Transforming Growth Factor- -activated Kinase 1 Is Essential for Differentiation and the Prevention of Apoptosis in Epidermis*

Transforming growth factor-activated kinase 1 (TAK1) is a member of the mitogen-activated protein (MAP) kinase family and is an upstream signalingmolecule of nuclear factorB (NFB). Given that NFB regulates keratinocyte differentiation and apoptosis, TAK1 may be essential for epidermal functions. To test this, we generated keratinocyte-specific TAK1-deficient mice fromMap3k7 flox/flox mice andK5-Cremice. The keratinocyte-specific TAK1-deficient mice were macroscopically indistinguishable from their littermates until postnatal day 2 or 3, when the skin started to roughen and wrinkle. This phenotype progressed, and the mice died by postnatal day 7. Histological analysis showed thickening of the epidermis with foci of keratinocyte apoptosis and intra-epidermal micro-abscesses. Immunohistochemical analysis showed that the suprabasal keratinocytes of the TAK1-deficient epidermis expressed keratin 5 and keratin 14, which are normally confined to the basal layer. The expression of keratin 1, keratin 10, and loricrin, which aremarkers for the suprabasal and late phase differentiation of the epidermis, was absent from the TAK1-deficient epidermis. Furthermore, the TAK1-deficient epidermis expressed keratin 16 and had an increased number of Ki67-positive cells. These data indicate that TAK1 deficiency in keratinocytes results in abnormal differentiation, increasedproliferation, and apoptosis in the epidermis.However, the keratinocytes from theTAK1-deficient epidermis induced keratin 1 in suspension culture, indicating that the TAK1-deficient keratinocytes retain the ability to differentiate. Moreover, the removal of TAK1 from cultured keratinocytes ofMap3k7 flox/flox mice resulted in apoptosis, indicating that TAK1 is essential for preventing apoptosis. In conclusion, TAK1 is essential in the regulation of keratinocyte growth, differentiation, and apoptosis.

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, which attaches to the basement membrane. After leaving the basement membrane, keratinocytes differentiate and form a multilayered epidermis instead of undergoing apoptosis. This keratinocyte differentiation is regulated by intracellular signaling pathways involving nuclear factor-B (NF-B), 2 the MAP kinase family, phosphatidylinositol 3-kinase, and protein kinase C (1)(2)(3).
In the epidermis, NF-B is found in the cytoplasm of basal cells (1). The functional blockade of NF-B by expressing dominant negative NF-B in transgenic mouse epidermis produced a hyperplastic epithelium in vivo (1). With deficiency of the p65/RelA subunit of NF-B, the epidermis is hyperplastic (4). Conversely, the overexpression of active p50 (a subunit of NF-B) and p65 in the transgenic epithelium produced hypoplasia and growth inhibition, suggesting a role for NF-B in negative cellular growth control (1). Furthermore, the expression of active p50 and p65 in keratinocytes inhibits cell cycle progression to G 1 arrest in vitro, which was associated with an increased p21 level (5) and CDK4 down-regulation (6). This CDK4 regulation is dependent on both the TNFR1 and JNK pathways (7). In addition to regulating differentiation and cell growth, NF-B protects keratinocytes from apoptosis, and the blockade of NF-B function in the epidermis by the expression of the dominant negative mutant IB␣ provoked premature spontaneous cell death (8). Although the roles of the NF-B pathway have been studied in keratinocytes, the upstream signal of the NF-B pathway has not been clarified fully.
