -Control) and CRT gene-transfected MDCK (MDCK-CRT1 and
-CRT2) cells. Specific oligonucleotide probes for E-cadherin, Snail, Slug, ␦EF1, SIP1, E12/E47, TWIST, and GAPDH were prepared and used for the assay as described under "Experimental Procedures." operated Ca 2ϩ entry was suppressed by overexpression of CRT in several cell types (30) and was inconsistent with the present results of CRT-overexpressing MDCK cells.
To further investigate why the influx of Ca 2ϩ was enhanced in CRT-overexpressing cells, we examined the expression of epithelial Ca 2ϩ channels in control and CRT-overexpressing cells. In Fig. 8A, the expression of epithelial Ca 2ϩ channels related to cellular Ca 2ϩ uptake, such as TRPV5, TRPV6 (31), and polycystin 2 (32), was examined by immunoblot analysis using specific antibodies in control and CRT-overexpressing cells. The expression of TRPV5 was stronger in the CRT-overexpressing cells. In contrast, TRPV6 was not detected in either of the cells. Polycystin 2 was similarly expressed in both control and CRT-overexpressing cells. In Fig.  8B, the expression of TRPV5 was also examined by immunofluorescence microscopy. The immunoreactive signals for TRPV5 showed a perinuclear vesicular distribution and were abundant in CRToverexpressing cells compared with less signals in control cells. In Fig. 8C, transcriptional levels of TRPV5 and -6 were examined in control and CRT-overexpressing cells by RT-PCR using specific primers as described under "Experimental Procedures." The results showed that the transcriptional expression of TRPV5 was detected in CRT-overexpressing cells but not in control cells under the experimental conditions. The expression of TRPV6 was not detected in either control or CRToverexpressing cells. For positive controls, RNAs from rat kidney and prostate were used for RT-PCR to detect TRPV5 and -6, respectively (33). Together, these results indicate that the expression of TRPV5 is apparently upregulated in CRT-overexpressing cells compared with controls, suggesting a functional link with the up-regulation of Ca 2ϩ influx in CRT-overexpressing cells. To investigate whether the basal level of Ca 2ϩ influx was affected in CRT-overexpressing cells, 45 Ca 2ϩ uptake was examined in control and CRT-overexpressing cells as described under "Experimental Procedures." As shown in Fig. 8D, the uptake of 45 Ca 2ϩ was enhanced in CRT-overexpressing cells, compared with controls. This seems to be compatible with the up-regulated expression of TRPV5 in CRT-overexpressing cells. Collectively, these results indicate that intracellular Ca 2ϩ homeostasis is apparently altered by overexpressed CRT in MDCK cells, and this may lead to sustained up-regulation of [Ca 2ϩ ] i in the cells.

Overexpression of CRT Downregulates E-cadherin Gene Expression through Alteration of Ca 2ϩ
Homeostasis-To determine whether the increase in [Ca 2ϩ ] i was part of the causative mechanism for E-cadherin gene repression in the gene-transfected cells, E-cadherin expression was examined in control cells treated with thapsigargin (5 M) or ionomycin (1 M) to increase [Ca 2ϩ ] i . RT-PCR analysis was performed to examine E-cadherin expression at the mRNA level. As shown in Fig. 9A, the mRNA levels of E-cadherin were decreased in both cases with thapsigargin and ionomycin in a time-dependent manner. The expression of E-cadherin was also examined by immunofluorescence microscopy after a 4-h treatment with thapsigargin or ionomycin (Fig. 9B). The results showed that the immunoreactive signal for E-cadherin was diminished in both cases with thapsigargin and ionomycin, compared with untreated controls. To investigate further the effect of increased [Ca 2ϩ ] i on the repressors of the E-cadherin gene, we examined mRNA levels of Slug by RT-PCR analysis in control cells treated with or without thapsigargin or ionomycin for 1 h. Expression of Slug was increased in both cases with thapsigargin and ionomycin, compared with untreated controls (Fig. 9C). The expression of Slug was also examined by immunofluorescence microscopy in control cells after 4 h of treatment with thapsigargin or ionomycin (Fig. 9D). The results showed that the immunoreactive signal for Slug was increased in both cases with thapsigargin and ionomycin, compared with untreated controls. In the cells treated with thapsigargin or ionomycin, increased signals for Slug were detected in the cytoplasm and nucleus and were similar with the signals in the CRT-overexpressing cells (MDCK-CRT1). Taken together, these results indicate that the rise in [Ca 2ϩ ] i increased the expression of Slug and decreased the expression of E-cadherin, suggesting a Ca 2ϩ -dependent repression of E-cadherin through Slug in MDCK cells.
