Phosphatidylinositol 3-Kinase Is a Key Regulator of Early Phase Differentiation in Keratinocytes*

The survival and growth of epithelial cells depend on adhesion to the extracellular matrix. Because epidermal keratinocytes differentiate as they leave the basement membrane, an adhesion signal may regulate the initiation of differentiation. Phosphatidylinositol 3-kinase (PI3K) is a fundamental signaling molecule that regulates the adhesion signal. Transfection of a dominant negative form of PI3K into keratinocytes using an adenovirus vector resulted in significant morphological changes comparable to differentiation and the induction of differentiation markers, keratin (K) 1 and K10. In turn, transfection with the constitutively active form of PI3K almost completely abolished the induction of K1 and K10 by differentiation in suspension cultures using polyhydroxyethylmethacrylate-coated dishes. PI3K activity was lost in suspension culture, except by cells bearing the constitutively active form of PI3K. These data demonstrate that blockade of PI3K results in differentiation and that activation of PI3K prevents differentiation. Furthermore, expression of the dominant negative form of PI3K significantly inhibited keratinocyte adhesion to the extracellular matrix and reduced the surface expression of α6 and β1 integrins in suspension culture. Moreover, expression of the active form of PI3K restored the mRNA levels of adhesion molecules that were reduced in suspension culture, including α3, α6, and β1 integrins, BP180, and BP230. In conclusion, loss of PI3K activity results in keratinocytes leaving the basement membrane and the initiation of a “default” differentiation mechanism.

The survival and growth of epithelial cells depend on adhesion to the extracellular matrix. Because epidermal keratinocytes differentiate as they leave the basement membrane, an adhesion signal may regulate the initiation of differentiation. Phosphatidylinositol 3-kinase (PI3K) is a fundamental signaling molecule that regulates the adhesion signal. Transfection of a dominant negative form of PI3K into keratinocytes using an adenovirus vector resulted in significant morphological changes comparable to differentiation and the induction of differentiation markers, keratin (K) 1 and K10. In turn, transfection with the constitutively active form of PI3K almost completely abolished the induction of K1 and K10 by differentiation in suspension cultures using polyhydroxyethylmethacrylate-coated dishes. PI3K activity was lost in suspension culture, except by cells bearing the constitutively active form of PI3K. These data demonstrate that blockade of PI3K results in differentiation and that activation of PI3K prevents differentiation. Furthermore, expression of the dominant negative form of PI3K significantly inhibited keratinocyte adhesion to the extracellular matrix and reduced the surface expression of ␣ 6 and ␤ 1 integrins in suspension culture. Moreover, expression of the active form of PI3K restored the mRNA levels of adhesion molecules that were reduced in suspension culture, including ␣ 3 , ␣ 6 , and ␤ 1 integrins, BP180, and BP230. In conclusion, loss of PI3K activity results in keratinocytes leaving the basement membrane and the initiation of a "default" differentiation mechanism.
Many cellular functions are modulated by interactions with the extracellular matrix (ECM). 1 The survival and growth of epithelial cells depends on adhesion to the ECM. Once cells leave the ECM, they may be eliminated by a form of apoptosis called anoikis (1). In malignant transformed cells, this apoptotic mechanism is likely lost, resulting in anchorage-independ-ent cell growth. The epidermis is a multilayered epithelial tissue, maintained by the precise regulation of keratinocyte proliferation, differentiation, and cell death. Cell growth is limited to the basal cell layer that attaches to the basement membrane. After leaving the basement membrane, keratinocytes differentiate and form a multilayered epidermis instead of undergoing apoptosis. Anti-apoptotic mechanisms, such as NF-B or Bcl-x, might prevent premature apoptosis in the epidermis (2,3), forming multilayers. Adhesion to the basement membrane is a negative regulator of keratinocyte differentiation, and the disruption of adhesion signals triggers terminal differentiation (4). This suggests that there are mechanisms that regulate the adhesion signal for determining the onset of differentiation. PI3K might be involved in this mechanism.
