Laminin α5 Is Required for Dental Epithelium Growth and Polarity and the Development of Tooth Bud and Shape*

In tooth development, the oral ectoderm and mesenchyme coordinately and reciprocally interact through the basement membrane for their growth and differentiation to form the proper shape and size of the tooth. Laminin α5 subunit-containing laminin-10/11 (LM-511/521) is the major laminin in the tooth germ basement membrane. Here, we have examined the role of laminin α5 (Lama5) in tooth development using laminin α5-null mouse primary dental epithelium and tooth germ organ cultures. Lama5-null mice develop a small tooth germ with defective cusp formation and have reduced proliferation of dental epithelium. Also, cell polarity and formation of the monolayer of the inner dental epithelium are disturbed. The enamel knot, a signaling center for tooth germ development, is defective, and there is a significant reduction of Shh and Fgf4 expression in the dental epithelium. In the absence of laminin α5, the basement membrane in the inner dental epithelium becomes discontinuous. In normal mice, integrin α6β4, a receptor for laminin α5, is strongly localized at the basal layer of the epithelium, whereas in mutant mice, integrin α6β4 is expressed around the cell surface. In primary dental epithelium culture, laminin-10/11 promotes cell growth, spreading, and filopodia-like microspike formation. This promotion is inhibited by anti-integrin α6 and β4 antibodies and by phosphatidylinositol 3-kinase inhibitors and dominant negative Rho-GTPase family proteins Cdc42 and Rac. In organ culture, anti-integrin α6 antibody and wortmannin reduce tooth germ size and shape. Our studies demonstrate that laminin α5 is required for the proliferation and polarity of basal epithelial cells and suggest that the interaction between laminin-10/11-integrin α6β4 and the phosphatidylinositol 3-kinase-Cdc42/Rac pathways play an important role in determining the size and shape of tooth germ.

The developing tooth is an excellent model for studying the molecular mechanisms of morphogenesis. Tooth development is regulated by sequential and reciprocal interactions between the neural crest-derived mesenchyme and the oral ectoderm (1)(2)(3). Tooth morphogenesis in mice starts at embryo day (E) 4 11.5, when the oral ectoderm invaginates into the underlying mesenchyme (4). Continuation of this process results in the formation of epithelial tooth buds at E13.5. Ectomesenchymal cells surrounding the bud form the dental papilla, which later develop into dentin-secreting odontoblasts and the tooth pulp. Following the bud stage, the tooth germ develops to the cap and bell stages, and dental epithelium differentiates into polarized and elongated enamelsecreting ameloblast.
The basement membrane is a thin, sheet-like extracellular matrix, which is present in most tissues. The basement membrane separates epithelium and mesenchyme and surrounds many cell types, such as muscle, fat, peripheral nerve, and endothelia. The basement membrane plays a role in organogenesis by supporting cells and providing signals for cell proliferation, migration, and differentiation (5)(6)(7). Basement membranes contain laminin, type IV collagen, perlecan, nidogen/entactin, and other molecules, but their structural compositions differ in tissues, developmental stages, and diseases, which reflects the diverse biological functions of the basement membrane. Laminin is a heterotrimeric glycoprotein, consisting of three genetically distinct ␣, ␤, and ␥ chains. There are five ␣ and three each of ␤ and ␥ chains, which form 15 isoforms of laminin (laminin-1 to -15) with different chain combinations (8,9). Laminin interacts with many extracellular matrix molecules and binds cells through cell surface receptors including integrins, syndecans, and ␣-dystroglycan. Laminin ␣ chains have a number of biological activities, including cell adhesion, migration, differentiation, angiogenesis, and tumor metastasis. The laminin ␣5 chain (Lama5) is expressed in many fetal and adult tissues and associates with the ␥1 chain (Lamc1) and either ␤1 (Lamb1) or ␤2 chains (Lamb2) to form laminin-10 (LM-511) and laminin-11 (LM-521 according to a recently proposed nomenclature) (9), respectively (10,11). Laminin-10/11 were shown to inhibit neurite outgrowth, whereas laminin-1 (LM-111), -2 (LM-211), and -4 (LM-221) promote neurite outgrowth, suggesting a unique role of laminin ␣5 in neural development (12). Mice lacking laminin ␣5 die in late development and are defective in anterior neuronal tube closure, digit formation, renal glomerulogenesis, lung development, and placentation (13)(14)(15). These results suggest that the ␣5 chain-containing laminin-10/11 has multiple functional roles for the development and establishment of tissue architecture.