Next, to further investigate whether the repression was caused by an rise in [Ca 2ϩ ] i in CRT-overexpressing cells, the effect of BAPTA-AM, a cell-permeable Ca 2ϩ chelator that reduces intracellular Ca 2ϩ levels, on the expression of E-cad- herin or Slug was examined in MDCK-CRT1. RT-PCR analysis was performed to examine E-cadherin expression at the mRNA level. As shown in Fig. 10A, the mRNA levels of E-cadherin were increased by BAPTA-AM (10 M) in a time-dependent manner. The expression of E-cadherin was also examined by immunofluorescence microscopy after 8 h of treatment with BAPTA-AM (Fig. 10B). The results showed that the immunoreactive signal for E-cadherin was increased by BAPTA-AM, compared with untreated MDCK-CRT1. Similar results were also obtained with MDCK-CRT2 cells (data not shown). To investigate further the effect of decreased [Ca 2ϩ ] i on the repressors of the E-cadherin gene, we examined mRNA levels of Slug by RT-PCR analysis in MDCK-CRT1 cells treated with BAPTA-AM. Expression of Slug was decreased by BAPTA-AM, compared with untreated MDCK-CRT1 cells (Fig. 10C). The expression of Slug was also examined by immunofluorescence microscopy in the gene-transfected cells after 8 h of treatment with BAPTA-AM (Fig. 10D). The results showed that the immunoreactive signal for Slug was decreased by BAPTA-AM, compared with untreated MDCK-CRT1 cells. Similar results were also obtained with MDCK-CRT2 cells (data not shown). Taken together, these results indicate that a drop in [Ca 2ϩ ] i decreased the expression of Slug and increased the expression of E-cadherin. They also suggest a Ca 2ϩ -dependent repression of E-cadherin through Slug in CRT-overexpressing cells.
In addition, the binding ability of E-box elements of the E-cadherin promoter was also examined by EMSA in control cells treated with or without thapsigargin (5 M) or ionomycin (1 M) for 2 h (Fig. 11A). The results showed that a positive gel-shift band appeared in response to the treatment with both thapsigargin and ionomycin in the case of E-box B, but not E-boxes A or C. The gel-shift band induced by thapsigargin or ionomycin was similar to that seen in the case of nuclear extracts from CRT-overexpressing cells (MDCK-CRT1). Next, to confirm the Ca 2ϩ -dependent up-regulation of E-cadherin gene expression, CRT-overexpressing cells were treated for 2 h with BAPTA-AM (10 M), and then the binding ability of E-box elements of the E-cadherin promoter was examined by EMSA (Fig. 11B). The results showed that the proteinbinding ability of E-box B apparently decreased in MDCK-CRT1 cells treated with BAPTA-AM, although that of E-box A also slightly decreased with BAPTA-AM. There was no influence of BAPTA-AM on the binding ability of E-box C. Similar results were also obtained with MDCK-CRT2 cells (data not shown). Taken together, these results indicate that overexpression of CRT in MDCK cells causes an increase in [Ca 2ϩ ] i , which up-regulates Slug expression, resulting in suppression of E-cadherin gene expression via E-box B in the gene promoter.

