CLCA Splicing Isoform Associated with Adhesion through β1-Integrin and Its Scaffolding Protein

Background: CLCA family members have been shown to be multifunctional proteins. Results: A rat splicing isoform was localized to the stratum basale of the epidermis. This isoform was found to alter β1-integrin expression and cell adhesion. Conclusion: This isoform may modulate the adhesive potential of undifferentiated epithelial cells. Significance: Cell-specific splicing might confer a novel function on the epithelium. We previously found that a rat CLCA homologue (rCLCA-f) modulates Ca2+-dependent Cl− transport in the ductal cells of the rat submandibular gland. CLCA proteins have been shown to be multifunctional, with roles in, for example, cell adhesion. Here, we describe the mRNA and protein expressions of a splicing isoform of rat rCLCA (rCLCA-t). This isoform is a 514-amino acid protein containing a C-terminal 59-amino acid that is distinct from the rCLCA-f sequence. Immunohistochemistry revealed rCLCA-t to be located in the basal cells of the rat submandibular gland excretory duct and the stratum basale of rat epidermis, whereas rCLCA-f was detected in cells during the process of differentiation. In a heterologous expression system, rCLCA-t was found to be a membrane protein present predominantly in the perinuclear region, and not to be either present on the cell surface or secreted. rCLCA-t failed to enhance ionomycin-induced Cl− conductance (unlike rCLCA-f). When compared with rCLCA-f, it weakened cell attachment to a greater extent and in a manner that was evidently modulated by intracellular Ca2+, protein kinase C, and β1-integrin. rCLCA-t was found to associate with RACK1 (receptor for activated C kinase) and to reduce expression of mature β1-integrin. Treatment of rat skin with rCLCA-t siRNA increased the expression of β1-integrin in the stratum basale of the epidermis. These results are consistent with cell-specific splicing of rCLCA mRNA playing a role in the modulation of the adhesive potential of undifferentiated epithelial cells.

ular candidates were put forward for Ca 2ϩ -activated Cl Ϫ channels, including CLCA (3-6), CLC-3 (7), and bestrophin (8), but it has now been identified as TMEM16A (9). The first reported CLCA clone, from bovine tracheal epithelium, was bCLCA1 (10), and this prototype was followed by the cloning of other members of the CLCA family. We previously identified a rat CLCA homologue (rCLCA), 2 and provided evidence suggesting that it is responsible for modulating Ca 2ϩ -dependent Cl Ϫ transport in salivary ductal cells (11). More recently, we presented in vivo evidence of a physiological role for rCLCA in transepithelial Cl Ϫ transport in the ductal system of rat submandibular gland (SMG) (12).
Genetic studies have indicated links between the CLCA family and certain secretory disorders (viz. cystic fibrosis and asthma) (13,14). Actually, CLCA proteins have been shown to be multifunctional, with roles in, for example, cell-cell adhesion, apoptosis, and cell-cycle control, and they appear to have a pathophysiological connection with tumor metastases (see Ref. 5 for review). First, a bovine isoform of CLCA (bCLCA2; Lu-ECAM-1) was reported to be an endothelial adhesion molecule (15)(16)(17), and then human CLCA2 (hCLCA2) and mouse CLCA1 (mCLCA1), which also possess specific binding motifs, were shown to bind ␤ 4 -integrin and to contribute to tumor metastases (18,19).
Each of these CLCA proteins consists of ϳ900 amino acids, and there are also reported to be various truncated isoforms, the physiological roles of which are largely unknown. The first truncated CLCA isoform to be described was derived from bCLCA2 (17). Among the four identified clones, the truncated version was predicted to be a secreted protein, and the authors supposed this to be an alternative splicing product from a single gene. In the case of hCLCA3, which is also reported to be secreted (20), the gene encodes the first one-third of the full-length hCLCA3 protein due to the presence of a stop codon. In mammary adenocarcinoma cell lines, mCLCAs are reportedly down-regulated when their splicing variants are expressed (21). Although the existence of these isoforms has been documented, it remains uncertain whether such splicing isoforms have physiological significance.
To identify another CLCA member present in rat tissue, using total RNA extracted from the ileum, we employed an RT-PCR-based strategy aided by primers based on the cDNA sequence conserved among mouse and rat members. Here, we report a truncated isoform of rCLCA (rCLCA-t) that we found to have characteristics distinct from those of the full-length isoform (rCLCA-f). Our results demonstrate that rCLCA-t exhibits a remarkable distribution among the undifferentiated epithelial cells of SMG and skin. Epidermal homeostasis has been proposed to be mediated by ␤ 1 -integrin, which plays regulatory roles in anchorage to the extracellular matrix and the onset of terminal differentiation (22). The present study suggests (a) that heterologous rCLCA-t expression may cause an attenuation of ␤ 1 -integrin-mediated cell adhesion, with no alteration in Cl Ϫ conductance, in human embryonic kidney 293 (HEK293) cells, and (b) that isoform-specific knockdown increases ␤ 1 -integrin expression in the basal layer of the skin. These results lead us to hypothesize that this cell-specific CLCA isoform might help to confer novel functions, such as modulation of adhesive potential, on undifferentiated epithelium.