In developing tooth, mRNA for laminin ␣3 and ␣5 are expressed by the dental epithelium, whereas mRNA for the other three laminin ␣ chains, ␣1, ␣2, and ␣4, are expressed primarily by the dental mesen-chyme (16 -19). Laminin-10/11 are the major laminins in the tooth basement membrane, which appears at the initial stage of tooth development. Laminin ␣5 expression is diminished when dental epithelium differentiates to enamel matrix-secreting ameloblasts. The basement membrane begins to degrade before the preameloblasts begin their differentiation to ameloblasts, and a new structure, the enamel matrix, is formed at the secretory stage (20). At the later terminal differentiation stage of ameloblasts, laminin ␣3 synthesis begins. Thus, laminin ␣5 is a major laminin ␣ chain in tooth basement membrane during early development. However, the role of laminin ␣5 in tooth development is not known.
Here we study the role of laminin ␣5 in tooth development using Lama5 knock-out mice. Mutant mice develop a smaller tooth germ and have defects in the enamel knot and cusp. In the tooth bud of mutant mice, distribution patterns of integrin ␣6␤4 are changed, and basal dental epithelial cell polarity is lost. Using primary dental epithelium culture, we show that PI3K and downstream molecules Rac1/Cdc42 regulate cell spreading and filopodia formation of the dental epithelium. These findings suggest that laminin ␣5 plays a role in cell polarity and proliferation of the dental epithelium and is essential for tooth morphogenesis.
Knock-out Mice for Lama5-We used two lines of Lama5-deficient mice. The original Lama5 knock-out mice were created by deleting a 3.5-kb fragment of Lama5 coding for amino-terminal domains of the protein (13). Heterozygous mice develop normally, but homozygous mice die at E14 -E18 with multiple tissue abnormalities (13). A second, independent line of Lama5 knock-out mice (Lama5 Trap mice) was generated from embryonic stem cells carrying an insertion into the intron of Lama5 of a gene trap vector (33). This vector, which was designed to trap genes encoding secreted and transmembrane proteins expressed in embryonic stem cells, lacks a promoter but has a splice acceptor and sequences encoding a transmembrane domain and a fusion between ␤-galactosidase and neomycin phosphotransferase II (␤geo) (34,35). Insertion into Lama5 resulted in a fusion protein containing the first 1763 amino acids of laminin ␣5 (domains VI through part of IVa), the transmembrane domain, and ␤geo (33). Although a major portion of the laminin ␣5 short arm is contained in the fusion protein, it is incapable of assembling with laminin ␤ and ␥ chains and is therefore nonfunctional. Indeed, Lama5 Trap homozygotes exhibit phenotypes identical to those observed in the original targeted knockout. 5 In addition, because the fusion protein is expressed under the control of endogenous Lama5 regulatory sequences, staining with 5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside (X-gal) can be used to assay which cells express Lama5 (33).