PI3K provides a universal survival signal that is involved in cell growth and anti-apoptosis. There are three classes of PI3K in mammalian cells (5)(6)(7). Most studies have examined class I PI3K, primarily because it is generally coupled to extracellular stimuli. Class I PI3K consists of a catalytic subunit and a regulatory subunit. There are four isoforms of catalytic subunit p110 (␣, ␤, ␥, and ␦) and five adaptor subunits: p85 (␣ and ␤), p55 (␣ and ␥), and p50␣ (5-7). Adhesion signals are transduced into intracellular signaling cascades via integrins through integrin-linked kinase (ILK) in a PI3Kdependent manner (8 -10). PI3K activates ILK by direct binding of phosphatidylinositol (3,4,5)-triphosphate, a predominant lipid product of PI3K (11), to ILK through a phosphatidylinositol phosphate-binding motif (9). Activation of PI3K is essential for ILK function. Focal adhesion kinase (FAK) has been implicated in the anchorage-dependent cell survival system (12). Although FAK has been shown to interact with PI3K in vitro, the physiological link between FAK and PI3K remains unclear (13). These PI3K pathways via integrin are disrupted when cells detach from the ECM, and consequently "default" apoptotic or differentiation mechanisms may be initiated.
Previously, we showed that apoptosis signal-regulating kinase (ASK1) regulates the late phase of keratinocyte differentiation (14). However, the intracellular signaling mechanism of the early phase of differentiation is poorly understood. Because epidermal keratinocytes undergo differentiation on leaving the ECM, we hypothesized that PI3K negatively regulates the early phase of differentiation by regulating the signals of adhesion to the ECM. Disruption of the interaction between integrins and the ECM results in a loss of PI3K activity (15,16). This negative signal may initiate keratinocyte differentiation. Here, we show that PI3K is a key regulator of early phase differentiation in keratinocytes.

EXPERIMENTAL PROCEDURES
Adenovirus Vectors-Adenovirus vectors (Ax) encoding a dominant negative mutant of the p85 subunit of PI3K (AxCA⌬p85) and a myristoylated p110 subunit of PI3K (AxCAMyr-p110, constitutively active form) were generous gifts from Dr. W. Ogawa, Second Department of Internal Medicine, Kobe University School of Medicine (17). Ax expressing a bacterial ␤-galactosidase gene (Ax␤-gal) was used as a control to exclude the effect of Ax itself.
For suspension culture, 2 ϫ 10 6 keratinocytes were plated onto 6-cm polyhydroxyethylmethacrylate (poly-HEMA)-coated dishes (1) in supplemented culture medium without bovine hypothalamic extract. Poly-HEMA-coated plates were made by adding 4 ml of a 10-mg/ml solution of poly-HEMA (Sigma Chemical Co., St. Louis, MO) in ethanol to the dish, drying, and repeating once, followed by extensive PBS washes. After the indicated time, cells were harvested by pipetting.
Adhesion Assay-After 24 h of Ax transfection at an m.o.i. of 10, keratinocytes were suspension-cultured for 6 h before cell adhesion. Then, 2 ϫ 10 5 cells were allowed to adhere to 3-cm type-I collagencoated dishes for 18 h in 2.0 ml of culture medium without bovine hypothalamic extract. Non-adherent cells were removed by gently washing the plates with PBS containing 1 mM CaCl 2 and 1 mM MgCl 2 . The number of adherent cells was quantified by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay. Briefly, 0.4 ml of 0.5% 3-(4,5-dimethyl-2-thiazole-2)-2,5-diphenyl-2H-tetrazolium bromide in PBS was added to 2.0 ml of fresh culture medium and further incubated for 4 h. The reaction was stopped by adding 4 ml of isopropyl alcohol containing 0.04 N HCl. The absorbance at 590 nm was measured, using the value of the medium alone as the background level. The data are given as the percentage of the control. Statistical analysis was performed using Student's t test (n ϭ 4).
Western Blotting-The analysis was performed using a Vistra ECF kit (Amersham Biosciences, Arlington Heights, IL) according to the manufacturer's instructions. Twenty g of protein were separated by 10% SDS-PAGE and transferred to a nitrocellulose membrane. The membrane was blocked with 5% nonfat dry milk in Tris-HCl, pH 7.4, 0.15 M NaCl, 0.05% Tween 20 and incubated in rabbit anti-K1 (Research Diagnostics Inc., Flanders, NJ) or monoclonal anti-K10 (Ab-1, NeoMarkers Inc., Fremont, CA) antibodies. After washing, the membrane was incubated with 1:2500 fluorescein-labeled goat anti-mouse IgG for 1 h. The signal was amplified with an anti-fluorescein antibody conjugated with alkaline phosphatase followed by the fluorescent substrate AttoPhos (Amersham Biosciences). Then, the membrane was scanned using FluoroImager (Molecular Dynamics Inc., Sunnyvale, CA). The intensity of each band was quantified with ImageQuaNT (Molecular Dynamics Inc.), referring to the control as one unit.