DISCUSSION
In this study, we have shown that overexpression of CRT repressed E-cadherin gene expression in MDCK cells, leading to alteration of cell characteristics such as morphology and motility. E-cadherin, a Ca 2ϩ -dependent transmembrane glycoprotein, plays a key role in the maintenance of intercellular adhesion, regulation of tissue morphogenesis, and cell polarity in epithelial cells, and disruption of E-cadherin expression or function causes invasion and metastasis (34 -37). Therefore, E-cadherin is thought to be an invasion suppresser (38,39). E-cadherin-mediated cell-cell adhesion plays a critical role in early embryonic development through a mechanism known as EMT (24). EMT also determines progression in epithelial tumors, occurring concomitantly with the acquisition of migratory properties followed by down-regulation of E-cadherin expression (40,41). In CRT-overexpressing MDCK cells, the morphology changed from a polarized, epithelial phenotype to a highly motile fibroblastoid-like, mesenchymal phenotype. In addition to a decrease in E-cadherin expression, the expression level of mesenchymal markers, such as N-cadherin and fibronectin, was apparently up-regulated in the gene-transfected cells. The conversion of the cadherin class is observed in EMT in general (42)(43)(44). These results strongly suggest that overexpression of CRT could cause EMT-like changes in MDCK cells.
Recently, it has been reported that EMT is caused by several transcriptional repressors of E-cadherin, such as the zinc finger factors (i.e. Snail (22,23), Slug (45,46), ␦EF1/ZEB1 (47,48), and SIP1/ZEB2 (28,49)) and the basic helix-loop-helix factors (i.e. E12/ E47 (27,50) and Twist (51,52)). The zinc finger factors have been described to directly repress transcription of the E-cadherin gene by binding to E-boxes (consisting of the sequence 5Ј-CANNTG) in the proximal E-cadherin promoter (23,28,29,53), and E-boxes are the recognition sites of the basic helixloop-helix family (54). Bolos et al. (46) reported that stable expression of Slug in MDCK cells leads to full repression of E-cadherin at the transcriptional level and triggers a complete EMT, although the binding affinity of Slug to the E-box is lower than that of Snail and E12/E47. Slug is also responsible for repression of the E-cadherin gene through binding to E-box C in human breast cancer cells (29). On the other hand, Snail is also a repressor of E-cadherin expression in various epithelial tumor cells (23). In SIP1-expressing human mammary cancer cells, mutation of E-box B (designated as E2-box 3) restored the E-cadherin promoter activity similarly to mutation of E-Box A (designated as E2-box1) (28). These studies indicate that the regulatory system for repression of the E-cadherin gene is complex, suggesting that the specific repressors could work in a cell type-dependent manner (29). In this study, we showed that overexpression of CRT induced the expression of Slug and that this enhanced the binding of Slug to E-box B in the E-cadherin gene promoter to repress E-cadherin expression in MDCK cells.
In addition to transcriptional repression, several molecular mechanisms that down-regulate E-cadherin function have been reported, including gene mutation (55), promoter methylation (56), and posttranslational modification of the cadherin-catenin complex (57). Indeed, we observed that, in CRT-overexpressing cells, the promoter activity of E-cadherin was suppressed to ϳ50% of the control level, despite extensive suppression of E-cadherin gene expression. Therefore, other regulatory mechanisms as stated above may also be involved in E-cadherin gene repression in CRT-overexpressing cells, suggesting that further investigation is required. CRT modulates cell adhesiveness via the expression of N-cadherin and vinculin and affects Wnt signaling pathways in L-fibroblasts (58,59). In that study, the authors found that CRT overexpression increased cell adhesion with a concomitant increase of N-cadherin and vinculin in fibroblasts. On the other hand, we found that overexpression of CRT in MDCK cells caused a loss of cellcell interaction and induced cell migration through the Matrigel (Figs. 1B, 3A, and 3B). In terms of the effect of overexpressed CRT on cell-cell interaction, the reason for the discrepancy is not yet clear. However, it is noteworthy that primary fibroblasts show a mesenchymal phenotype with less expression of E-cadherin, compared with the levels in the epithelial cell-like MDCK cells (23). This suggests that the basal mechanisms for cell adhesion and motility differ between these cell types. Despite the discrepancy in cell morphology, the expression of N-cadherin and vinculin was induced by overexpression of CRT in both fibroblasts (59) and MDCK cells (this study), indicating that the present result does not necessarily contradict previous findings using CRT-overexpressing fibroblasts, in respect of the expression of N-cadherin and vinculin (58,59). Very recently, Afshar et al. (60) reported that cellular migration and binding to collagen type V were apparently suppressed in embryonic fibroblasts from CRT knock-out mice, indicating that the cellular level of CRT is important for the regulation of cell motility. Cell surface expression of CRT has been reported in several cell types (3). In addition, Goicoechea et al. (61) reported that cell surface CRT functions as a receptor for thrombospondin to contribute to focal-adhesion disassembly of the cell. Although it is not clear how cell surface CRT contributes to thrombospondin-mediated signaling and regulation for cell migration, no significant expression of CRT was observed on the surface of CRToverexpressing MDCK cells. Taken together, these findings suggest that CRT plays a vital role in the regulation of cell-cell interaction and cell motility in a variety of cell types.