Preparation of Tissue Sections and Immunohistochemistry-E12.5, E13.5, E14.5, and E17.5 heads were dissected out and fixed with 4% paraformaldehyde in phosphate-buffered saline overnight at 4°C. Tissues were embedded into OCT compound (Sakura Finetechnical Co.) for frozen sectioning or dehydrated into xylene through a graded ethanol series and embedded in Paraplast paraffin (Oxford Laboratories). Sections were cut at 8 m by a cryostat (2800 Frigocut, Leica, Inc.) for frozen sections and at 10 m on microtome (RM2155, Leica, Inc.) for paraffin sections. For morphological analysis of molars and incisors, sections were stained with Harris hematoxylin (Sigma) and eosin Y (Sigma). LacZ staining for E13.5 heterozygous and homozygous heads was performed as described previously (36). Immunohistochemistry was performed on sections that were incubated in 1% bovine serum albumin/phosphate-buffered saline for blocking for 1 h and incubated with primary antibody. We used antibodies to collagen IV, integrin ␣3 (6B3), integrin ␣6 (GoH3), integrin ␤1 (9EG7), integrin ␤4 (346-11A), and PCNA. Primary antibodies were detected by fluorescein isothiocyanate-conjugated secondary antibodies (Jackson ImmunoResearch) for integrins or Histomouse immunostaining kit (Zymed Laboratory, Inc.) for collagen IV and PCNA. Tissue and cell samples of immunohistochemistry were examined under a fluorescence microscope and confocal microscope (Zeiss). Images were recorded and imported into Metamorph 4.5 and NIH Image 1.6 software.
In Situ Hybridization-In situ hybridization was performed on paraffin sections of E14.5 molars with the SureSite II system (Novagen). Shh and Fgf-4 cDNA for the RNA probe were gifts from Lillian Shum (NIH). Lef-1 cDNA (465 bp) was prepared by reverse transcription-PCR using RNA from E15 mouse molar tooth germ with Lef-1 primers, 5Ј-CCCATCCTCACTGTCAGGCGACAC-3Ј and 5Ј-CCTGTACCT-GAAGTCGACTCC-3Ј, and cloned into the pCR4-TOPO vector (Invitrogen). The Lef-1 cDNA was confirmed by DNA sequencing. The Lef-1 cDNA plasmid were digested by NotI for antisense probe and PstI for sense probe, and then cRNAs were synthesized by T3 RNA polymerase for antisense and T7 RNA polymerase for sense probes. [ 33 P]UTPlabeled RNA probes were prepared according to the manufacturer's instructions for the SureSite II system. The labeled transcripts were alkaline hydrolyzed to 150 -200 nucleotides and used for hybridization.
Dental Epithelial Cell Culture-Molars from E13.5 or E14.5 mouse embryos were dissected, and dental epithelial cells were prepared and cultured as described previously (20). Briefly, molars were treated with 0.1% collagenase, 0.05% trypsin, and 0.5 mM EDTA for 10 min, and the dental epithelium was separated from the dental mesenchyme. The dental epithelium was treated further with 0.1% collagenase, 0.05% trypsin, and 0.5 mM EDTA for 15 min, and then pipetted up and down well. Cells were cultured in keratinocyte-SFM medium supplemented with epidermal growth factor, bovine pituitary extract, 50 g/ml penicillin G, and 50 g/ml streptomycin (Invitrogen). To remove mesenchymal cells, cells were cultured in this medium for 7 days. During the culture, the medium was changed every 2 days.
Cell Proliferation Assays-Dental epithelial cells were cultured at 1.0 ϫ 10 5 cells/a 60-mm diameter dish coated with laminin-1 (Invitrogen), laminin-2 (Invitrogen), laminin-10/11 (Invitrogen), type I collagen (Cellmatrix, NITTA GERATIN, Japan), or fibronectin (Invitrogen) for 3 days. The cell proliferation assays were performed by incorporation of 5-bromo-2Ј-deoxyuridine (BrdUrd) using the BrdUrd labeling and detection kit (Roche Applied Science). Three days after culturing on each substrate, the cells were incubated with 10 M BrdUrd and 10% fetal bovine serum for 1 h. The cells were fixed with 4% paraformaldehyde and phosphate-buffered saline and permeabilized by 0.5% Triton X-100. The cells were incubated with 1% bovine serum albumin and Tris-buffered saline for 1 h and then with anti-BrdUrd antibody. Anti-BrdUrd antibodies were visualized by fluorescein isothiocyanate-conjugated secondary antibody. Nuclear staining was performed with DAPI.