Identification of rCLCA-t and Generation of Stable Expression Cell Lines-Male
Wistar rats (6 weeks old) were anesthetized with sodium pentobarbital, permission for the procedures used having been granted by the Animal Research Committee of Fukuoka Dental College. Total RNA was isolated from rat ileum, and then the reverse-transcribed cDNA was subjected to PCR, utilizing PfuUltra DNA polymerase (Stratagene, La Jolla, CA) and primers previously employed for the PCR cloning of rCLCA-f (11). To confirm the existence of the contiguous cDNA comprising these PCR products, we amplified the entire ORF for rCLCA-t along with part of its 5Ј-and 3Ј-untranslated regions. The PCR product was cloned into a pCR-XL-TOPO vector (Invitrogen).
For the transient expression study, the cDNA was subcloned into mammalian expression vector pIRES-hrGFP-1a (Stratagene). Cell lines exhibiting stable, tetracycline (TC)-inducible expression were also generated for histidine-tagged rCLCA-f and rCLCA-t cDNA using a Flp-In T-REx system together with Flp-In-293 cells (Invitrogen), the parental cell line of which is HEK293 cells. Fifteen nucleotides were inserted 76 bp downstream of the start codon in either construct to produce a histidine tag (one intrinsic His plus five additional His, see Fig. 1). The cells were treated with TC (1 g/ml) for 24 h prior to the experiments.
Immunoblotting Analysis-Rat SMG and skin samples were frozen and crushed in liquid nitrogen. The tissues and transfected cells were homogenized in lysis buffer (150 mM NaCl, 1 mM EGTA, 1 mM MgCl 2 , and 20 mM Tris-Cl, pH ϭ 7.4) in the presence of protease inhibitor mixture (Sigma). The lysate was centrifuged for 10 min to obtain the 1,000 ϫ g supernatant (Sup), and then the Sup was centrifuged for 15 min to collect 6,500 ϫ g Sup. Centrifugation at 100,000 or 200,000 ϫ g for 2 h separated the 6,500 ϫ g Sup into the cytosolic fraction and the membrane pellet. All procedures were performed at 4°C.
For isolation of ␤ 1 -integrin on the cell surface, cells were labeled with Sulfo-NHS-SS-Biotin (Pierce). The cells were lysed in the presence of 1% Triton X-100, and the cell lysate was applied to NeutrAvidin-agarose beads and incubated for 60 min. The flow-through fraction was collected for the detection of nonsurface proteins. After the column had been washed, the Sulfo-NHS-bound surface protein was eluted by incubation with dithiothreitol (DTT).
To condense the histidine-tagged rCLCA proteins released into the culture medium, cells grown in a 75-cm 2 flask were incubated with 5 ml of serum-free medium for 5 h. The collected medium was concentrated using centrifugal filters (cutoff ϭ 30 kDa; Microcon YM-30; Millipore Corp.) followed by incubation with cobalt-affinity resins (BD TALON; BD Biosciences). The bound protein was eluted with 200 mM imidazole buffer. All procedures were performed at 4°C.
Cell Adhesion Assay-Cells were dispersed and incubated in Ca 2ϩ -containing HEPES-buffered physiological salt solution (PSS). The cells (0.5-1.0 ϫ 10 6 cells/well) were allowed 20 min to adhere to a collagen-coated plate (24 wells) during incubation at 37°C in the presence of several reagents followed by a gentle wash three times with low Ca 2ϩ (10 M)-containing HEPES-PSS. To quantify adherence, the attached cells were incubated with tetrazolium salt for 2 h at 37°C (Cell Counting Kit-8; Dojindo, Kumamoto, Japan). Then, the absorbance of the supernatant at 450 nm was measured. To obtain a control value, the absorbance of the supernatant derived from nontreated cells was measured without the washing procedure.
Epidermal Application of siRNA-A small interference RNA (siRNA) to target the rCLCA-t gene was synthesized in a purified and annealed duplex form (GeneDesign, Inc., Osaka, Japan). The sense sequence of the annealed siRNA duplex form was 5Ј-CGGAUAUGACAGUUGGAGAdTdT-3Ј. Scrambled RNA controls were also designed. The specificity of the siRNA was tested in HEK293 cells that stably expressed the CLCA isoform following the use of a transfection reagent (RNAiFect, Qiagen). siRNA or the scrambled RNA (200 g) was incorporated into 2 mg of biodegradable cationized gelatin microspheres (CGM) (MedGel E50; MedGEL Corp., Kyoto, Japan).
Each type of RNA-incorporated microspheres was suspended in saline and injected subcutaneously into both earlobes in 7-week-old rats. Such incorporated compounds have been reported to be released from the hydrogel after its degradation by endogenous enzymes secreted from the surrounding tissues (23). One week after the injection, the rats were sacrificed, both ears were removed, and 7-m-thick frozen sections were examined in an immunofluorescence study using the same concentrations of antibodies as those shown in the previous section. ImageJ was used for quantifying the intensity of the fluorescence.
Statistics-All values are presented as means Ϯ S.E. (n ϭ number of observations). Statistical analysis was performed using a one-way ANOVA or multivariate ANOVA followed by a post hoc Bonferroni's t test. A grouped t test was employed when two groups were to be compared. A p value less than 0.05 was considered to be statistically significant.

