The Recruitment of Phosphatidylinositol 3-Kinase to the E-cadherin-Catenin Complex at the Plasma Membrane Is Required for Calcium-induced Phospholipase C-γ1 Activation and Human Keratinocyte Differentiation*

Calcium induces epidermal keratinocyte differentiation, but the mechanism is not completely understood. We have previously demonstrated that calcium-induced human keratinocyte differentiation requires an intracellular calcium rise caused by phosphatidylinositol 3-kinase (PI3K)-dependent activation of phospholipase C-γ1. In this study we sought to identify the upstream signaling pathway necessary for calcium activation of PI3K and its subsequent activation of phospholipase C-γ1. We found that calcium induces the recruitment of PI3K to the E-cadherin-catenin complex at the plasma membrane of human keratinocytes. Knocking-down E-cadherin, β-catenin, or p120-catenin expression blocked calcium activation of PI3K and phospholipase C-γ1 and calcium-induced keratinocyte differentiation. However, knocking-down γ-catenin expression had no effect. Calcium-induced PI3K recruitment to E-cadherin stabilized by p120-catenin at the plasma membrane requires β-catenin but not γ-catenin. These data indicate that the recruitment of PI3K to the E-cadherin/β-catenin/p120-catenin complex via β-catenin at the plasma membrane is required for calcium-induced phospholipase C-γ1 activation and, ultimately, keratinocyte differentiation.

The normal epidermis undergoing a well defined program of terminal keratinocyte differentiation displays a characteristic calcium gradient from a low level of calcium ions in the stratum basale to a progressive increase with the level of calcium ions reaching its maximal density in the outer stratum granulosum (1). This calcium gradient is important for maintaining normal epidermal differentiation. Loss of the calcium gradient as a consequence of epidermal barrier disruption leads to increased proliferation and decreased differentiation (2). A disturbed calcium gradient has been observed in psoriatic epidermis in vivo favoring increased proliferation and decreased differentiation (3). In vitro, many of normal keratinocyte differentiation-specific processes can be reproduced by culturing cells in media containing high calcium concentrations. The addition of calcium to these cells in culture induces a rapid and homogeneous response, triggering a differentiation program that closely resembles that of differentiating keratinocytes in vivo (4 -7). However, very little is known about early molecular events triggering keratinocyte differentiation. Calcium-induced keratinocyte differentiation is accompanied by induction of phospholipase C (PLC) 2 levels, increased phosphoinositide metabolism, and subsequently, increased inositol 1,4,5-triphosphate (IP 3 ) production and intracellular calcium concentration (8 -11). PLC catalyzes the hydrolysis of phosphatidylinositol 4,5bisphosphate (PIP 2 ) to IP 3 and diacylglycerol. These molecules regulate the mobilization of intracellular calcium and protein kinase C activation, respectively (12,13). To date, four types of phosphoinositide-specific PLC isozymes have been identified: PLC-␤, PLC-␥ , PLC-␦, and PLC-⑀. Each type contains several subtypes (14). We have previously demonstrated that PLC-␥1 is required for the extracellular calcium-induced intracellular calcium rise and differentiation of human keratinocytes (15). PLC-␥1 activation and human keratinocyte differentiation induced by calcium require phosphatidylinositol 3-kinase (PI3K) activation (16). PI3K converts PIP 2 to phosphatidylinositol 3,4,5-triphosphate (PIP 3 ), which binds to the N-terminal PH domain (17) and the C-terminal SH 2 domain (18,19) of PLC-␥1 to activate PLC-␥1 (16). However, the mechanism by which calcium activates PI3K responsible for PLC-␥1-mediated human keratinocyte differentiation remains unclear.