E-cadherin appears to be an important target molecule in the mechanism influenced by CRT. To investigate the mechanism whereby CRT up-regulates the expression of Slug to repress the E-cadherin gene, we focused on Ca 2ϩ homeostasis influenced by CRT overexpression in MDCK cells, because CRT plays an important role in the regulation of Ca 2ϩ homeostasis in the ER (1). We found that thapsigargin-sensitive intracellular Ca 2ϩ stores in the ER were apparently increased in CRT-overexpressing cells, compared with controls. This was consistent with previous reports concerning the intracellular Ca 2ϩ stores in cells overexpressing CRT (13,30,62,63). On the other hand, it has been reported that CRT modulates store-operated Ca 2ϩ influx (1,30,64). In previous studies, decreased Ca 2ϩ influx was observed in stable CRT-overexpressing cells (13,30,63), but not in cells transiently transfected with the CRT gene (65). In the present study, when the cells in Ca 2ϩ -free medium were stimulated with thapsigargin, the level of [Ca 2ϩ ] i was apparently elevated more in CRT-overexpressing cells than controls. Then store-operated Ca 2ϩ influx was examined by re-adding Ca 2ϩ to the external medium, and the results showed that the [Ca 2ϩ ] i was significantly elevated in CRT-overexpressing cells compared with controls. This was unexpected and inconsistent with previous findings of the suppressive effect of CRT overexpression on store-operated Ca 2ϩ influx (13,30,63). To explain why the Ca 2ϩ influx was up-regulated by overexpressed CRT, the expression of several epithelial Ca 2ϩ channels (i.e. TRPV5, TRPV6, and Polycystin 2) was examined in control and CRToverexpressing cells. The results showed that, in CRT-overexpressing cells, the expression of TRPV5 was apparently up-regulated and the basal level of Ca 2ϩ influx from external medium was also increased. TRPV5 and TRPV6 are highly Ca 2ϩ -selective epithelial channels in the TRP cation channel superfamily, and they function to (re) absorb Ca 2ϩ in the epithelial cells of specific organs, such as kidney, intestine, testis, esophagus, ileum, and colon (31). The causative relation between overexpressed TRPV5 and increased Ca 2ϩ influx seen in CRT-overexpressing cells seemed to be consistent with previous findings in MDCK cells stably transfected with the TRPV5 gene (66). Recently, Pigozzi et al. (67) has reported that a change in Ca 2ϩ stores in the ER influenced the gene regulation of some TRP channels, including TRPC1, TRPC3, and TRPV6 in a prostate cancer cell line, suggesting that the expression of these TRP channels may be regulated by alteration of cellular Ca 2ϩ homeostasis. However, in the present study, it is not clear how the expression of TRPV5 is specifically up-regulated in the CRToverexpressing MDCK cells. Taken together, these results suggest that stable overexpression of CRT in MDCK cells may cause an imbalance of Ca 2ϩ homeostasis, and this leads to the up-regulation of TRPV5, resulting in the increase in the basal level of Ca 2ϩ uptake. On the other hand, in previous reports, levels of [Ca 2ϩ ] i in the resting state were not significantly different between control and CRT-overexpressing cells (13,30,63). Thus, the up-regulation of the [Ca 2ϩ ] i state in CRT-overexpressing MDCK cells seems to be unusual compared with other cell types, and further investigation is required to clarify the mechanism involved.