Cell Spreading and Filopodia-like Microspike Formation-Approximately 1 ϫ 10 4 cells were plated onto glass coverslips coated with laminin-1 or laminin-10/11 and allowed to spread for 1, 3, and 12 h at 37°C in keratinocyte-SFM medium. To determine whether cell spreading and filopodia-like microspike formation on laminin-10/11 were dependent on integrins, cells were preincubated with 10 g/ml inhibitory antibodies against integrin ␣6 (GoH3), integrin ␣3 (6B3), integrin ␤1 Then cells were fixed as described above and stained with rhodamine-conjugated phalloidin for F-actin. Preliminary studies indicated that round cells that have not begun to spread have cell areas Ͻ400 m 2 and microspikes Ͻ2 m. Hence, we define those cells with areas Ͼ400 m 2 as spread, cell areas Ͻ400 m 2 as not spread, and those cells with Ͼ2 m microspikes as filopodia-positive cells.
DNA Transfection-For transfection, primary dental epithelial cells with ϳ80% confluence on glass coverslips in 24-well culture plates were transfected with Lipofectamine 2000 reagent (Invitrogen), according to the manufacturer's instructions. 1-g expression vectors for AU-5tagged attached active and dominant negative forms of RhoA (pcEF/ RhoQL and pcEF/RhoN19), Rac1 (pcEF/RacQL and pcEF/RacN17), and Cdc42 (pcEF/Cdc42QL and pcEF/Cdc42N17) were used. These plasmids were gifts from Silvio Gutkind (NIH). After a 3-h incubation with transfection reagents, cells were washed with keratinocyte-SFM medium, cultured for 2 days on a medium containing epidermal growth factor and bovine pituitary extract, and then used for analysis of cell spreading and filopodia-like microspike formation.
Organ Culture-Initiation stage (E12.5), bud stage (E13.5), and early cap stage (E14.5) tooth germs of mandibular first molars were dissected out from ICR mice and cultured for periods of up 7 days using chemically defined medium in a modification of Trowell's system (37)(38)(39). The explants were cultured using BGJb medium (Invitrogen) supplemented with 100 g/ml ascorbic acid, 10% fetal bovine serum, and penicillin-streptomycin on a filter (Millipore) in an atmosphere containing 5% CO 2 . The medium was changed every 2 days.

Defects in Tooth Germ Development of Lama5
Knock-out Mice-Two different lines of Lama5 mutant mice (13,33) were used. Using Lama5 Trap mice, which express the ␤geo gene (lacz/neo) under the control of the endogenous Lama5 promoter, we found that Lama5 was expressed specifically in the dental epithelium, but it did not appear in the dental mesenchyme (supplemental Fig. 1), agreeing with a previous report (17).
We next analyzed abnormalities of tooth germ development in E12.5-E17.5 embryos by hematoxylin and eosin staining and immunostaining for type IV collagen, the major basement membrane collagen. In E12.5 (initiation stage) and E13.5 (bud stage) heterozygous mice, the oral ectoderm thickens and invaginates into the ectomesenchyme (Fig.  1A, a and c). During these stages, the epithelium forms a single cell layer of the inner and outer epithelium, which surrounds multilayer dental epithelial cells including the stratum intermedium and stellate reticulum. However, in mutant mice, formation of a distinct single cell layer of the dental epithelium was not observed (Fig. 1A, b and d). In E14.5 control tooth germ (cap stage), the structure of the enamel knot, which determines the tooth shape including cusp, starts to form in the middle of the inner epithelium layer (Fig. 1Ae). In contrast, in mutant mice, such a structure is not formed, and the size of tooth germ is significantly smaller than control (Fig. 1Af). However, type IV collagen is immunostained on the basal laminar of the dental epithelium similar to that of control mice, indicating that the basement membrane is formed in mutant tooth germ despite the absence of the laminin ␣5 chain (Fig. 1A,  g and h). At E17.5 control tooth germ (bell stage), cusps form, and the inner dental epithelial layer expands. At this stage, the inner dental epithelium of control mice is polarized, forms an elongated shape, and starts to synthesize enamel matrix proteins essential for enamel formation after birth (Fig. 1B, a and b). In mutant mice, the inner dental epithelium is not polarized and remains in a round shape (Fig. 1B, d and  e). In addition, the basement membrane is discontinuous and disintegrated in some portions in the inner dental epithelium, but not the outer dental epithelial region (Fig. 2Bf). This is in contrast to the presence of a continuous smooth layer of the basement membrane surrounding the inner and outer dental epithelium in control mice (Fig. 1Bc).