A number of studies indicate that PI3K is recruited in epithelial cell types by E-cadherin to cell-cell contacts to approach its plasma membrane substrate, PIP 2 (20 -24). The recruitment of PI3K to the plasma membrane is one of the mechanisms by which the kinase is activated (25). The cadherin family is a group of glycoproteins that forms homophilic calcium-dependent binding between adjacent cells. E-cadherin is the most abundant member of classical cadherins in epidermal keratinocytes and is found in all layers of the epidermis (26). E-cadherin function is not limited to cell-cell adhesion in the epidermis; E-cadherin can also participate in signaling events between and within cells that * This work was supported by NIAMS, National Institutes of Health Grant P01AR39448. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1  affect epidermal keratinocyte differentiation. Blocking E-cadherin function using E-cadherin antibody blocks calcium-induced phosphorylation of Akt, which is a downstream target of PI3K and, therefore, blocks calcium-induced differentiation of mouse keratinocytes (24). During calcium-induced mouse keratinocyte differentiation, PI3K increasingly associates with the cadherin-catenin protein complex (24).
Mice lacking E-cadherin in the epidermis manifest a loss of adherens junctions and reduced epidermal differentiation (27,28). In keratinocyte-derived squamous cell carcinoma cells, reduced expression of E-cadherin is associated with reduced expression of differentiation markers (29 -33).
Inactivating mutations in the E-cadherin gene have been found in patients with epithelial cell carcinomas (34,35). These observations suggest that keratinocyte differentiation requires E-cadherin. Epidermal keratinocytes also contain multiple types of catenins. ␤and ␥-catenin bind in a mutually exclusive way to the catenin binding domain of the E-cadherin cytoplasmic tail. The N-terminal portion of both ␤and ␥-catenin interacts with ␣-catenin, which links the E-cadherin to the underlying actin cytoskeleton (36). P120-catenin, another member of the catenin family, was originally described as a substrate for Src-and receptor-tyrosine kinases (37,38) and later was identified as a catenin. p120-catenin also binds to the cadherin cytoplasmic tail, but p120-catenin binds to the juxtamembrane domain rather than the distal region where ␤and ␥-catenin bind (39 -43). Unlike ␤and ␥-catenin, p120-catenin does not interact with ␣-catenin (44). As has been extensively reported for E-cadherin, p120-catenin expression is also frequently reduced to very low or undetectable levels in the major human carcinoma types (45), including the squamous cell carcinoma with poor differentiation (46). In the normal human epidermis, p120catenin is expressed faintly in the stratum basale but intensely in the stratum granulosum and finally disappears in the stratum corneum (46), which correlates with the epidermal differentiation. In our present study we sought to examine the potential contribution of E-cadherin/catenin-activated PI3K to the PLC-␥1 activation by calcium and determine whether this signaling pathway is required for calcium-induced human keratinocyte differentiation.

EXPERIMENTAL PROCEDURES
Cell Culture-Human keratinocytes were isolated from neonatal human foreskins as described previously (47). Briefly, human keratinocytes were isolated from newborn human foreskins by trypsinization (0.25% trypsin, 4°C, overnight), and primary cultures were established in serum-free medium (medium 154CF with human keratinocyte growth supplement, Cascade Biologics, Portland, OR) containing 0.07 mM calcium. The second-passage human keratinocytes were plated with the serum-free medium containing 0.03 mM calcium and used in the subsequent experiments. To initiate differentiation, the calcium concentration was raised to 1.2 mM.