Transforming growth factor-␤-activated kinase 1 (TAK1), a member of the MAP kinase kinase kinase family, is a signaling molecule upstream from NF-B and is encoded by the Map3k7 gene. TAK1 was originally identified as a signaling molecule activated by transforming growth factor ␤ (9). TAK1 is also involved in IL-1 signaling and TNF-␣-induced activation of NF-B and MAP kinases (MAPKs). Activated TAK1 is recruited to TRAF6 and TRAF2, in response to the IL-1 and TNF receptors, respectively (10 -12). TAK1 forms a complex with TAK1-binding protein (TAB) 1, TAB2, and TAB3 (10,(12)(13)(14). The activated TAK1 complex phosphorylates IB kinases (IKKs) and MAPK kinase 6, which activate NF-B and MAPKs, respectively. The transfection of TAK1 small interfering RNA markedly inhibited TNF-␣-and IL-1-mediated activation of NF-B (11). Recently, TAK1 was shown to be essential for innate and adaptive immune responses by generating B cellspecific TAK1 deficiency (15).
Given that TAK1 is an upstream signaling molecule of the NF-B pathway, TAK1 might regulate keratinocyte growth, differentiation, and apoptosis. However, the function of TAK1 in epidermal keratinocytes has not been studied. To address this issue, we generated keratinocyte-specific TAK1-deficient mice by using Cre-recombinase transgenic mice under the control of the K5 promoter.

EXPERIMENTAL PROCEDURES
Generation of Keratinocyte-specific TAK1-deficient Mice Using Gene Targeting with the Cre Transgene-The targeting construct has been described previously (15). We generated keratinocyte-specific TAK1-deficient mice by breeding Map3k7 flox/flox mice with mice carrying the Cre transgene under the control of the keratin 5 promoter (K5-Cre) (16). The Map3k7 flox/flox mice were bred with K5-Cre mice to generate K5-Cre/Map3k7 flox/ϩ mice. Subsequently, the K5-Cre/ Map3k7 flox/ϩ mice were bred to Map3k7 flox/flox mice to generate K5-Cre/Map3k7 flox/flox mice. The genotype of each mouse was confirmed using PCR and Western blot analysis. The TAK1 primer sequences were 5Ј-GGAACCCGTGGATAAGTG-CACTTGAAT-3Ј and 5Ј-GGCTTTCATTGTGGAGGTA-AGCTGAGA-3Ј. The amplified products were 320 bp for the floxed allele and 280 bp for the wild-type allele. The Cre-recombinase primers were 5Ј-TTACCGGTCGATGCAACGAGT-GATG-3Ј and 5Ј-TTCCATGAGTGAACGAACCTGGTCG-3Ј. Isolated keratinocytes were cultured overnight, and the adherent keratinocytes were harvested for Western blot analysis (see Fig. 1D). This protocol was approved by the Institutional Review Board of Ehime University School of Medicine.
Trans-epidermal Water Loss (TEWL)-TEWL was measured using a Tewameter TM210 (Integral Co., Tokyo, Japan) according to the manufacturer's instructions. The data are expressed as the means Ϯ S.E. The statistical significance was determined using the paired Student's t test. The differences were considered statistically significant at p Ͻ 0.01.
Histological Analysis-Mouse skin was fixed in 3.6% formaldehyde, dehydrated, and embedded in paraffin. Four-m sections were stained with hematoxylin and eosin.
To analyze differentiation markers, keratin (K) 5, K14, K1, K10, and loricrin, the paraffin-embedded sections were deparaffinized, blocked with 10% goat serum, and reacted with the first antibodies overnight at 4°C. After washing with phosphate-buffered saline, the first antibodies were detected using a peroxidase staining kit (ImmPRESS; Vector Laboratories, Burlingame, CA) and visualized with the chromogen 3-amino-9-ethyl-cabazole according to the manufacturer's instructions. For K16 and Ki67 staining, the deparaffinized sections were boiled in 10 mM citrate buffer, pH 6.0, for 40 min and cooled at room temperature for 20 min for antigen retrieval. Subsequently, K16 and Ki67 were stained following the same procedure as for the differentiation markers.
Confocal Laser Scanning Microscopy-Frozen skin sections (4 m) were blocked with 10% goat serum and reacted overnight at 4°C with rabbit anti p50 and p65. After washing with phosphate-buffered saline, the sections were incubated with Alexa Fluor 488-conjugated donkey anti-rabbit IgG for 30 min at room temperature. The stained specimens were observed under a LSM 510 microscope (Carl Zeiss, Jena, Germany). The images were captured using LSM 510 software.