Molecular Cloning of a Truncated Mutant of Rat CLCA-
Previously, we identified a rat CLCA member, rCLCA, from rat ileum (903 amino acids) (11) (rCLCA-f in Fig. 1 and in supplemental Fig. S1; named rCLCA2 in the present GenBank TM / European Molecular Biology Laboratory (EMBL) nomenclature). The same procedure was employed here to find a new isoform using primers designed for PCR based on the cDNA sequence that is identical among mCLCA1, -2, and -4. Among the nine positive clones obtained from the amplified PCR product, six clones (2.1 kb in length) were almost identical to the central part of rCLCA-f cDNA (GenBank/EMBL accession number AB119249). The remaining clones (named rCLCA-t) were shorter (2.0 kb), and but for a missing 101 bp in the middle of the sequence, they were also identical to rCLCA-f.
We amplified the entire ORF for rCLCA-t along with part of its 5Ј-and 3Ј-untranslated regions. rCLCA-t was found to consist of 3,208 bases, of which 20 bases comprise the 5Ј-untranslated region. Because the absence of 101 bp in the rCLCA-t mRNA sequence (between nucleotides 1,367 and 1,467 of rCLCA-f mRNA) causes a frameshift, a stop codon (UAA) appears in the middle of the entire length of the mRNA (Fig.  1A). Because the rCLCA-t isoform was found to possess 1,646 bases in the 3Ј-untranslated region followed by a poly(A) ϩ tail, the ORF was predicted to be shorter, encoding a 514-amino acid protein ( Fig. 1B and supplemental Fig. S1). This protein possessed an N-terminal signal sequence, but lacked the significant C-terminal hydrophobic domain reported to be present in rCLCA-f (11). The amino acid sequence in the C-terminal region downstream of the frameshift (59 amino acids) was entirely different from that of any isoform previously reported for CLCA members (supplemental Fig. S1, italic letters). Excluding this region, this isoform was 84, 72, and 56% identical to mCLCA1, Lu-ECAM-1, and hCLCA1, respectively (supplemental Fig. S1).
We performed a BLAST search using a rat genomic sequence (24), with rCLCA-f and rCLCA-t cDNAs as the query sequences (25). The locus of the rCLCA gene appears to be 2q44, and the region was found to be identical to the predicted mRNA sequence XM_575054. On the basis of a comparison between the rat genome and rCLCA-f mRNA, the rCLCA gene was predicted to consist of at least 14 exons interspersed with 13 introns and to span 22,293 bp from the start to the end of the cDNA. Based on the above argument, the ninth exon appeared to be lost when rCLCA-t mRNA was transcribed and spliced ( Fig. 1A). A cDNA sequence homologous to rCLCA-t was found in the mouse cDNA database (26; accession number BC010260), although it has not been characterized further.
Localization of Truncated rCLCA Isoform Specifically in Undifferentiated Cells of Rat SMG and Skin-We tested in RT-PCR and immunoblotting experiments whether rCLCA isoforms are expressed in rat SMG and skin, using specific primers and antibodies (Fig. 1A). Firstly, we confirmed that the primers recognized each isoform in a heterologous expression system. Amplification of rCLCA-t cDNA at the predicted nucleotide size of 427 bp was detected (using the P-T pair of primers) specifically in transiently rCLCA-t-transfected cells (Fig. 1B, T). Using these primers, no expression of rCLCA-t mRNA was detected in rat SMG, whereas marked expression was found in rat skin. In contrast, the P-F pair amplified rCLCA-f cDNA at the nucleotide size of 455 bp in the rCLCA-f-transfected cells (Fig. 1B, F), but produced only a small amount of pseudo-products in T. Using these primers, an abundant expression of rCLCA-f mRNA was observed in both SMG and skin.
Rat SMG and skin were lysed and analyzed by immunoblot using CT1 antibody (Fig. 1C). Its specificity between the two rCLCA isoforms is shown in the supplemental text and supplemental Fig. S2. This antibody recognized a 60-kDa-sized band in the rCLCA-t-transfected cells. This corresponded to the size of rCLCA-t protein (Fig. 1A). CT1 antibody failed to detect an apparent 60-kDa signal in rat SMG lysate or its immunoprecipitant. This antibody, however, strongly recognized the band in total lysate and in 200,000 ϫ g membrane pellet and Sup fractions obtained from rat skin.
To examine whether the truncated isoform is distributed only in specific regions in rat SMG and skin, immunohistochemistry was performed. Using CF1 antibody, intense staining of rCLCA was observed in the ductal epithelia of the intralobular ducts in SMG ( Fig. 2A). CF1 antibody that had been preabsorbed with the antigen peptide did not stain the ductal cells (Fig. 2B). CT1 antibody failed to stain epithelial cells in the acinar and intralobular ducts (Fig. 2C). Interestingly, immunoreactivity to CT1 antibody was found in a few rounded or pyramidal cells at the base of the epithelium in SMG excretory ducts ( Fig. 2D and inset). rCLCA-t was largely limited to the perinuclear region, with little surface distribution. The preabsorbed CF1 antibody did not stain such cells (Fig. 2E). CK14 is known to be a specific marker of undifferentiated basal cells (27), and a selective immunoreactivity to an anti-CK14 antibody confirmed that such cells were undifferentiated (Fig. 2F).