Cell Lysate Preparation, Western Analysis, and Coimmunoprecipitation-Keratinocyte total lysate was prepared using 2% SDS. Briefly, human keratinocytes in cell culture plates were washed twice with PBS and then incubated in PBS containing 2% SDS, Complete TM protease inhibitors (Roche Applied Science), and 4-(2-aminoethyl)benzenesulfonyl fluoride (AEBSF, EMD Biosciences, San Diego, CA) for 1 min. Cells were scraped into microcentrifuge tubes, incubated at 4°C for 30 min, and pelleted by centrifugation. The supernatant was collected for the determination of protein concentration and Western analysis. Keratinocyte membrane lysate was prepared using the Mem-PER eukaryotic membrane protein extraction reagent kit (Pierce). The protein concentration of the lysate was measured by the bicinchoninic acid (BCA) protein assay kit (Pierce). Equal amounts of protein were electrophoresed through reducing SDS-PAGE and electroblotted onto polyvinylidene fluoride microporous membranes (Immobilon-P, 0.45 M, Millipore). After incubation in blocking buffer (100 mM Tris base, 150 mM NaCl, 5% nonfat milk, and 0.5% Tween 20), the blot was incubated overnight at 4°C with appropriate primary antibodies: polyclonal antibody against human E-cadherin, p120-catenin, or transglutaminase-1 (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) at a dilution of 1:200, monoclonal antibody against human ␤or ␥-catenin (Santa Cruz) at a dilution of 1:200, polyclonal antibody against Keratin 1 (K1) and keratin 5 (K5) (Covance Research Products, Inc., Denver, PA) at 1:10,000, monoclonal antibodies against human involucrin (Sigma Aldrich) at a dilution of 1:2,000, polyclonal antibody against human Akt or phosphorylated Akt (Cell Signaling Technology, Inc.) at a dilution of 1:1,000, polyclonal antibody against human p85␣ (Santa Cruz), which is the regulatory subunit of class Ia PI3K at a dilution of 1:200, monoclonal antibody against human integrin ␣2 (plasma membrane marker), BIP (endoplasmic reticulum marker), or GM130 (cis-Golgi marker) at a dilution of 1:250 (BD Bioscience, San Jose, CA). After incubation in the blocking buffer, the membranes were incubated for 1 h with the appropriate anti-IgG secondary antibody conjugated to horseradish peroxidase (Amersham Biosciences) in the blocking buffer. After a second series of washes, bound antibody complexes were visualized using the SuperSignal ULTRA chemiluminescent kit (Pierce). To analyze protein complex formation at the plasma membrane by coimmunoprecipitation, equal amounts of membrane protein (500 g) extracted with Mem-PER eukaryotic membrane protein extraction reagent kit (Pierce) were incubated with 2 g of polyclonal antibody against human E-cadherin for 1 h at room temperature or overnight at 4°C and then with 20 l of Ultra-Link immobilized protein G (Pierce) for 1 h at 4°C. The lysateantibody-agarose bead mixture was washed four times with PBS and then analyzed by Western analysis with antibodies against PI3K-p85␣ as described above. In a reverse approach, membrane lysates were analyzed by co-immunoprecipitation with PI3K-p85␣ antibodies followed by Western blotting with E-cadherin antibodies.
PI3K Activity Assay-The activity of PI3K was determined by the PI3K enzyme-linked immunosorbent assay kit (Echelon Research Laboratories, Salt Lake City, UT) that detects the formation of PIP 3 from PIP 2 . Briefly, cells in 100-mm dishes were washed 3 times with ice-cold buffer A (137 mM NaCl, 20 mM Tris-HCl, pH 7.4, 1 mM CaCl 2 , 1 mM MgCl 2 , and Halt phosphatase inhibitor mixture from Pierce) and then incubated with ice-cold lysis buffer (buffer A plus 1% Nonidet P-40, Complete TM protease inhibitors, and 4-(2-aminoethyl)benzenesulfonyl fluoride) for 20 min on ice. Cells were scraped into microcentrifuge tubes and centrifuged for 10 min to sediment insoluble material. PI3K in the supernatant was immunoprecipitated from the supernatant containing 500 g of protein using 5 l of polyclonal antibodies against PI3K-p85 (Upstate, Charlottesville, VA) for 1 h at 4°C and then 20 l of UltraLink immobilized protein G for 1 h at 4°C. After a series of washes, the conjugated beads were then assayed for PI3K activity according to the manufacturer's protocol. The activity of PI3K was normalized to the protein content in the immunoprecipitates and expressed as percentages of the normalized activity in the presence of control siRNA and 0.03 mM calcium.
PLC-␥1 Activity Assay-PLC-␥1 activity was determined by measuring accumulation of IP 3 according to the experimental procedure described (16,48). Cells in 100-mm dishes were washed with PBS containing Halt phosphatase inhibitor mixture and 4-(2-aminoethyl)benzenesulfonyl fluoride and then incubated with PBS containing 1% Nonidet P-40 (Nonidet P-40), Halt phosphatase inhibitor mixture, and Complete TM protease inhibitors for 5 min. Cells were scraped into microcentrifuge tubes and incubated at 4°C on a rotator for 1 h. PLC-␥1 was immunoprecipitated from the supernatant using 2 g of polyclonal PLC-␥1 antibodies (Santa Cruz) for 1 h at 4°C and then 20 l of UltraLink immobilized protein G for 1 h at 4°C.