TdT-mediated dUTP Nick End Labeling (TUNEL)-Keratinocyte apoptosis was detected by the TUNEL method using an in situ cell death detection kit (Roche Applied Sciences). After being deparaffinized, the sections were treated with 20 g/ml proteinase K in 10 mM Tris-HCl, pH 7.4, for 15 min at room temperature and labeled according to the manufacturer's instructions. The labeled cells were observed under a fluorescence microscope.
Keratinocyte Culture-Primary mouse keratinocytes were isolated from newborn mouse skin. The skin samples were cut into 3-5-mm pieces and incubated with 250 units/ml dispase (Godoshusei, Tokyo, Japan) in Dulbecco's modified Eagle's medium overnight at 4°C. After the epidermis was separated from the dermis, the epidermal sheets were incubated in a 0.25% trypsin solution for 10 min at 37°C and teased with forceps. The keratinocytes were collected by centrifugation and were cultured further in CnT-02 medium (CellnTec, Bern, Switzerland).
For suspension culture, keratinocytes were plated onto 6-cm polyhydroxyethylmethacrylate (poly-HEMA)-coated plates (3). The poly-HEMA-coated plates were made by adding 4 ml of a solution containing 10 mg/ml poly-HEMA (Sigma-Aldrich) in ethanol to the dish, drying, and repeating once, followed by extensive phosphate-buffered saline washes. After 24 h, the cells were harvested by pipetting.
Western Blotting-Keratinocytes were harvested on ice with lysis buffer containing 1.0% Nonidet P-40, 1% sodium deoxycholate, 0.1% SDS, 150 mM NaCl, 20 mM Tris-HCl, pH 7.5, 1 mM EDTA, and 0.1% protease inhibitor (Sigma-Aldrich). The proteins (20 g) 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, and 0.05% Tween 20 and was incubated with the appropriate first antibody. After washing, the membrane was incubated with fluorescein-labeled goat anti-mouse IgG (1:2,500 dilution) for 1 h. The signal was amplified with an anti-fluorescein antibody conjugated with alkaline phosphatase, followed by the fluorescent substrate AttoPhos (Molecular Dynamics, Sunnyvale, CA). The membrane was scanned using FluoroImager (Molecular Dynamics), and the intensity of each band was quantified with ImageQuant (Molecular Dynamics), referring to the control as one unit.
Reverse Transcriptase (RT)-PCR-Total RNA samples were isolated using Isogen (Nippon Gene, Tokyo, Japan), and the TAK1 mRNA expression was analyzed by RT-PCR using RT-PCR High Plus (Toyobo, Osaka, Japan). The PCR product (337 bp) was sequenced to confirm the accuracy of amplification. The primer sequences for TAK1 were 5Ј-AGTGCTGACATG-TCTGAAAT-3Ј and 5Ј-TTCGAACACTGCCATGGATT-3Ј, and those for the internal standard glyceraldehyde-3-phosphate dehydrogenase were 5Ј-ACCACAGTCCATGCCAT-CAC-3Ј and 5Ј-TCCACCACCCTGTTGCTGTA-3Ј. The intensity of each band was quantified using NIH Image. The data are presented as fold induction relative to the control signal, set at 1 unit.
Adenovirus Vector-Adenovirus vector (Ax) encoding Crerecombinase was generated using the COS-TPC method (17). Virus stocks were prepared by standard procedures. Concentrated, purified virus stocks were prepared using a CsCl gradient, and the virus titer was checked using a plaque formation assay. We infected keratinocytes with Ax at a multiplicity of infection of 100. Empty Ax-1W vector was used as a control.