In skin, CT1 antibody predominantly stained the stratum basale, which is a single layer of undifferentiated epithelial cells (basal cells) (Fig. 3A). The preabsorbed CT1 antibody did not stain the stratum basale (Fig. 3B). Anti-CK14 antibody recognized basal cells and also suprabasal cells, probably due to their slow turnover (Fig. 3C). In contrast, CF1 antibody stained suprabasal layers during the process of differentiation: most intensively in the stratum spinosum and weakly in the stratum granule, which were recognized by anti-CK10 (Fig. 3, D and F). Little staining was seen when preabsorbed CF1 antibody was used (Fig. 3E). Interestingly, CT1 staining was very intense and spread across the basal as well as suprabasal layers of the skin wounds (Fig. 3, G and H).
Heterologous Expression of rCLCA Isoforms in the Membrane Fraction of HEK293 Cells-We investigated the subcellular distribution and post-translational processing of rCLCA isoforms by immunoblotting using a CN1 antibody (Fig. 4A). HEK293 transfected transiently with the empty vector did not exhibit immunoreactivity to CN1 antibody (Fig. 4A, lane V). In the rCLCA-t-transfected cells (Fig. 4A, lane T), the antibody recognized a 60-kDa-sized band in 1,000 ϫ g Sup. In the 100,000 ϫ g membrane pellet fractions, substantial immunoreactivity was detected in the Triton X-100-insoluble subfraction, but there Recognition sites for the three rCLCA antibodies (Ab) are shown (Y). The proposed post-translational processing site is indicated by a closed triangle. aa, amino acids. B, mRNA expression was examined in rat SMG and skin and in HEK293 cells transfected with either rCLCA-t (T) or rCLCA-f (F) cDNA. P-T and P-F denote primer pairs designed to recognize rCLCA-t and rCLCA-f, respectively (see panel A). C, expressions of rCLCA proteins from the lysate of rat SMG (100 g) and rCLCA-t-transfected HEK293 cell (25 g). The immunoprecipitant (IP) was obtained from the SMG lysate (300 g for the left lane and 450 g for the right lane) using CT1 antibody. The expression was also examined for the lysate, 1,000 ϫ g, 6,500 ϫ g, and 200,000 ϫ g pellet (Ppt), and 200,000 ϫ g Sup obtained from rat skin. Immunodetection was performed using CT1 antibody. IB, immunoblot.
was much less in the soluble subfraction, suggesting weak solubility of this protein in the detergent. There was no immunoreactivity in the cytosolic fraction.
In contrast, in the rCLCA-f-transfected cells (Fig. 4A, lane F), the antibody recognized 86-and 120-kDa-sized bands in the 1,000 ϫ g Sup, and the immunoreactivities were almost equal between the Triton X-100-soluble and -insoluble membrane fractions, indicating a relatively easy solubilization of this isoform. Immunoreactivity was lowest in the cytosolic fraction. The 86-and 120-kDa-sized rCLCA-f bands corresponded to the N-terminal part and the entire length of the protein, respectively (Fig. 1A), the former probably being produced by posttranslational proteolysis (closed triangle). This modification was not detected for the rCLCA-t protein.
Then, we examined whether rCLCA isoforms might be expressed heterologously on the plasma membrane by investigating the TC-inducible expression of rCLCA-t and -f in stably transfected cells (sT and sF cells). Surface expression was assessed by detecting the biotinylation of membrane proteins on these cells. Streptavidin-HRP failed to recognize the  . Immunostaining in rat skin. A, immunoreactivity to CT1 antibody is shown in basal cell layers of skin. B, CT1 antibody that had been preabsorbed with the antigen peptide (pep) did not stain the basal cells. C, anti-CK14 antibody mainly recognized undifferentiated layers (see "Results"). D, CF1 antibody stained more differentiated cell layers. E, CF1 antibody that had been preabsorbed with the antigen peptide did not stain the undifferentiated layers. F, anti-CK10 antibody recognized the stratum spinosum and stratum granule. G and H, rat skin wounds. In G, unwounded (UW) and wounded (W) epithelia can be seen. Hair follicles were also stained. B, stratum basale; S, stratum spinosum; G, stratum granule; C, stratum corneum. Counterstaining: hematoxylin. Bars ϭ 100 m. FEBRUARY 15, 2013 • VOLUME 288 • NUMBER 7 60-kDa-sized, CT1 antibody-immunoreactive protein in the lysate, although it did recognize the 86-kDa-sized, CF1 antibody-immunoreactive protein (Fig. 4B). Next, rCLCA-t protein was immunoprecipitated from the cell lysate using CN1 antibody. The immunoreactivity of the 60-kDa protein was confirmed by immunoblotting using CT1 antibody. Streptavidin again failed to detect this immunoprecipitated protein, but did recognize the 86-kDa rCLCA-f protein immunoprecipitated using CN1 antibody (Fig. 4B). These results suggest that rCLCA-t is unlikely to be a surface protein, whereas the N-terminal 86-kDa portion (modified post-translationally) of rCLCA-f is a surface protein.
Several members of the CLCA protein family have been shown to be secreted into the extracellular environment (28 -30). Because the amino acid sequence of rCLCA-t lacks half of the C-terminal part of the full-length CLCA, it seemed possible that this isoform is secreted from cells too. However, in cells stably transfected with histidine-tagged rCLCA-t cDNA, immunoreactivity for a 60-kDa band was found in the cell lysate, but not in the elution from the medium (Fig. 4D). This implies that rCLCA-t is unlikely to be a secreted protein. In the histidine-tagged rCLCA-f-transfected cells, 120-and 86-kDa bands were immunoreactive in the cell lysate, but only the 86-kDa band was detectable in the medium (Fig. 4D). This suggests that the post-translationally modified N-terminal part, but not the unmodified rCLCA, may be secreted.