Shh and Fgf-4 Expression and Cell Proliferation of Dental Epithelium-It has been suggested that tooth shape is regulated in part by signaling molecules from the enamel knots (40). To examine the mechanism involved in the abnormal tooth morphogenesis observed in Lama5 knock-out mice, we performed in situ hybridization for Fgf-4, Shh, and Lef-1. Expression of Fgf-4 and Shh mRNA was localized in the enamel knot in heterozygous mice (Fig. 2). In mutant mice, mRNA expression levels of these genes were significantly reduced (Fig. 2). In contrast, Lef-1 was expressed in the enamel knot and its surrounding dental mesenchyme in both heterozygous and mutant mice (Fig. 2).
Because Shh and Fgf-4 are known to promote cell proliferation, we examined cell proliferation in E14.5 tooth germ by PCNA staining (Fig.  3). In wild-type mice, the inner and outer epithelia and surrounding mesenchyme cells are positive for PCNA, indicating that these cells are proliferating, except for epithelial cells within the enamel knot (Fig.  3Aa). The cells in the enamel knot stop proliferation and die by apoptosis to form cusps (41,42). In mutant tooth germ, the total number of proliferating dental epithelial cells is reduced by about 50%. Epithelial cells in the region corresponding to the enamel knot continue to proliferate, indicating that a functional enamel knot structure is not formed in mutant tooth germ (Fig. 3Ab).
In mutant tooth germ, integrin subunits ␣6 and ␤4 were not intensely localized at the basal lamina but diffusely expressed around the cell surface (Fig. 3B, d and h). The expression patterns of ␣3 and ␤1 were the same in control and mutant mice. These results suggest that integrin ␣6␤4 binds laminin ␣5 in normal tooth germ and that without laminin ␣5, ␣6␤4 cannot be localized at the basal layer of the dental epithelium.
Promotion of Cell Proliferation of the Dental Epithelium by Laminin-10/11-In laminin ␣5-deficient mice, the tooth germ is small, and proliferation of dental epithelium is reduced. To test the role of laminin ␣5 in cell proliferation in culture, we prepared primary dental epithelial cells from tooth germs, plated them on dishes coated with substrates, and measured the number of proliferating cells by BrdUrd incorporation (Fig. 4). Laminin-10/11 increased the number of proliferating cells, whereas laminin-1 and laminin-2 did not. Fibronectin and collagen I also increased the number of proliferating cells but to a lesser extent than laminin-10/11. These results suggest that laminin-10/11 promotes the proliferation of dental epithelial cells in vivo, in agreement with the observation in mutant tooth germs.
Laminin-10/11 Promotes Cell Spreading and Filopodia Formation of the Dental Epithelium through Integrin ␣6␤4-In mutant tooth, cell polarity of dental epithelium is lost, and integrin ␣6␤4 localization is changed. Because cell spreading and filopodia formation regulate cell polarity (43)(44)(45), we next examined the activity of laminin ␣5 for cell spreading and filopodia formation in culture. We found that laminin-10/11 promoted cell spreading by 2-fold within a 1-2-h incubation when the dental epithelium was plated on laminin-10/11-coated plates (Fig. 5, A and B). This spreading was partly inhibited by anti-integrin ␣6 antibody but not by anti-integrin ␣3 antibody (Fig. 5C), suggesting involvement of integrin ␣6 in cell spreading. In contrast to laminin-10/  , heterozygous (a, c, e, and g) and mutant (b, d, f, and h) molars at the initiation (E12.5; a and b), bud (E13.5; e and f), and cap (E14.5, e-h) stages were stained with hematoxylin and eosin (a-f) and immunostained with anti-collagen IV antibody (g and h). At the initiation and bud stages, the dental epithelium invaginates into the ectomesenchyme and forms a single layer of the inner and outer dental epithelium, which surrounds a multilayer of dental epithelium including stellate reticulum and stratum intermedium (a and c); however, in mutant mice, there is no such distinct monolayer of the dental epithelium (b and d).