Calcium Induces PI3K Recruitment to the E-cadherin-Catenin Complex at the Plasma Membrane-We have previously
shown that the activation of PLC-␥1 by calcium requires PI3K in human keratinocytes (16). PI3K catalyzes the local production of PIP 3 (49), which activates PLC-␥1 (16,18). PI3K forms a complex with E-cadherin after calcium stimulation (24). Accordingly, we sought to test whether calcium induces PI3K recruitment to the E-cadherin-catenin complex at the plasma membrane via its p85 subunit. This would place PI3K near its substrate. To address this issue, the plasma membrane of human keratinocytes treated with 1.2 mM calcium for 2 min was isolated, and the complex formation was analyzed by immunoprecipitation with antibodies against PI3K-p85␣ followed by Western analysis with antibodies against E-cadherin, ␤-catenin, ␥-catenin, and p120-catenin. The results show that calcium induces PI3K-p85␣ to form a complex with E-cadherin, ␤-catenin, ␥-catenin, or p120-catenin at the plasma membrane (Fig. 1A). These results were confirmed by a reverse approach using antibodies against E-cadherin for immunoprecipitation FIGURE 1. Calcium induces PI3K-p85␣ recruitment to the E-cadherincatenin complex at the plasma membrane of human keratinocytes. Cultured human keratinocytes were treated with 1.2 mM calcium for 2 min. The cells were harvested, and the plasma membrane and total cell lysates were extracted. The lysates were analyzed by immunoprecipitation (IP) with antibodies against PI3K-p85␣ followed by Western analysis with E-cadherin, ␤-catenin, ␥-catenin, or p120-catenin (A). In a reverse approach, detection of PI3K-p85␣, ␤-catenin, ␥-catenin, or p120-catenin in the E-cadherin immunoprecipitates was accomplished by Western analysis with the appropriate antibodies. Rabbit IgG and protein G beads alone were included in the co-immunoprecipitation assay and used as negative controls (B). The protein levels of E-cadherin, ␤-catenin, ␥-catenin, or p120-catenin, PI3K-p85␣, integrin ␣2 (plasma membrane marker), BIP (endoplasmic reticulum marker), and GM130 (cis-Golgi marker) in total cell lysates and plasma membrane lysates were determined by Western analysis (C). and then antibodies against PI3K-p85␣, ␤-catenin, ␥-catenin, or p120-catenin for Western analysis (Fig. 1B). Rabbit IgG and protein G beads alone were included in the co-immunoprecipitation assay, showing that there was no background with these negative controls (Fig. 1, A and B). Two minutes of exposure of cells to 1.2 mM calcium raised levels of PI3K-p85␣, E-cadherin, ␤-catenin, ␥-catenin, and p120-catenin at the plasma membrane but not in the total cell lysate (Fig. 1C). In contrast, other divalent cations (5 mM magnesium, 5 mM barium, 5 mM strontium, and 0.1 mM zinc) had no effect (data not shown). Western analysis with antibody against the plasma membrane marker integrin ␣2, endoplasmic reticulum maker BIP, or cis-Golgi maker GM130 confirmed that only the integrin ␣2 antibody immunoreacted with the plasma membrane lysate, confirming the purity of the plasma membrane lysate (Fig. 1C). These data indicate that calcium induces the formation of the E-cadherincatenin complex and its recruitment of PI3K-p85␣ to the plasma membrane.