Next, we performed immunocytochemistry using permeabilized, stably transfected cells. CT1 antibody predominantly recognized the TC-induced expression of rCLCA-t in the perinuclear area of the transfected cells, with no staining being evident near the cell membrane. In contrast, rCLCA-f appeared to be expressed in more peripheral areas of the cell than rCLCA-t, with intense linear labeling on the cell-to-cell interface, probably indicating its presence on the plasma membrane (Fig. 4E,  arrowheads).
rCLCA-t Does Not Alter Ca 2ϩ -induced Cl Ϫ Conductance-We examined whether rCLCA-t might exhibit an electrophysiological function similar to that reported for rCLCA-f: namely, an ability to increase Ca 2ϩ -induced Cl Ϫ conductance (11). Fig.  5, A and B, show typical current traces recorded during step pulses or voltage ramps before and after ionomycin (2 M) was added (to cause a drastic increase in the intracellular Ca 2ϩ concentration). In the Western blot study, there was estimated to be a similar amount of immunoreactivity per sT or sF cell between rCLCA-t and surface rCLCA-f (86 kDa), suggesting that the expression levels of these isoforms are almost the same (Fig. 5C). The summarized data (Fig. 5D) show that, in the absence of TC (ϪTC), the background current was negligible in the sT cells, whereas a small increase could be observed in sF cells, suggesting a leaky expression of the rCLCA protein under the present experimental conditions. After TC treatment FIGURE 4. Western blotting and immunocytochemistry using transfected HEK293 cells. A, the 1,000 ϫ g and 100,000 ϫ g Sup and the 1% Triton X-100-soluble (Sol) and -insoluble (Insol) membrane pellet (Ppt) were obtained from transiently transfected cells (F, rCLCA-f; T, rCLCA-t; V, vehicle). Protein loading was 10 g per lane for 1,000 ϫ g and 100,000 ϫ g Sup, and 50 g per lane for soluble and insoluble fractions. * denotes specific bands at 120 and 86 kDa for rCLCA-f, and 60 kDa for rCLCA-t. IB, immunoblotting. B, cells subjected to stable transfection with either rCLCA-t (sT) or rCLCA-f (sF) were surfacebiotinylated, lysed, and immunoprecipitated using CN1 antibody. The biotinylated proteins in the cell lysate and its immunoprecipitate (IP) were detected using streptavidin (SA)-HRP (*). Immunoreactive bands were detected using CT1 or CF1 antibody (*). C, peptide-N-glycosidase F (PNGase)-induced migration of rCLCA-f and rCLCA-t bands to lower molecular weights. Closed triangle, glycosylated. Open triangle, nonglycosylated. D, culture medium taken from the transfected cells and total cell lysate were subjected to the affinity resins, and the bound CLCA was eluted and applied to Western blotting using CT1 or CF1 antibody. E, immunocytochemistry using anti-CLCA antibodies in permeabilized sT or sF cells. Arrowheads indicate intense staining of rCLCA-f on the plasma membrane. Counterstaining: hematoxylin. Bar ϭ 50 m.
(ϩTC), the current was not altered during voltage steps or ramps in sT cells (Fig. 5, A, B, and D, upper), but it was substantially enhanced in sF cells (Fig. 5, A, B, and D, lower). This lack of such an electrophysiological function for rCLCA-t is supported by the above evidence that there is no surface expression of this isoform.
rCLCA-t Decreases Integrin-dependent Cell Attachment in Response to Increased Intracellular Ca 2ϩ -During patch clamp experiments, we noticed that sT cells were easily detached from collagen-coated coverglasses. We made a quantitative comparison between cells transfected stably with one of the rCLCA isoforms and mock-transfected cells for their adherence to collagen-coated plates (24-well) in the presence or absence of either ionomycin or thapsigargin (TG), each of which increases intracellular Ca 2ϩ (Fig. 6). In the mock and sF cells, adhesion capability tended to decrease to ϳ75-85% and ϳ65-80% of control, respectively, when either ionomycin (2 M) or thapsigargin (5 M) was added, depending on the period of drug application. In sT cells, the reduction in adhesion capability induced by either drug was greater than in the above two cell types regardless of the period of drug application (Fig. 6).
We then evaluated cell adherence using another protocol; in this one, the cells were suspended in ionomycin-containing PSS immediately before plating. A 20-min incubation with ionomycin substantially impaired adhesion capability in all groups ( Fig.  7A; columns with horizontal hatching), although it was suppressed most strongly in sT cells. This reduction was partially reversed by subsequent additional (for 10 or 20 min) incubation with EGTA to chelate external Ca 2ϩ and to avoid Ca 2ϩ influx into the cells. In all groups, adhesion capability was also restored by incubation with phorbol-12, 13-dibutyrate (PDBu, 100 nM), which activates protein kinase C (Fig. 7A).