The size of tooth germ at the bud and cap stages is smaller in mutant (f) than heterozygous mice (e). The basal lamina is clearly formed in both heterozygotes (g) and mutants (h). B, E17.5 heterozygous (a-c) and mutant (d-f) molars were stained with hematoxylin and eosin (a, b, d, and e) and immunostained with anti-collagen IV antibody (c and f). In a heterozygous molar, three cusps are observed, and the inner dental epithelium is enlarged and polarized (a and b); however, in mutants, the size of molars and cusps is small, and the inner epithelium is round and not polarized (d and e). In a control molar, a distinct basement membrane layer is visualized by collagen IV immunostaining between the inner dental epithelium and mesenchyme (c), but in mutants, the basement membrane layer is disorganized, particularly the region contacting the inner dental epithelium (f). ie, inner dental epithelium; sr, stellate reticulum; dp, dental pulp; oe, outer dental epithelium. 11, laminin-1 did not promote cell spreading during a 1-3-h incubation (Fig. 5, A and B).
We next examined the effect of laminin-10/11 on filopodia formation of the dental epithelium. Laminin-10/11 promoted filopodia-like microspike formation, but laminin-1 did not (Fig. 6, A and B). Double-  d, f, and h). Integrins ␣3, ␣6, and ␤4 are expressed in the dental epithelium and strongly localized at the basal layer of the epithelium of heterozygotes (a, c, and g), whereas integrin ␤1 is expressed in both epithelium dental mesenchyme with higher expression in mesenchyme (e and f). In mutant molars, integrins ␣6 and ␤4 are not localized at the basal layers but distributed around the cell surface of dental epithelium (d and h). Localization of integrins ␣3 and ␤1 remained the same in homozygote and mutant mice (a, b, e, and f). ie, inner dental epithelium; sr, stellate reticulum; dp, dental pulp; oe, outer dental epithelium. Primary dental epithelial cells were cultured in dishes coated with various substrates. Cells were then incubated with BrdUrd for 1 h, and BrdUrd-positive cells were counted as described under "Experimental Procedures." Laminin-10/11 (LN1-/11) promoted the number of proliferating cells, whereas laminin-1 (LN1) and laminin-2 (LN2) did not. Fibronectin (FN) and collagen I (Col I) increased the number of proliferating cells to a lesser extent than laminin-10/11. These experiments were repeated at least three times with similar results. Cell proliferation in each matrix in primary cultured dental epithelial cells was compared with uncoated dishes. Statistical analysis was performed using analysis of variance ( * , p Ͻ 0.01).

FIGURE 5. Cell spreading of the dental epithelium on laminin-1 and laminin-10/11 and its inhibition by anti-integrin antibodies.
A, cell spreading on laminin-1 (LN1) and laminin-10/11 (LN10/11). Primary dental epithelial cells were plated on a dish coated with or without laminin-1 and laminin-10/11 for 1 h. Cells were incubated with laminin-1 or laminin-10/11 for 1 h. B, time course of cell spreading. Laminin-1 did not promote cell spreading during a 1-3-h incubation (b), whereas laminin-10/11 strongly promoted cell spreading. C, inhibition of cell spreading on laminin-10/11 by anti-integrin antibodies. Cell spreading is inhibited by anti-integrin ␣6 but not by anti-integrin ␣3.  Ab and B). C, inhibition of filopodia formation by anti-integrin antibodies. Filopodia formation is inhibited by anti-integrin ␣6 and ␤4 antibodies (Ad and C) but not by anti-integrin ␣3 and ␤1 antibodies (C).
immunostaining showed that integrin ␣6 and F-actin were colocalized in the microspike structure (data not shown). Formation of the filopodia-like structure was inhibited by anti-integrin ␣6 and ␤4 antibodies, but not by anti-integrin ␣3 and ␤1 antibodies (Fig. 6, A and C). These results suggest that the interaction of laminin-10/11 and integrin ␣6␤4 is important for cell spreading and filopodia formation of dental epithelium.