LDH Assay-Cell death was analyzed quantitatively by measuring LDH release using an LDH assay kit (Kyokutokogyo, Tokyo, Japan). Keratinocytes were cultured on 6-cm dishes, and Ax was transfected. At the indicated time, 100 l of supernatant were harvested, centrifuged, and stored at Ϫ70°C until the LDH assay was performed according to the manufacturer's instructions. The LDH of living cells was obtained by cell lysis with 0.1% Tween 20. LDH release was expressed as a percentage of the total LDH, which was obtained by summing the LDH released and the LDH of living cells. The data are expressed as the means Ϯ S.E. The statistical significance was determined using the paired Student's t test (n ϭ 5). The differences were considered statistically significant at p Ͻ 0.01. Keratinocyte-specific TAK1-deficient mice were generated by breeding Map3k7 flox/flox mice with mice carrying the Cre transgene under the control of the keratin 5 promoter (K5-Cre) (16). A, schematic representation of the targeting strategy. A targeting construct was designed to flank exon 2 of the Map3k7 gene and the neo r gene with loxP sites (triangles). Cre-mediated removal was performed to generate the deleted (⌬) allele of Map3k7. The genotype of each mouse was confirmed by PCR (B and C). B, a primer pair was designed to include the LoxP site (triangle), so that the PCR products represent the floxed (320 bp) and wild-type (280 bp) alleles of Map3k7. C, the PCR product represents Cre-recombinase. D, RT-PCR analysis showed the Cre-mediated keratinocyte-specific deletion of TAK1 in K5-Cre/Map3k7 flox/flox mice. mRNA isolated from the epidermis, heart, kidney, and lung of K5-Cre/ Map3k7 flox/flox mice at birth was subjected to RT-PCR to determine TAK1 expression. The faint band from the epidermis detected at the expected size is likely the result of the presence of nonkeratinocyte cells in the epidermis, such as Langerhans cells or melanocytes. Glyceraldehyde-3-phosphate dehydrogenase is an internal standard. E, Western blot analysis showed the Cremediated deletion of TAK1 from the keratinocytes of K5-Cre/Map3k7 flox/flox mice. Keratinocytes isolated from Map3k7 flox/flox and K5-Cre/Map3k7 flox/flox mice at birth were subjected to Western blot analysis to detect TAK1. There was a faint band in the keratinocytes from K5-Cre/Map3k7 flox/flox mice, which was slightly lower than that for Map3k7 flox/flox mice. This may represent TAK1⌬ (A), which lacks the ATP-binding site required for kinase activity (15). ␤-Actin is an internal standard. In this manuscript, K5-Cre/Map3k7 flox/flox mice and Map3k7 flox/flox mice are referred as TAK1-KO and control mice, respectively. FIGURE 2. The macroscopic phenotype of TAK1-KO mice. A, appearance of TAK1-KO mice at P6. The skin is rough and wrinkled, with mild scaling. The TAK1-KO mice weighed approximately half the control weight. B, TEWL in TAK1-KO mice at P6. The TEWL value, a measure of epidermal barrier function, in TAK1-KO mice (n ϭ 5) was 6.9-fold that in the control mouse (n ϭ 7). The data are expressed as the means Ϯ S.E. The statistical significance was determined using the paired Student's t test. * The difference was considered statistically significant at p Ͻ 0.01. AUGUST 4, 2006 • VOLUME 281 • NUMBER 31
Rough and Wrinkled Skin with Disrupted Barrier Function, Growth Retardation, and Early Mortality in TAK1-KO Mice-At birth, the TAK1-KO mice were macroscopically indistinguishable from their littermates until postnatal day 2 or 3 (P2-P3), when the skin started to roughen and wrinkle. This phenotype progressed with mild scaling, and the TAK1-KO mice subsequently died by P7 ( Fig. 2A). At this stage, the mice weighed approximately half the control weight. To evaluate epidermal barrier function, we measured TEWL at P6 (Fig. 2B). The TEWL value in TAK1-KO mice was 6.9-fold that in the control, indicating that the barrier function had been disrupted. Given that the epidermal barrier function arises during keratinocyte differentiation, the regulation of keratinocyte differentiation might be disturbed in TAK1-KO mice. The Map3k7 flox/ϩ and Map3k7 ϩ/ϩ mice with or without K5-Cre showed no pathological phenotypes.