HEK293 cells have been shown to express ␤ 1 -integrin proteins (31,32), most of which form receptors for collagen (33). We therefore speculated that rCLCA-t might alter the expression of endogenous ␤ 1 -integrin protein. To test whether activation of ␤ 1 -integrin can rescue the adherence of sT cells (Fig.  7B), we examined the effect of ionomycin in cells pretreated with an activating anti-␤ 1 -integrin antibody (TS2/16) (34). The pretreated cells were exposed to PSS with or without ionomycin (2 M, 20 min, 37°C) at 1 to 20 min after the beginning of the adhesion (37°C). In the absence of the activating antibody, Intensity of Western blot obtained using CN1 antibody was quantified (inset), and the relative values in a single cell were calculated and normalized to that of the 86-kDa band observed in sF cells. n ϭ 4. D, summary of difference current density (n ϭ 4 -6) evoked by voltage steps (300 ms) from Ϫ50 mV to Ϫ80 mV or 80 mV. Current density was calculated by dividing current amplitude by membrane capacitance. Ionomycin-induced current density was obtained by subtracting the density before ionomycin perfusion from that during the perfusion. *, p Ͻ 0.05 versus corresponding value in sT cells (Student's or Welch's t test). pF, picofarads. FEBRUARY 15, 2013 • VOLUME 288 • NUMBER 7 ionomycin reduced adhesion to 50% of control in sT cells, but only to 90 -94% of control in the mock and sF cells. Pretreatment with TS2/16 (2.5 g/ml, 20 min, room temperature) partially restored adhesion capability (73% of control) in sT cells (Fig. 7B).

Splicing Isoform of Rat CLCA
Typically, ␤ 1 -integrin protein is first synthesized (87 kDa) and then core-glycosylated to generate the so-called "premature form" (110 kDa), which dimerizes with an ␣ subunit in the endoplasmic reticulum. The dimer is then transported to the Golgi apparatus where it matures to a 130-kDa form that expresses complex N-linked glycan (35,36).
Cell lysate was obtained from mock, sT, and sF cells that had been pretreated with either ionomycin or PDBu for 20 min. Western blots obtained using anti-␤ 1 -integrin antibody revealed that ionomycin decreased the expression of mature ␤ 1 -integrin (130 and 110 kDa) in all three types of cells, whereas PDBu tended to increase it. The inhibitory action of ionomycin was most marked in sT cells (Fig. 7C).
rCLCA-t Alters the Subcellular Localization of the Mature Form of ␤ 1 -Integrin-We examined the localization of endogenous ␤ 1 -integrin in immunofluorescence experiments (Fig.  8A). In mock and sF cells, ␤ 1 -integrin was located near the surface membrane (arrowheads) and in the perinuclear region. In contrast, ␤ 1 -integrin became localized mainly to the perinuclear site after transfection of rCLCA-t (arrows).
Next, mock and sF cells were incubated with ionomycin (2 M, 5-20 min, 37°C). The drastic increase in the intracellular Ca 2ϩ concentration did not alter the localization of ␤ 1 -integrin in these cells (Fig. 8A). In sT cells, the ␤ 1 -integrin distribution obtained during ionomycin treatment was much less marked in surface membranes, but more marked in perinuclear regions (arrows). Some of the other cells exhibited strikingly distinct morphology, the nucleus having become rounded and shrunken.
The activating anti-␤ 1 -integrin antibody TS2/16 (5 g/ml, 30 min, 4°C) increased the peripheral ring-like immunostaining of ␤ 1 -integrin in sT and sF cells in the presence or absence of ionomycin, indicating a tighter adhesion to the  collagen-coated coverglass (Fig. 8B, arrows). Occasionally, ionomycin still caused ␤ 1 -integrin to locate around the nucleus of some sT cells.
The surface and subcellular expressions of the glycosylated forms of ␤ 1 -integrin were examined by means of surface biotinylation techniques. The 130-kDa ␤ 1 -integrin was found to be surface-biotinylated (Fig. 9A), indicating that a mature, fully glycosylated type (␤ 1 ) is localized on the cell surface. Stable transfection with rCLCA-t tended slightly to attenuate the intensity of the 130-kDa band in the total cell lysate, and it significantly decreased (by 20%) the expression of the surfacebiotinylated 130-kDa form (Fig. 9B).
In contrast, the 110-kDa premature ␤ 1 -integrin was found to be present intracellularly (Fig. 9A). The intensity of the 110-kDa band in the total lysate and intracellular unbiotinylated form tended to be increased by transfection with rCLCA-t and to be inhibited by rCLCA-f (Fig. 9B). Immunofluorescence obtained using anti-calnexin antibody revealed that ␤ 1 -integrin colocalized extensively with calnexin, a marker of endoplasmic reticulum, in sT cells (Fig. 9C).
Among proteins known to interact with ␤ 1 -integrin and to promote cell adhesion, we focused on a scaffolding intracellular protein, RACK1, a receptor for activated C kinase that is reported to be modified by PKC and intracellular Ca 2ϩ (37). RACK1 was mostly localized in intracellular regions, but some was found to be associated with biotinylated proteins on surface membranes (Fig. 9A). When compared with that in mock or sF cells, the surface expression of RACK1 was lower in sT cells, in which the surface localization of ␤ 1 -integrin was attenuated (Fig. 9A).