␤-Catenin Mediates PI3K-p85␣ Recruitment to E-cadherin at the Plasma Membrane-Among E-cadherin-associated catenin members, ␤-catenin has been shown to bind directly to PI3K-p85␣ (22, 50) but not ␥-catenin or p120catenin. To determine which catenin is required for E-cadherin complex formation with PI3K in human keratinocytes, the expression of ␤-catenin, ␥-catenin, or p120-catenin in human keratinocytes was knocked down by siRNAs. Cells were harvested after stimulation with 1.2 mM calcium for 2 min. Total cell lysates were isolated, and the protein levels for E-cadherin, ␤-catenin, ␥-catenin, p120-catenin, and integrin ␣2 were determined by Western analysis. In parallel experiments membrane lysates were isolated. The complex was analyzed by co-immunoprecipitation with an antibody against E-cadherin followed by Western blotting with an antibody against PI3K-p85␣ or immunoprecipitation with an antibody against PI3K-p85␣ followed by Western blotting with an antibody against E-cadherin. The results show that the siRNA for ␤-catenin, ␥-catenin, or p120-catenin specifically inhibited more than 90% of the protein expression of ␤-catenin, ␥-catenin, or p120-catenin, respectively (Fig. 2, A  and B). Although the basal and calcium-stimulated levels of E-cadherin were reduced by the p120-catenin knockdown (Fig. 2, A and B) due to the disrupted stabilization function of p120-catenin at the plasma membrane (42,51), the E-cadherin level was not affected by ␤-catenin knockdown. The ␥-catenin level was not affected by either p120 or ␥-catenin knockdown. The integrin ␣2 level in the lysates was used as a control, showing equal loading in each lane of the blot (Fig. 2,  A and B). Inhibition of ␤-catenin expression blocked calcium-induced PI3K-p85␣ recruitment to E-cadherin at the plasma membrane, but inhibition of ␥-catenin expression did not affect this recruitment (Fig. 2C). Inhibition of p120 catenin shows no binding between E-cadherin and PI3K-p85␣ due to the reduced E-cadherin level in the plasma membrane lysate. These data suggest that ␤-catenin, but not ␥-catenin, is required for PI3K-p85␣ recruitment to E-cadherin at the plasma membrane, whereas p120-catenin is critical for the stability of the entire complex.
E-cadherin, ␤-Catenin, and p120-Catenin Are Required for PI3K Activation by Calcium-We have previously demonstrated that calcium-induced human keratinocyte differentiation requires PI3K-mediated PLC-␥1 activation (16). We then wanted to know whether E-cadherin and catenins are the upstream signaling components necessary for calcium-activated PI3K. To answer this question, human keratinocytes were treated with siRNA for E-cadherin, ␤-catenin, ␥-catenin, or p120-catenin for 3 days followed by 1.2 mM calcium for 1 h. The total cell lysates were immunoprecipitated by an antibody against PI3K-p85. The PI3K activity was then determined by measuring the conversion from PIP 2 to PIP 3 and normalized to the protein level of PI3K-p85 in the immunoprecipitate. The results show that knocking down E-cadherin, ␤-catenin, or p120-catenin expression blocked the calcium activation of PI3K but knocking down ␥-catenin expression had no effect (Fig. 3, A-C). Western analysis demonstrated that the siRNA for E-cadherin, ␥-catenin, or p120-catenin reduced the protein levels of E-cadherin, ␥-catenin, or p120-catenin more than 90% in the same experiment (Fig. 3, D-F). The threonine phosphorylation of Akt, a downstream effect after PI3K activation, was assayed by Western analysis. The results show that stimulation of Akt phosphorylation by calcium was blocked by knocking down E-cadherin, ␤-catenin, or p120-catenin but not by knocking down ␥-catenin (Fig. 3, D-F). The level of total Akt was not affected by the siRNA treatments. These data indicate that E-cadherin, ␤-catenin, and p120-catenin, but not ␥-catenin, are required for the activation of PI3K by calcium. ␤-Catenin mediates PI3K-p85␣ recruitment to E-cadherin at the plasma membrane of human keratinocytes. Cultured human keratinocytes were transfected with siRNA for ␤-catenin, ␥-catenin, or p120-catenin. Transfected cells were incubated for 72 h at 37°C with 5% CO 2 . Cells were harvested after stimulation with 1.2 mM calcium for 2 min. Total cell lysates were isolated, and the protein levels for ␤-catenin, ␥-catenin, p120-catenin, and integrin ␣2 were determined by Western analysis (A and B). In some experiments membrane lysates were isolated. The complex formation was analyzed by immunoprecipitation (IP) with an E-cadherin antibody followed by Western blotting (WB) with a PI3K-p85␣ antibody or immunoprecipitation with a PI3K-p85␣ antibody followed by Western blotting with an E-cadherin antibody (C).