Hyperplasia, Apoptosis, and Micro-abscess Formation in the Epidermis of TAK1-KO Mice-The epidermis of TAK1-KO mice at P0 was histologically indistinguishable from that of control mice, except for a few apoptotic cells (arrow). However, the histological analysis of the skin of TAK1-KO mice at P6 (Fig. 3A) showed many apoptotic cells (arrow) and marked thickening of the epidermis without a granular layer ( g), indicating abnormal epidermal differentiation. The thickened epidermis contained foci of keratinocyte apoptosis (Fig. 3B, left  panel, arrow) and intra-epidermal micro-abscesses (Fig. 3B,  right panel, arrowhead), which were associated with apoptotic cells (arrow). The oral mucosa showed a similar phenotype to that of the epidermis (Fig. 3A). Clinically, there were no apparent erosions or ulcers on the oral mucosa (data not shown).  No microbial pathogens were seen in the micro-abscesses on staining the specimens with methylene blue (data not shown), indicating that the micro-abscesses were not caused by infection. Because the micro-abscesses were associated with keratinocyte apoptosis, one possible mechanism for micro-abscess formation is the release of cytokines or chemokines from dead keratinocytes.
Abnormal Differentiation and Increased Proliferation of Keratinocytes in TAK1-KO Mice-To analyze the differentiation status of epidermal keratinocytes, we performed immunohistochemical analysis of the skin using antibodies against differentiation markers (Fig. 4). The expression of differentiation markers of TAK1-KO epidermis at P0 was similar to that of controls. The expression pattern started to change at P2, and the changes became more marked at P6. The expression of K5 and K14 is normally confined to the basal cell layer, as seen in the control. However, the suprabasal keratinocytes of TAK1-KO mice at P6 expressed K5 and K14. K1, K10, and loricrin are markers for the suprabasal and late phase differentia-tion of keratinocytes and are normally expressed on the suprabasal keratinocytes and in the upper epidermis, respectively. In addition to the ectopic expression of K5 and K14, the expression of K1, K10, and loricrin was absent on the viable epidermal keratinocytes of TAK1-KO mice at P6. Furthermore, the epidermis of TAK1-KO mice at P6 expressed K16, a marker for inflammatory and hyperproliferative epidermal keratinocytes; K16 expression was absent from control mice. These data indicate that keratinocyte differentiation was disturbed by removing TAK1 from keratinocytes.
We further analyzed growth and apoptosis in the epidermis using Ki67 staining and the TUNEL method (Fig. 5). The staining of Ki67 revealed increased proliferation of keratinocytes in the epidermis of TAK1-KO mice at P6 (Fig. 5A), in contrast to the increased apoptosis shown with the TUNEL method (Fig.  5B). Furthermore, Ki67-positive keratinocytes were present in the suprabasal layer of the epidermis of TAK1-KO mice. The expression of p50 and p65 in the basal layer of the epidermis of TAK1-KO mice was decreased at P6 (Fig. 5C).
Induction of K1 in the Keratinocytes of TAK1-KO Mice in Suspension Culture-Based on the immunohistochemical results, we postulated that the keratinocytes of TAK1-KO mice do not express K1 by differentiation. To test this, we isolated and cultured keratinocytes from the epidermis of newborn mice. Although the number of Ki67-positive keratinocytes was increased in the epidermis of TAK1-KO mice, the in vitro growth of keratinocytes obtained from TAK1-KO mice was significantly reduced. Therefore, to study the ability of keratinocytes to differentiate, freshly isolated keratinocytes were seeded on culture plates and incubated overnight; adherent cells were subjected to suspension culture (3) to induce differentiation. Unexpectedly, the keratinocytes from TAK1-KO mice induced K1 in the suspension culture, as did those from control mice (Fig. 6).