Cell Sorter Analysis-Cell sorter analysis was performed as previously described (19). Keratinocytes were harvested from the dishes with trypsin-EDTA (adherent cells) or by pipetting (suspension culture) and filtered through a cell-strainer (Becton Dickinson, Mountain View, CA). Cells (1 ϫ 10 6 ) were incubated with anti-␣ 6 (4F10, Chemicon International, Temecula, CA) or anti-␤ 1 (P5D2, Chemicon International) antibodies, or mouse IgG (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) in PBS containing 1 mM CaCl 2 and 1 mM MgCl 2 on ice for 30 min. After washing, the cells were reacted with fluorescein isothiocyanate-conjugated anti-mouse antibody for 30 min on ice. The labeled cells were analyzed with a flow cytometer (Becton Dickinson, Co., CA). The data are reported as the mean fluorescence intensity (MFI) or percent MFI referring to the control as 100%. Statistical analysis was performed using Student's t test (n ϭ 4).

RESULTS
We infected normal human keratinocytes with Ax, at a multiplicity of infection (m.o.i.) of 10. After 24 h, gene expression was found in almost all keratinocytes (data not shown). Transfection of Ax␤-gal, AxCAMyr-p110, or AxCA⌬p85 did not cause any apoptosis. Expression of the dominant negative form of PI3K (AxCA⌬p85) in keratinocytes resulted in significant morphological changes comparable to differentiation (Fig. 1). There were no apparent morphological changes in keratinocytes transfected with either the constitutively active form of PI3K (AxCAMyr-p110) or the control vector Ax␤-gal, compared with the cells without Ax.
After transfection of Ax, the expression of differentiation markers was analyzed by RPA (Fig. 2). The dominant negative form of PI3K (AxCA⌬p85) significantly enhanced the expression of early phase differentiation markers, K1 and K10 mRNA. However, late phase differentiation markers, such as transglutaminase-1, loricrin, and involucrin mRNA, were not induced. Wortmannin, a PI3K inhibitor, also enhanced K1 and K10 mRNA expression. Western blot analysis confirmed the induction of K1 and K10 protein by the dominant negative form of PI3K (Fig. 3A). We next analyzed the PI3K activity after the transfection of Ax (Fig. 3B). The dominant negative form of PI3K (AxCA⌬p85) almost completely inhibited the PI3K activity. By contrast, the active form of PI3K (AxCAMyr-p110) enhanced the PI3K activity. This suggested that inhibition of PI3K activity induces early phase differentiation.
A suspension culture is a keratinocyte differentiation model, in which the interaction of keratinocytes with the ECM is disrupted (4). We employed a suspension culture with poly-HEMA-coated culture plates, which inhibit cell-to-ECM interaction, but not cell-to-cell interaction (20). First, we analyzed the expression of differentiation markers in suspension culture with RPA. Transglutaminase-1, involucrin, K1, and K10 mRNA started to increase at 12 h and persisted up to 48 h (Fig.  4A). Loricrin mRNA increased at 48 h. Next, PI3K activity was analyzed in suspension culture (Fig. 4B). PI3K activity decreased within 3 h of suspension culture. This indicates that an adherent signal is essential for the activation of PI3K.
We then analyzed whether the active form of PI3K inhibits the induction of differentiation markers by suspension culture.
After transfection of Ax, keratinocytes were suspension-cultured for 24 h. Transfection of the constitutively active form of PI3K (AxCAMyr-p110) almost completely abolished the induction of early phase differentiation markers, K1 and K10 mRNA (Fig. 5), whereas the expression of late phase differentiation markers, transglutaminase-1 and involucrin, was not affected. Other Ax had no effect on the expression of differentiation markers. This result was confirmed by Western blot analysis (Fig. 6A). The induction of K1 and K10 protein by suspension culture was abolished by expression of the constitutively active form of PI3K (AxCAMyr-p110). The PI3K activity decreased in suspension culture (Fig. 6B), as shown in Fig. 4B. However, cells bearing the constitutively active form of PI3K (AxCAMyr-p110) retained PI3K activity in suspension culture. This indicates that activation of PI3K prevents early phase differentiation, and blockade of PI3K activity results in early phase differentiation.