E-cadherin, ␤-Catenin, and p120-catenin Are Required for PLC-␥1 Activation by Calcium, and Calcium-induced PLC-␥1
Membrane Localization Is Dependent on PI3K Activity-To determine the role of E-cadherin, ␤-catenin, ␥-catenin, and p120-catenin in mediating calcium activation of PLC-␥1, cultured human keratinocytes were transfected with siRNA for E-cadherin, ␤-catenin, ␥-catenin, or p120-catenin. Transfected cells were incubated for 3 days. Cells were harvested after stimulation with 1.2 mM calcium for 1 h. The cell lysates were immunoprecipitated with a PLC-␥1 antibody and incubated with PIP 2 . PLC-␥1 activity was determined by measuring the IP 3 produced. The results show that the induction of PLC-␥1 activity by calcium was blocked by knocking down E-cadherin, ␤-catenin, or p120-catenin but not by knocking down ␥-catenin (Fig. 4A). These data indicate that E-cadherin, ␤-catenin, and p120-catenin, but not ␥-catenin, are required for the activation of PLC-␥1 by calcium.
We have previously demonstrated that calcium-induced PLC-␥1 activation requires PIP 3 produced by PI3K. To determine whether PI3K is required for PLC-␥1 recruitment to the plasma membrane, cultured human keratinocytes were preincubated with the PI3K inhibitor LY294002 and then treated with 1.2 mM calcium for 1 h. The level of PLC-␥1 in the plasma membrane lysates was determined. The results show that 1.2 mM calcium raised the levels PLC-␥1 at the plasma membrane but not in the total cell lysate (Fig.  4B). Western analysis with antibody against the plasma membrane marker integrin ␣2, endoplasmic reticulum maker BIP, or cis-Golgi maker GM130 confirmed that only the integrin ␣2 antibody immunoreacted with the plasma membrane lysate, confirming the purity of the plasma membrane lysate (Fig. 4B). These results indicate that PI3K activity is required for PLC-␥1 recruitment to the plasma membrane by calcium, presumably by its production of PIP 3 to which PLC-␥1 is bound and activated.
E-cadherin, ␤-Catenin, and p120catenin Are Required for Calcium-induced Human Keratinocyte Differentiation-To determine whether E-cadherin, ␤-catenin, ␥-catenin, and p120-catenin are required for calcium-induced human keratinocyte differentiation, human keratinocytes were pretreated by the siRNA for E-cadherin, ␤-catenin, ␥-catenin, or p120-catenin for 3 days before exposing to 1.2 mM calcium for 24 h. E-cadherin, ␤-catenin, ␥-catenin, p120-catenin, and keratinocyte differentiation markers were determined by Western analysis. The results show that calcium treatment increased the protein expression of E-cadherin, ␤-catenin, ␥-catenin, and p120-catenin within 24 h (Fig. 5, A-C). The siRNA for E-cadherin, ␤-catenin, ␥-catenin, or p120-catenin selectively reduced E-cadherin, ␤-catenin, ␥-catenin, or p120-catenin protein levels more than 90% in human keratinocytes (Fig. 5, A-C). The level of ␥-catenin was not changed even when the E-cadherin complex was disrupted as a consequence of E-cadherin knockdown, presumably due to the stabilization of ␥-catenin by the desmosomal cadherin complex. The basal and especially the calcium-stimulated levels of ␤-catenin and p120-catenin were reduced as a consequence of E-cadherin complex disruption, although the reduction in p120-catenin was not as marked as that of ␤-catenin (Fig. 5A). These observations are consistent with previous studies showing that p120-catenin is more stable than ␤-catenin in the absence of E-cadherin (28,52). Similarly, the basal and calciumstimulated levels of E-cadherin and ␤-catenin were reduced by the p120-catenin knockdown (Fig. 5B), as expected from the fact that the E-cadherin complex is stabilized by p120-catenin at the plasma membrane (42,51). However, ␥-catenin was not affected by p120-catenin knockdown (Fig. 5B). Unlike p120catenin, knocking down ␤or ␥-catenin did not affect the levels of E-cadherin possibly because of the compensation by ␥-catenin and ␤-catenin, respectively (Fig. 5, B and C). Indeed, ␤-catenin levels were increased by knocking down ␥-catenin perhaps FIGURE 3. E-cadherin, ␤-catenin, and p120-catenin are required for calcium-induced PI3K activation. Cultured human keratinocytes were transfected with siRNA for E-cadherin, ␤-catenin, ␥-catenin, or p120catenin. Transfected cells were incubated for 72 h at 37°C, 5% CO 2 . Cells were harvested after stimulation with 1.2 mM calcium for 1 h. The class Ia PI3K in the cell lysates was immunoprecipitated by antibodies against the PI3K regulatory subunit p85, and PI3K activity was determined by analyzing PIP 3 production generated from PIP 2 using enzyme-linked immunosorbent assay as described under "Experimental Procedures." The level of PI3K-p85 in the p85 immunoprecipitate was determined by Western analysis (A-C). The threonine phosphorylation of Akt (p-Akt), and the total Akt was determined by Western analysis. The protein levels for E-cadherin, ␤-catenin, ␥-catenin, and p120-catenin were determined for knocking down efficiency (D-F).