Apoptosis of Keratinocytes on Depletion of TAK1 in Culture-Because of the difficulty in culturing TAK1-KO keratinocytes, we depleted TAK1 in cultured keratinocytes obtained from the control mice (Map3k7 flox/flox ). An adenovirus vector carrying Cre-recombinase (Ax-Cre) was transfected to the cultured keratinocytes. Empty Ax-1W vector was used as a control. As shown in Fig. 7, the expression of TAK1 was reduced to 0.4-fold that in control mice at 48 h after the transfection of Ax-Cre. The reduction of TAK1 in cultured keratinocytes resulted in apo-  Freshly isolated keratinocytes were seeded on culture plates and incubated overnight. To induce differentiation, the adherent cells were subjected to suspension culture using poly-HEMA-coated plates. Western blot analysis was performed to analyze the expression of K1. ␤-Actin is an internal standard. The intensity of each band was quantified relative to the control, set at one unit. ptosis 72 h after transfection (Fig. 7C), indicating that TAK1 prevents keratinocyte apoptosis. The quantitative analysis of cell death based on measurements of LDH release revealed that the apoptosis started at 48 h and was completed at 72 h (D).

DISCUSSION
Although TAK1 is evolutionarily conserved and is thought to be an important intracellular signaling molecule, little is known about its physiological roles. In Xenopus embryonic development, TAK1 is involved in mesoderm induction mediated by bone morphogenetic protein, a member of the transforming growth factor ␤ family (18). TAK1-deficient Drosophila are viable and fertile, but they do not produce antibacterial peptides and are highly susceptible to Gram-negative bacteria infection (19). In HeLa cells, it is thought that TAK1 is involved in the TNF receptor and IL-1 receptor/Toll-like receptor-mediated signaling pathways upstream from NF-B and MAPKs, based on small interfering RNA inhibition of TAK1 expression (11). Because TAK1-deficient mice are embryonically lethal (15), by utilizing embryonic fibroblasts, a loss of response to IL-1␤ and TNF in TAK1-deficient cells was demonstrated (15). To study the role of TAK1 in B cell function, mice with B cell-specific TAK1 deficiency were generated (15). The B cell-specific TAK1-deficient mouse showed that TAK1 was indispensable for B cell activation in response to Toll-like receptor ligands (CpG DNA, poly(I:C), and LPS), CD40, and B cell receptor cross-linking (15). Furthermore, TAK1-deficient B cells failed to activate NF-B and JNK in response to IL-1␤, TNF, and Toll-like receptor ligands. Therefore, TAK1 has essential functions in the Toll-like receptor-, IL-1 receptor-, TNF receptor-, and B cell receptor-mediated cellular responses. In addition, we showed that TAK1 is essential for regulating keratinocyte differentiation and preventing keratinocyte apoptosis in the epidermis.
TAK1-KO mice die by P7. Because the barrier function of the epidermis was severely disturbed, as shown by TEWL, the impaired homeostasis of body fluids resulting from water loss through the skin may have caused the early mortality. Another possible cause of the early mortality is impaired feeding. The body weight at death was approximately half that of the control mice. The lips of the TAK1-KO mice showed signs of scaling, and the histological study revealed massive keratinocyte apoptosis (data not shown), as in the oral mucosa. These findings suggested that the diseased lip and oral mucosa affected nursing, resulting in severe growth retardation, which might cause the early mortality.