Combining Triton X-100 (1%) and sodium deoxycholate (0.1%) was found to produce a detergent that dissolved both RACK1 and rCLCA-t together (supplemental Fig. S3). Under the Triton X-100 ϩ sodium deoxycholate condition, RACK1 was successfully co-immunoprecipitated by CT1 antibody, but not by CF1 antibody. Reciprocally, rCLCA-t, but not rCLCA-f, was co-immunoprecipitated by anti-RACK1 antibody (Fig. 9D). Any association of ␤ 1 -integrin with either rCLCA isoform was barely detectable, although an interaction was confirmed with ␣ 2 -integrin (Fig. 9D), which is known to associate with a ␤ 1 -integrin to form a receptor for collagen (33). These results suggest the existence of a specific interaction between RACK1 and rCLCA-t in the membrane fraction.
Knockdown of rCLCA-t Alters Expression of ␤ 1 -Integrin in Skin-To examine the influence of rCLCA-t over the expression of ␤ 1 -integrin in skin, an siRNA was generated for isoformspecific knockdown of rCLCA-t (see supplemental Fig. S4). Transfection of an siRNA together with CGM has been reported to have a long-lasting effect on the expression of the target molecule (23). In control (scrambled RNA ϩ CGMtreated) skin, a confocal microscopic study showed that rCLCA-t displayed an intracellular localization. Surface expression of ␤ 1 -integrin was detected in cells in the basal layers (Fig.  10A). In siRNA ϩ CGM-treated skin, expression of ␤ 1 -integrin was significantly augmented in the area within which rCLCA-t localization was depressed (Fig. 10B). When compared with the control skin, siRNA ϩ CGM-treated skin exhibited a weaker intensity rCLCA-t band (CT1 antibody) in Western blots, but a greater ␤ 1 -integrin intensity (Fig. 10C). In rat skin, RACK1 was co-immunoprecipitated by CT1 antibody, and rCLCA-t was Splicing Isoform of Rat CLCA FEBRUARY 15, 2013 • VOLUME 288 • NUMBER 7 reciprocally co-immunoprecipitated by anti-RACK1 antibody (Fig. 10D). Immunofluorescence reveals that RACK1 extensively colocalized with rCLCA-t, but not with ␤ 1 -integrin (Fig. 10E).

DISCUSSION
In the present study, we report a novel splicing isoform of rCLCA. Comparison of its cDNA sequence with the rat genome suggests that it is likely to be an exon 9 splicing variant. We initially supposed that rCLCA-t might be a secreted protein because it possessed an N-terminal signal sequence and had no significant hydrophobic domain in the other portion. However, the present results indicate that rCLCA-t is unlikely to be translocated to the plasma membrane or secreted into the culture medium. Therefore, such a limited subcellular localization of the splicing variant suggests that it might perform a cellular function distinct from that of the full-length isoform, rCLCA-f. Indeed, we were unable to detect any electrophysiological activity of this isoform.
An additional striking finding was that the distribution of rCLCA-t differed from that of rCLCA-f in the epithelial cells of rat SMG and skin. Immunohistochemistry clearly demonstrated a predominant perinuclear expression of rCLCA-t in undifferentiated basal cells and a reciprocal expression of rCLCA-f in differentiated epithelial cells. In addition, dense expression of rCLCA-t was found broadly across the epithelial cell layers of wounded skin, where reproduction of the epidermis takes place. These observations raise the possibility that the truncated isoform is localized specifically to subserve a function specific to epithelial cells with an undifferentiation status. Previously, reciprocal patterns of expression of truncated isoforms in epidermis have been reported for other genes closely related to cell differentiation and proliferation. For example, the p63 gene encodes two isoforms, the full TAp63 and a truncated⌬Np63 (38), consistent with p63 performing dual roles (39,40). Likewise, the reciprocal expression of the rCLCA isoforms reported here could indicate roles in the regulation of cell growth and differentiation in salivary glands and epidermis.
Our results reveal an inhibitory effect of rCLCA-t on cell adhesion, suggesting reduced expression or recruitment of adhesion molecules to the cell membrane. Interestingly, we found that rCLCA-t altered both the maturation of ␤ 1 -integrin and its subcellular distribution; specifically, rCLCA-t decreased the surface expression of the fully glycosylated form of ␤ 1 -integrin. It is reasonable to speculate that rCLCA-t may associate directly with ␤ 1 -integrin. There are published examples of expressions of certain membrane proteins (vacuolar H ϩ -ATPase subunit and low density lipoprotein receptor-related protein-1) being accompanied by shifts both in the degree of maturation and in the subcellular localization of ␤ 1 -integrin (36,41). Previously, CLCA family members have been reported to possess a von Willebrand A domain, which is known to be involved in the cell adhesion mediated by ␣and ␤-integrins (42). Because we were unable to support this idea by co-immunoprecipitation analysis, it is possible that rCLCA-t affects integrin functions indirectly, through an adaptor protein.  