DISCUSSION
We have previously identified a pathway required for calcium-induced human keratinocyte differentiation. This pathway is mediated by PI3K-dependent activation of PLC-␥1, which induces an intracellular calcium rise (15,16). Extracellular calcium activates PI3K, which phosphorylates PIP 2 to generate PIP 3 at the plasma membrane. PIP 3 in turn activates PLC-␥1, which cleaves PIP 2 into diacylglycerol and IP 3 . Diacylglycerol remains in the inner layer of the plasma membrane. IP 3 diffuses through the cytosol and binds to the IP 3 receptors in the endoplasmic reticulum and Golgi causing the intracellular calcium rise, which triggers keratinocyte differentiation (15,16,55). In cultured human or mouse keratinocytes activities of PI3K and PLC-␥1 are induced by calcium within minutes and further stimulated up to 1 h before the onset of keratinocyte differentiation. Inhibition of this activation blocks calcium-induced keratinocyte differentiation (16,24). In our present study we have explored the earlier molecular events required for PI3K recruitment to the E-cadherin complex and its activation. We have found that in human keratinocytes extracellular calcium induces an increase in the concentration of E-cadherin and PI3K and their association with the E-cadherin complex at the plasma membrane. Knocking down E-cadherin blocks calcium activation of PI3K and PLC-␥1 and, ultimately, keratinocyte differentiation, indicating that the plasma membrane-associated E-cadherin acts as a primary driver for the recruitment and activation of PI3K. Indeed, translocation to the membrane alone is sufficient to activate PI3K (25). The activated PI3K converts PIP 2 to PIP 3 , which then activates PLC-␥1 at the plasma membrane to trigger keratinocyte differentiation via increasing intracellular calcium (15,16). Our data FIGURE 4. E-cadherin and ␤and p120-catenin are required for calcium-induced PLC-␥1 activation, and calcium-induced PLC-␥1 membrane localization is dependent on PI3K activity. A, cultured human keratinocytes were transfected with siRNA for E-cadherin, ␤-catenin, ␥-catenin, or p120-catenin. Transfected cells were incubated for 72 h at 37°C, 5% CO 2 . Cells were harvested after stimulation with 1.2 mM calcium for 1 h. The PLC-␥1 in the cell lysates was immunoprecipitated with a PLC-␥1 antibody, and the PLC-␥1 activity was assayed as described under "Experimental Procedures." The level of PLC-␥1 in the PLC-␥1 immunoprecipitate was determined by Western analysis. B, cultured human keratinocytes were preincubated with 10 M LY294002 or vehicle Me 2 SO for 1 h and then treated with 1.2 mM calcium for 1 h. The cells were harvested, and the plasma membrane and total cell lysates were extracted. The protein levels of PLC-␥1, integrin ␣2 (plasma membrane marker), BIP (endoplasmic reticulum marker), and GM130 (cis-Golgi marker) in plasma membrane lysates and total cell lysates were determined by Western analysis. also indicate that calcium-activated Akt requires E-cadherin. PIP 3 generated by E-cadherin-activated PI3K brings phosphoinositide-dependent kinase 1 (PDK1) and Akt to the plasma membrane and facilitates phosphorylation of Akt by PDK1. This phosphorylation stimulates the catalytic activity of Akt (56), which protects keratinocytes from apoptosis (24). The anti-apoptotic pathway is likely to extend the life span of keratinocytes and allow differentiation triggered by the increased intracellular calcium to proceed.