The role of PI3K in keratinocyte adhesion to the ECM was then assessed. After transfection of Ax, cells were plated on type-I collagen-coated dishes. Expression of the dominant negative form of PI3K (AxCA⌬p85) inhibited keratinocyte adhesion to the dishes to 43% (Fig. 7). The effect of other vectors was minimal. Next, we studied whether PI3K regulates the expression of ␣ 6 and ␤ 1 integrins using cell-sorter analysis. Expression of the active or dominant negative form of PI3K did not affect the expression of ␣ 6 or ␤ 1 integrins on adherent keratinocytes (data not shown). When the cells were suspensioncultured, the expression of ␣ 6 and ␤ 1 integrins decreased within 24 h (Fig. 8). Then, we analyzed the effects of the active and dominant negative forms of PI3K on the integrin expression at 6 h of suspension culture (Fig. 9). Cells bearing the dominant negative form (AxCA⌬p85) of PI3K further decreased the expression of ␣ 6 and ␤ 1 integrins to 83 and 74% of control, respectively. Next we analyzed the mRNA expression of molecules that mediate adhesion to the ECM. These include integrins, BP180, and BP230. BP180 and BP230 are components of the hemidesmosome that promote adhesion to the basement membrane (21). After transfection of Ax, keratinocytes were suspension-cultured for 24 h. The expression of mRNA was analyzed by RPA. Suspension culture significantly reduced the mRNA expression of integrins and BPs (Fig. 10). The active form of PI3K (AxCAMyr-p110) restored the mRNA levels of adhesion molecules, including ␣ 3 , ␣ 6 , and ␤ 1 integrins, BP180, and BP230 (Fig. 10). This suggests that PI3K maintains the expression of adhesion molecules in the basal cell layer. Combined, PI3K regulates keratinocyte adhesion to the ECM. DISCUSSION We have shown that inhibition of PI3K initiates differentiation and that activation of PI3K prevents differentiation. In adherent cells, PI3K is activated by an adhesion signal (15,16). This indicates that, at the basal cell layer, activated PI3K drives the keratinocytes to grow and not to differentiate. A decrease in PI3K activity results in a loss of keratinocyte adhesion to the ECM (Fig. 5), which might be followed by keratinocyte migration to the suprabasal layer. In the suprabasal layer, a "default" differentiation mechanism may initiate, which is suppressed by PI3K at the basal layer. Therefore, PI3K is a key regulator of the initiation of keratinocyte differentiation. Because Akt is activated on adhesion to the ECM and to protect against anoikis (16), Akt is a candidate downstream signaling molecule of PI3K. We repeated our experiments, including the constitutively active form of Akt, and found that Akt did not reduce the expression of early phase differentiation markers induced by suspension culture (data not shown). Therefore, some other molecules may participate downstream from PI3K. Such molecules may include ILK or glycogen synthase kinase (GSK) 3␤. ILK phosphorylates both Akt (22,23) and GSK3␤ (9, 10) and regulates their activity. Phosphorylation of GSK3␤ results in the activation of the ␤-catenin/ Lef-1 pathway (24, 25). Overexpression of wild-type ILK suppresses differentiation of mouse mammary glands (26). Although Akt had no effect on keratinocyte differentiation, it is . The expression of transglutaminase-1, loricrin, involucrin, K1, and K10 mRNA was analyzed by RPA. Control: without Ax. GAPDH was used as an internal standard. The intensity of each band was quantified and standardized with GAPDH. The data are reported as -fold induction, referring to the signal of the control as 1 unit. still possible that ILK or GSK3␤ affects keratinocyte differentiation by interacting with some molecules other than Akt.
With the initiation of differentiation, keratinocytes downregulate integrin expression and migrate to the suprabasal layer in vivo. During suspension-induced differentiation, adhesion to the ECM is down-regulated in two stages (19,27). First, the ability of the integrin to bind the ligand is reduced. Second, the integrin is lost from the cell surface. Although expression of the dominant negative form of PI3K inhibited the adhesion to 43%, the levels of ␣ 6 and ␤ 1 integrin expression were still high (83 and 74% of control, respectively). A decrease in integrin expression by the dominant negative form of PI3K may account for part of the loss of cell adhesion. Because the expression of the dominant negative form of PI3K had a minimal effect on integrin expression, the dominant negative form of PI3K might also inhibit the binding activity of integrins. Integrins are molecules that can transduce both "inside-out" and "outside-in" signals (28). Adhesion regulation by PI3K is reported in other types of cell. In T lymphocytes, PI3K regulates cell adhesion to the ECM via ␤ 1 integrin, without changing its expression level (29,30), as occurs in keratinocytes (19,27). Moreover, the active form of PI3K restored the mRNA levels of adhesion molecules that were reduced by suspension culture in keratinocytes. This indicates that activation of PI3K by adhesion plays an important role in maintaining the adhesion molecules in the basal cell layer. Thus, PI3K regulates keratinocyte adhesion to the ECM.