To our knowledge, whether the expression of Slug is controlled in a Ca 2ϩ -mediated manner has not been investigated. In this study, we showed that, in MDCK cells, the transcription of Slug was up-regulated by thapsigargin or ionomycin, which increased [Ca 2ϩ ] i (Fig. 9), leading to an increase in intensity of the gel-shift band of the protein-DNA complex of E-box B in the E-cadherin promoter (Fig. 11A). These results were consistent with the finding that E-cadherin expression was suppressed through the up-regulation of Slug expression in MDCK cells by treatment with thapsigargin or ionomycin. Conversely, in CRToverexpressing cells treated with BAPTA-AM, the expression of Slug was suppressed and the expression of E-cadherin was increased. This indicates that the altered regulation of E-cadherin expression through Slug was dependent on an increased [Ca 2ϩ ] i in CRT-overexpressing cells. Collectively, these results strongly support the notion that overexpression of CRT suppresses E-cadherin expression by up-regulating Slug expression FIGURE 11. A, MDCK-Control cells were treated with or without thapsigargin (Tg, 5 M) or ionomycin (Io, 1 M) for 2 h, and nuclear extracts were prepared. The protein-binding activity of E-boxes was then examined by EMSA as described under "Experimental Procedures" using radiolabeled nucleotide probes for E-boxes A-C in the canine E-cadherin gene promoter. Arrowheads indicate shifted bands of protein complex with 32 P-labeled probes. Data represent three independent experiments. B, MDCK-CRT1 cells were treated with BAPTA-AM (10 M) for 2 h, then nuclear extracts were prepared. The proteinbinding activity of E-boxes was then examined by EMSA as described under "Experimental Procedures" using radiolabeled nucleotide probes for E-boxes A-C. Data represent three independent experiments. through the increased level of Ca 2ϩ . Zhao et al. (68) reported that the Slug gene is a downstream target of the transcription factor MyoD, and is up-regulated by MyoD through the E-boxes in its promoter. Furthermore, a Ca 2ϩ -responsive E-box element, CaRE2, was identified to function in the transcription of the brain-derived neurotrophic factor gene in neurons (69), suggesting a possible regulation of the Ca 2ϩ -induced up-regulation of E-box elements in the Slug gene. In terms of these findings, further investigation is required to clarify how the Slug gene is up-regulated by the increased [Ca 2ϩ ] i in CRToverexpressing cells.
In terms of the relationship between CRT and cancer, proteomic analysis has revealed a new functional role of CRT for the early diagnosis of cancers. CRT is proposed as a new tumor marker of bladder cancer (70) and autoantibodies to CRT isoforms have utility for the early diagnosis of pancreatic cancer (71). These reactions are not indicative of malignant properties of CRT but, rather, are markers of immunogenicity and anticancer responses (72,73). On the other hand, another report demonstrated that CRT is overexpressed in the nuclear matrix in hepatocellular carcinoma, compared with normal liver tissue, suggesting a relationship between overexpressed CRT and malignant transformation (74). In this study, we found that overexpression of CRT conferred migratory ability on epithelial-like MDCK cells. To gain invasive ability is a characteristic of malignancy, although the increase of invasiveness induced by CRT does not directly indicate that CRT is an oncogenic factor. MDCK cells transfected with an expression vector for the Snail gene, a zinc finger transcription factor like Slug, gained invasive and angiogenic properties through the repression of the E-cadherin gene and behaved like carcinomas (75). However, cell growth of Snail-overexpressing MDCK cells is less than that of control cells. These results are compatible with our present findings. In addition, a recent report indicated that Snail overexpression in MDCK cells up-regulated the matrix metalloproteinase-9 (MMP-9) gene, suggesting some contribution to cell invasiveness (76). Taken together, these findings suggest that CRT has a potential function to increase cell invasion by suppressing E-cadherin expression through zinc finger transcription factors like Slug, resulting in a contribution to the establishment of invasive characteristics in malignancies. To test this hypothesis, further investigations, including studies in vivo, will be required.
In conclusion, in the present study, we demonstrated a novel function of CRT in cell adhesion and motility of the epithelium. The enhanced cell invasiveness mediated through E-cadherin gene repression was regulated by the gene repressor, Slug, via altered Ca 2ϩ homeostasis caused by overexpression of CRT in epithelial MDCK cells.