Both ␤and ␥-catenin bind to the same intracellular domain of E-cadherin in a mutually exclusive fashion. Both catenins can compensate for the function of each other in cell adhesion (57). Among catenin members, only ␤-catenin has been reported to have a direct association with PI3K-p85␣ (22,50). Our results show that calcium-induced PI3K recruitment and activation, PLC-␥1 activation, and keratinocyte differentiation require ␤-catenin but not ␥-catenin. It is likely that recruitment and activation of PI3K by E-cadherin in the plasma membrane is via ␤-catenin. ␤-catenin may serve as a bridge between E-cadherin and PI3K. In the absence of ␤-catenin, the bridge is lost, and PI3K is unable to be associated with E-cadherin in the plasma membrane even though the membrane-bound E-cadherin/␥-catenin/ p120-catenin complex apparently is still intact as indicated by the lack of effect of ␤-catenin knockdown on E-cadherin, ␥-catenin, and p120-catenin levels in the cell. ␥-Catenin cannot compensate for defects in PI3K signaling resulting from ␤-catenin knockdown, although elevated ␥-catenin levels compensate for the adhesive role of ␤-catenin (53,54), indicating that ␥-catenin cannot form this bridge between PI3K and E-cadherin. p120-catenin is another catenin member directly associated with E-cadherin. Our data indicate that p120-catenin is required for E-cadherin binding to PI3K, calcium-induced activation of PI3K and PLC-␥1, and keratinocyte differentiation. Unlike other catenin members, p120-catenin association with E-cadherin is not necessary for the link to the cytoskeleton (44), although this association is important for cell adhesion (58). There are several lines of evidence demonstrating mutual dependence of E-cadherin and p120-catenin at the plasma membrane. Calcium-induced homophilic binding of E-cadherin prolongs the half-life of E-cadherin and allows binding to attract p120-catenin to the plasma membrane (52). In the absence of E-cadherin, p120-catenin becomes translocated to the cytoplasm (52). Unlike ␤-catenin or ␥-catenin, p120-catenin seems to be required for maintaining the E-cadherin complex at the plasma membrane, as indicated by the reduction in levels of these proteins when p120catenin levels are knocked down. It may do so by stabilizing E-cadherin at the plasma membrane. E-cadherin is degraded in the absence of p120-catenin (42,51). However, p120-catenin does not seem to be required for E-cadherin synthesis or trafficking to the plasma membrane (42,51), although other types of classical cadherins have been shown to be dependent on p120-catenin for such protein trafficking (43,59). Our results not only confirm these mutually dependent effects of E-cadherin and p120-catenin in human keratinocytes but also highlight that in normal human keratinocytes both E-cadherin and p120-catenin at the plasma membrane are required for PI3K activation. When E-cadherin or p120-catenin expression is inhibited, the E-cadherin-catenin complex is dissociated, and PI3K becomes stranded and inactive in the cytoplasm. When E-cadherin and p120-catenin levels are normal, calcium activation stabilizes the E-cadherin-catenin complex at the plasma membrane. These complexes recruit and activate PI3K via ␤-catenin to trigger keratinocyte differentiation.
Mice lacking E-cadherin show reduced epidermal keratinocyte differentiation, consistent with our in vitro data for human keratinocytes. However, mice lacking ␤-catenin or p120-catenin in the epidermis have been reported to show little impact on epidermal differentiation (54,60,61). In contrast, our current in vitro data show clearly that both ␤-catenin and p120-catenin are essential for human keratinocyte differentiation. The reasons for this apparent discrepancy between the in vivo and in vitro data are unclear. It is possible that mouse and human epidermal keratinocytes have different mechanisms in the regulation of differentiation by E-cadherin/catenins.
vates the kinase. PLC-␥1 activated by PIP 3 produced by PI3K in the plasma membrane triggers keratinocyte differentiation through increasing intracellular calcium concentration. This proposed mechanism is summarized in Fig. 6.