TAB1 and TAB2 were identified as adaptor proteins of TAK1 using yeast two-hybrid screening (10,13). TAB1 binds to TAK1 constitutively and induces kinase activity on stimulation with IL-1. TAB2 is an adaptor molecule linking TRAF6 and TAK1. With IL-1 stimulation, TAB2 translocates from the cell membrane to the cytosol and binds to TRAF6 and TAK1. TAB3 is a TAB2 homologue that interacts with TRAF6 and TRAF2 on stimulation with IL-1 and TNF, respectively (12). Co-transfection with small interfering RNA s directed against both TAB2 and TAB3 inhibited both the IL-1-and TNF-induced activation of TAK1 and NF-B (12). These results suggest that TAB2 and TAB3 play redundant roles as mediators of TAK1 activation in IL-1 and TNF signal transduction. Although TABs have been FIGURE 7. Apoptosis on depletion of TAK1 from cultured keratinocytes. TAK1 expression was depleted in cultured keratinocytes of the control mice (Map3k7 flox/flox ). An adenovirus vector carrying Cre-recombinase (Ax-Cre) was transfected into the cultured keratinocytes at a multiplicity of infection of 100. The empty vector Ax-1W was used for the control. The expression of TAK1 mRNA and protein were analyzed using RT-PCR (A) and Western blotting (B). The intensity of each band was quantified relative to the control, set at one unit. Glyceraldehyde-3-phosphate dehydrogenase and ␤-actin are internal standards. There was a weak band in the cells treated with Ax-Cre that was slightly lower than that of Ax-1W at 48 h (B). This might represent TAK1⌬ (15), as in Fig. 1E. At 72 h after transfection, the cell morphology was observed under a phase contrast microscope, and apoptotic cells were detected using the TUNEL method (C). Keratinocytes of C57/BL6 (BL6) mice were also transfected with Ax vector as a control. Cell death was quantified by measuring LDH release (D). Ax was transfected to the cultured keratinocytes, and the culture supernatant was harvested for LDH assay at the indicated time. LDH release was expressed as the percentage of total LDH (mean Ϯ S.E.), which was obtained by summing the LDH release and the LDH of living cells. The statistical significance was determined using the paired Student's t test (n ϭ 5). * and **, differences were considered statistically significant for p Ͻ 0.01. identified as TAK1 adaptor proteins, the genetic study of these proteins suggested that they do not always function together as a TAK1 complex. The neural tube development was abnormal in the TAK1-deficient embryo (20). This phenotype is substantially different from those of TAB1-deficient (abnormal cardiovascular and pulmonary morphogenesis) and TAB2-deficient (liver degeneration and apoptosis) mice (21,22). These data suggested not only that TAK1 and the TAB complex function as a single unit, but that each component has a distinct role during development (20). Furthermore, the roles of TABs in keratinocytes have not yet been studied. Therefore, molecular interaction between TABs and TAK1 and their roles in keratinocytes should be clarified.
Genetic studies have shown that mutations in the human NEMO/IKK-␥ gene are the cause of incontinentia pigmenti or Bloch-Sulzberger syndrome (23). Disruption of the NEMO/ IKK-␥ gene causes female mice to develop patchy skin lesions with massive granulocyte infiltration and hyperproliferation and increased apoptosis of keratinocytes (24,25). Diseased animals show severe growth retardation and early mortality (24,25). Although the skin lesion of TAK1-KO mice was not patchy, the clinical appearance, histological findings, and early mortality were similar to those in IKK-␥ deficiency. The formation of patchy lesions in IKK-␥ deficiency is thought to be the result of a chimerism in the IKK-␥-deficient mouse. Furthermore, the epidermis-specific deletion of IKK-␤/IKK2 increased apoptosis and the abnormal expression of differentiation markers, including K6, K14, K10, and loricrin (26), resembling the characteristics of TAK1-KO mice. The development of similar skin phenotypes among mice with TAK1-, IKK-␤-, or IKK-␥-deficient epidermal keratinocytes indicates the disruption of a common pathway in these mice. Given that TAK1 activates the IKK complex, the signal from TAK1 to NF-B might be disrupted in the IKK-␤or IKK-␥-deficient epidermis.