The proliferative and anti-apoptotic roles of PI3K have been studied; however, PI3K is not recognized as a regulator of keratinocyte differentiation. Several intra-cellular signaling A, adherent cells under phase contrast microscopy. B, the number of adherent cells was quantified by the 3-(4,5-dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide assay. Briefly, 0.4 ml of 0.5% 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide in PBS was added to 2.0 ml of culture medium and further incubated for 4 h. After stopping the reaction, the absorbance at 590 nm was measured, referring to the control value as 100%. Statistical analysis was performed using Student's t test (n ϭ 4). *, statistically significant (p Ͻ 0.01). Control: without Ax.
pathways have been identified as regulators of keratinocyte differentiation, such as the ASK1 (14), p38 MAPK (14,31), MAPK (32,33), protein kinase C (34,35), and NF-B (2, 36 -38) pathways. The upper epidermis expresses ASK1, which induces late phase differentiation markers, including transglutaminase-1, involucrin, and loricrin (14). Epidermal keratinocytes express ␣, ␦, ⑀, , and isoforms of protein kinase C (34, 39 -42). It has been suggested that protein kinase C is involved in the transition process from the spinous to the granular layer (43). We examined whether PKC isoforms were involved in the dominant negative form of PI3K-induced differentiation. Activation of PKC isoforms was determined by analyzing the subcellular distribution of PKC isoforms using Western blotting. The redistribution of PKC from the soluble fraction to a particulate fraction is a useful indicator of PKC activation (44). However, expression of the dominant negative form of PI3K (AxCA⌬p85) did not affect the subcellular distribution of PKCs (data not shown), indicating that PKCs are not activated by the dominant negative form of PI3K (AxCA⌬p85). Furthermore, both ASK1 and protein kinase C regulate involucrin promoter activity via p38 MAPK (14,31). Therefore, ASK1, p38 MAPK, and protein kinase C seem to be the regulators of late phase differentiation in keratinocytes. On the other hand, PI3K, NF-B, and MAPK are implicated in the early phase of differenti- transfection with Ax, keratinocytes were suspension-cultured for 6 h, and then ␣ 6 and ␤ 1 integrin expression was analyzed with a cell sorter, as in Fig. 8. A, dotted line, control (without Ax). Solid lines, Ax␤-gal, AxCA⌬p85, and AxCAMyr-p110. B, the data are reported as the MFI and percent MFI referring to the control as 100%. Statistical analysis was performed using Student's t test (n ϭ 4). *, statistically significant (p Ͻ 0.01). ation. The epidermis of IKK␣-deficient mice shows abnormal differentiation (37). Because NF-B translocates from the cytoplasm to the nucleus at the suprabasal layer (37), NF-B seems to play an important role in the initiation of differentiation. Integrin ligation activates MAPK and negatively regulates differentiation, as does PI3K (32,33). MAPK also regulates cell adhesion to the ECM (32,33). The identities of these regulators, their relationships, and the overall regulatory system remain unclear and should be further elucidated.
␤ 1 integrin generates at least two signals in keratinocytes. One signal is a clustering of receptors into a focal adhesion with the polymerization of actin filaments (32,33,45,46). The other signal is a negative stimulus for differentiation (47) that is independent of receptor clustering (33,48). The predominant integrins in the epidermis are ␣ 3 ␤ 1 and ␣ 6 ␤ 4 (49). Although ␤ 1 integrin regulates differentiation in an in vitro model, ␤ 1 integrin is not essential for the differentiation mechanism in a gene-targeted mouse model. The proliferative potential is markedly reduced in the ␤ 1 null epidermis; however, the differentiation program is preserved (48,50). The differentiation process may be compensated for by other mechanisms, such as ␤ 4 integrin. By contrast, in ␤ 4 null mice, basal-like keratin-5positive cells were seen in the spinous layer and apoptotic cells in the basal cell layer (51). This suggests that the differentiation program is disturbed. There may be a distinction between the signal transduction mechanisms of the two ␤-integrin subunits.
In conclusion, activation of PI3K inhibits early phase differentiation, and a decrease in PI3K activity results in loss of adhesion to the ECM and the initiation of early phase differentiation. Therefore, PI3K is a key regulator of early phase differentiation in keratinocytes.