4 -6). The intensities of the immunoblot bands obtained from sT and sF cells were normalized to that obtained from mock cells. **, p Ͻ 0.01 versus mock cells (multivariate ANOVA followed by a post hoc Bonferroni's t test). C, immunofluorescence study showing that ␤ 1 -integrin (TS2/16) colocalized with calnexin (CNX), a marker of endoplasmic reticulum, in sT cells. Bar ϭ 10 m. D, interaction of rCLCA-t with RACK1. rCLCA isoforms, RACK1, and ␤ 1 -integrin were immunoprecipitated by the corresponding antibody using the Triton X-100 plus sodium deoxycholate-soluble fractions obtained from mock, sT, or sF cells. The co-immunoprecipitant (IP) was subjected to immunoblotting (IB). FIGURE 10. Modulation of ␤ 1 -integrin expression by rCLCA-t in rat skin. A, changes in rCLCA-t and ␤ 1 -integrin in the stratum basale of rat earlobe following epidermal application of siRNA. Immunofluorescence labeling was detected for rCLCA-t and ␤ 1 -integrin in the stratum basale following epidermal application of siRNA. Green, CT1 antibody. Red, anti-␤ 1 -integrin antibody (HM␤1-1). Bar ϭ 20 m. DIC, differential interference contrast. B, densitometric analysis of expressions of rCLCA-t and ␤ 1 -integrin. The intensities of the immunofluorescence measured within the basal layer at the injection site were normalized to that measured at a noninjected site. n ϭ 13-14. **, p Ͻ 0.01, ***, p Ͻ 0.001 versus the control RNA-injected earlobe (Student's t test). C, Western blot of 6,500 ϫ g Sup obtained from rat earlobe at 1 week after injection into the earlobe of microspheres incorporating either rCLCA-t siRNA or control (cont) RNA. CT1, anti-RACK1, and anti-␤-actin antibodies were used. Densitometric analysis of the blot is shown. n ϭ 4. D, interaction of rCLCA-t with RACK1 in rat skin. rCLCA-t and RACK1 were immunoprecipitated by the corresponding antibodies. The co-immunoprecipitant was subjected to immunoblotting (CT1 and anti-RACK1 antibodies). Representative result of three experiments. E, immunofluorescence labeling for rCLCA-t (CT1), RACK1, and ␤ 1 -integrin in rat skin, showing that rCLCA-t colocalized with RACK1, but not with ␤ 1 -integrin (HM␤1-1). Bar ϭ 20 m. FIGURE 11. Model illustrating how ␤ 1 -integrin expression may be attenuated by expression of rCLCA-t. The scaffolding protein RACK1 is presumed to facilitate the surface expression of the mature form of ␤ 1 -integrin (left). rCLCA-t interacts with RACK1 intracellularly (presumably within the Triton X-100insoluble membrane fraction). This interaction is likely to inhibit recruitment of the mature form to the surface membrane and thereby to decrease the adhesive potential of the cell (right).
Proteins that have been shown to interact with ␤-integrin cytoplasmic domains include actinin, paxillin, talin, focal adhesion kinase, integrin-linked kinase-1, and so on. We found that an increase in intracellular Ca 2ϩ (induced by ionomycin or thapsigargin) led to a reduction in the adhesion of rCLCAt-transfected cells, which PDBu and TS2/16 partly restored. Therefore, phosphorylation/dephosphorylation of some adaptor or scaffolding proteins interacting with rCLCA-t is likely to be important for its regulation of integrins. RACK1, a member of the G␤ superfamily, is known to bind to activated PKC (43), to interact with the cytoplasmic tail of ␤ 1 -integrin, and to promote cell adhesion (37,44). Activation of PKC by phorbol esters and the resultant phosphorylation of RACK1 is reported to be necessary for the association of RACK1 with ␤ 1 -integrin (37). In addition, knockdown of RACK1 by siRNA has been shown to attenuate cell adhesion to collagen (45). In contrast, a Ca 2ϩ /calmodulin-dependent protein phosphatase, calcineurin, has been shown to dephosphorylate RACK1 and to block RACK1-mediated functions (46). Previously, it was shown that signaling proteins, such as ␤ 1 -integrin and phosphatase 2A, compete for binding to a common site on RACK1 and that the binding preference is switched according to phosphorylation status (47). Our findings of (a) a specific association between rCLCA-t and RACK1 and (b) a reduced surface expression of both ␤ 1 -integrin and its scaffolding protein, RACK1, indicate a possible mechanism by which rCLCA-t might inhibit recruitment of integrins to RACK1 and so impair their adhesive capability (Fig. 11). Through such an interaction between these molecules, an inhibitory effect of rCLCA-t on the expression of ␤ 1 -integrin in focal adhesion could result in a reduced resistance to the cell detachment induced by compounds that increase intracellular Ca 2ϩ .
We found that an isoform-specific knockdown of rCLCA-t increased the expression of ␤ 1 -integrin in the basal layers of rat skin. Because most ␤ 1 -integrins adhere to the extracellular matrix, these are likely to provide a physical support within tissues and also to regulate cell function (33). Reportedly, integrins, including ␤ 1 -integrin, function not just as direct mediators of adhesion, but also as regulators of epidermal stratification, differentiation, and morphogenesis (48). Signal transduction through ␤ 1 -integrin represents an instruction to keratinocytes not to differentiate (22). The above results point to the interesting hypothesis that a causal relationship may exist between rCLCA-t expression and the integrin-mediated cell adhesion and differentiation of native SMG and skin. Intriguingly, a previous study pointed out a close association between CLCA and basal-cell adhesion (a human isoform, hCLCA2, being shown to be colocalized with ␤ 4 -integrin in the basal cells of human stratified epithelia) (49).
Overall, the findings made in the present study suggest that the truncated isoform of rCLCA is likely to be expressed preferentially in undifferentiated epithelial cells. We also demonstrated that the isoform may play a role in the modulation of the adhesive potential of undifferentiated epithelial cells. It would be interesting to investigate further the role of the cell-specific rCLCA isoform in regulating the differentiation of epithelial cells.