Although apoptosis was almost 100% at 72 h in Fig. 7D, TAK1 expression was still 0.4-fold 48 h after Ax-Cre transfection (Fig.  7, A and B). There are three possibilities for this discrepancy: 1) The band in the Western blot of cells transfected with Ax-Cre was slightly lower than that of Ax-1W at 48 h (B); this may represent TAK1⌬, which lacks the ATP-binding site required for kinase activity (15), as in Fig. 1E. Therefore, the active TAK1 has already disappeared, and a small amount of inactive TAK1 (TAK1⌬) appeared instead. 2) Cells started to die 48 h after Ax-Cre transfection. Because the dead cells detached spontaneously or on washing the culture dishes, and only living cells were subjected to analysis, it is most likely that cells expressing 0.4-fold TAK1 were still alive, whereas the complete loss of TAK1 had resulted in the apoptosis of cells that were not included in the analysis. At 72 h, TAK1 knockdown may be complete, and all the cells underwent apoptosis.
3) The transfection efficiency of Ax-Cre might not be equal for all cells. In less transfected cells, TAK1 expression might not be reduced. This might cause the 0.4-fold TAK1 expression at 48 h. Moreover, the high expression of Cre caused the complete loss of TAK1 and apoptosis at 48 h, but these cells were not included in the analysis, as discussed in the second possible explanation. One reason or a combination of these reasons might cause this discrepancy.
A possible role for NF-B in negative cellular growth control via cell cycle regulation has been suggested (1,5,6). Growth inhibitory genes are induced by p65 in keratinocytes but not in other cell types (27). In the p65-deficient mouse, the number of Ki67-positive keratinocytes increased in the epidermis, and keratinocyte proliferation increased in culture (4). Based on the increased number of Ki67-positive cells in TAK1-KO mice seen in this study, we postulate that cell growth increased. However, the proliferation of TAK1-KO keratinocytes is reduced in vitro. IKK-␤-deficient keratinocytes are also hypoproliferative in vitro, despite an increased number of Ki67-positive cells in the epidermis (26). One possible mechanism to account for an increased number of Ki67-positive cells and decreased proliferation in vitro is the induction of apoptosis that overtakes proliferation. However, this cannot explain the discrepancy between TAK1 and IKK-␤ deficiencies and p65 deficiency. Similar phenotypic discrepancy has also been found between IKK␣/IKK1-deficient mice and others; the phenotypes of IKK␣-deficient mice are somewhat different from those of mice deficient in p65, IB kinase 2, or IKK␤. Mice lacking IKK␣ died perinatally and showed severely impaired limb outgrowth and abnormal epidermal differentiation (28,29). These results suggest that the phenotypic differences among these deficient mice result from the loss of individual upstream signaling molecules of NF-B that do not completely impair NF-B function. The loss of adhesion to the extracellular matrix in suspension culture strongly induces keratinocyte differentiation (3). Although the epidermis of TAK1-KO mice showed abnormal differentiation, the keratinocytes of TAK1-KO mice differentiated to express K1 in suspension culture. Similarly, IKK-␤-deleted keratinocytes retained the ability to differentiate in suspension culture despite the failure of epidermal differentiation (26). These data indicate that the regulation of keratinocyte differentiation by TAK1 and IKK-␤ is independent of the initiation of differentiation by the loss of adhesion to the extracellular matrix.
Keratinocytes in the basal cell layer of control mice expressed p50 and p65, whereas the expression of p50 and p65 in the basal cell layer of TAK1-KO mice was decreased. One possible cause of the decreased expression is abnormal differentiation. High expression levels of p50 and p65 are normally found in the keratinocytes of the basal layer. However, the keratinocytes in the TAK1-KO basal layer did not possess a characteristic feature of the basal layer, the expression of K5 and K14. This abnormal differentiation might cause decreased expression of p50 and p65. As a result of decreased p50 and p65 expression, the NF-B signal is scarcely transduced to the downstream signaling pathway, in addition to the lack of TAK1. In conclusion, TAK1 is essential in regulating keratinocyte growth, differentiation, and apoptosis.