PI3K-Rac1/Cdc42 Signaling Pathways Regulate Dental Epithelial Cell Spreading and Filopodia Formation-To identify signaling pathways involved in cell spreading and filopodia formation mediated by laminin-10/11, we plated dental epithelium on laminin-10/11 with several kinase inhibitors and measured filopodia formation after a 1-h incubation (Fig.  7, A and B). Fig. 7A shows representative staining patterns for F-actin with and without wortmannin, and Fig. 7B represents quantitative inhibition data. PI3K inhibitors wortmannin and LY294002 inhibited filopodia formation on a laminin substrate. Herbimycin, a general tyrosine kinase inhibitor, also inhibited filopodia formation. However, Src inhibitor PP1, MEK inhibitor PD98059 and p38 inhibitor SB203580 did not inhibit filopodia formation.
Rho family tyrosine kinases make up some of the down stream molecules of PI3K and regulate cell polarity, cell spreading, filopodia formation and migration. To examine the effect of these kinases on cell spreading and filopodia formation, we transfected expression vectors for active or dominant negative forms of these kinases into dental epithelium (Fig. 8, A and B). On noncoated plates, there was no effect of these proteins on cell spreading and filopodia formation (data not shown). On laminin-10/11-coated plates, active Rac1 (RacQL) significantly promoted cell spreading, and dominant negative Rac1 (RacN17) inhibited cell spreading. The active form of Cdc42 (Cdc42QL) did not enhance cell spreading, and dominant negative Cdc42 (Cdc42N17) did not inhibit cell spreading significantly. We found the opposite results of these proteins on filopodia formation. Dominant negative Cdc42N17 strongly inhibited filopodia formation, and conversely active Cdc42QL strongly inhibited filopodia formation. Active RacQL also enhanced filopodia formation, and dominant negative RacN17 inhibited filopodia formation although to a lesser extent compared with Cdc42. In contrast, both active and dominant negative forms of Rho did not affect cell spreading and filopodia formation.
Reduced Size and Cusp Formation of Tooth Germ in Organ Culture by Anti-integrin Antibody and PI3K Inhibitor-To confirm further that integrin/PI3K pathways regulate tooth morphogenesis, we cultured tooth germs prepared from wild-type mouse embryos at three different stages, E12.5, E13.5, and E14.5. Antibody to integrin ␣6 inhibited the growth of tooth germ as well as cusp formation in E12.5 and E13.5 tooth germ culture ( Fig. 9 and Table 1). In E14.5 tooth germ culture, which had already formed the enamel knot and cusp structures, there was less inhibition by anti-integrin ␣6 antibody. Integrin ␤4 antibody showed less inhibition (Table 1). Wortmannin inhibited tooth germ growth and cusp formation similar to integrin ␣6. Other kinase inhibitors and integrin antibodies did not inhibit growth or cusp formation.

DISCUSSION
In this study, we demonstrate that the laminin ␣5 chain is essential for tooth morphogenesis. The tooth germ of Lama5-deficient mice is small and has defective cusp development. Normal tooth development initiates the formation of a dental laminar by thickening the oral ectoderm, which subsequently invaginates into the dental mesenchyme and differentiates into the inner and outer dental epithelia and other cell types. Fluorescence staining with rhodamine-conjugated phalloidin of filopodia of dental epithelial cells incubated on laminin-10/11 in the presence of wortmannin for 1 h. A color scale with numerical values equivalent to the intensity of phalloidin for all images is displayed. Red represents high F-actin staining, and blue is low staining. Filopodia formation of dental epithelial cells on laminin-10/11 is inhibited by wortmannin. B, inhibition of filopodia formation by dental epithelial cells on laminin-10/11 by kinase inhibitors. Cells were preincubated with kinase inhibitors and were plated on laminin-10/11 for 1 h. Filopodia formation on laminin-10/11 without kinase inhibitor was set as 100. General kinase inhibitor herbimycin and PI3K inhibitors wortmannin and LY294002 inhibit filopodia formation, but Src family kinase inhibitor PP1, MEK inhibitor PD98059, and p38 inhibitor SB203580 do not. The inner dental epithelium is polarized and forms a single cell layer, which reciprocally interacts with the underlying dental mesenchyme through the basement membrane. In mutant mice, the initial invagination of the dental ectoderm occurs, but the tooth bud size is small, and proliferation of dental epithelium is reduced, suggesting the role of laminin ␣5 in cell proliferation. Consistent with the in vivo observation, laminin ␣5-containing laminin-10/11 promotes the proliferation of primary dental epithelial cells in culture. At a later stage (E17.5), the basement membrane in the inner dental epithelium is patchy and disintegrated in some parts, areas that perhaps receive intense mechanical pressure. These results suggest that laminin-10/11 is involved in structural stability of the basement membrane. Lack of basement membrane integrity may also contribute to the defects in cell growth and differentiation.
In laminin ␣5-null mice, expression levels of Shh and Fgf-4 are reduced in the enamel knot. At the cap stage of normal tooth development, Shh and Fgf-4 are expressed primarily by dental epithelium in the enamel knot and promote proliferation of the epithelium and mesenchyme surrounding the enamel knot (46 -48). In normal tooth development, both dental epithelium and mesenchyme express Gli1, a transcriptional effector of hedgehog signaling, and Shh signaling promotes cell proliferation. On the other hand, the epithelium within the enamel knot does not express Gli1 and eventually undergoes apoptosis. In mutant mice, reduced expression of Shh by the dental epithelium in the enamel knot likely causes reduced proliferation of the surrounding epithelium and mesenchyme, resulting in part in smaller tooth germs. Skin of Lama5-null mice shows a failure of hair germ elongation with reduced expression of Shh, and Gli1, similar to tooth (49).
Laminin has been implicated in epithelial morphogenesis (50,51). For example, laminin regulates Madin-Darby canine kidney cell polarity in epithelial cysts (43,52,53) and salivary gland morphogenesis (54,55). Laminin-1 secreted by myoepithelium requires polarity of breast luminal epithelial cells (56). Laminin ␤1 and ␥1 knock-out mice cannot produce laminin-1 and laminin-10 and die at the peri-implantation stage (57,58). Laminin ␣1 knock-out mice survive slightly longer than laminin ␤1 and ␥1 knock-out mice but die during early embryogenesis prior to the onset of tooth development (58). During tooth development, dental epithelium produces laminin ␣3 and ␣5, whereas the dental mesenchyme produces other laminin ␣ chains. Laminin ␣3 (Lama3)-null mice showed abnormal late differentiation of ameloblasts resulting in enamel hypoplasia (18). Ameloblasts in Lama3 mutant mice are defective in cell polarity and enamel matrix production. The expression pattern of laminin ␣3 and ␣5 is different: laminin ␣3 is expressed by differentiated ameloblasts from the secretory stage to the maturation stage, whereas laminin ␣5 is expressed by preameloblasts in the presecretory stage but not by secretory ameloblasts. Thus, both laminin ␣3 and ␣5 regulate dental epithelial morphology and phenotype, but laminin ␣5 deficiency causes earlier, more severe abnormalities during tooth germ formation. A study with laminin ␣2 knock-out mice revealed that laminin ␣2 is not required for dental epithelial cell growth and differentiation but is essential for odontoblast differentiation, consistent with the expression pattern of laminin ␣2 (19).
Cell spreading requires actin polymerization and reorganization, extension of lamellipodia or filopodia, and formation of new integrinsubstrate adhesions (59,60). These cellular processes are associated with cell polarization of epithelial and neuronal cells and migration of tumor cells (43,45,52,53,61). In Lama5 knock-out mice, cell polarity of the dental epithelium is disturbed at the early stage of tooth germ development, and a single inner epithelial cell layer is not formed. It is reported in cells derived from aggressive, late stage tumors that integrin ␣6␤4 is localized at the membrane protrusions associated with the migration front such as filopodia, lamellipodia, and retraction fibers (62). We found that laminin-10/11 promotes spreading and filopodia formation of normal dental epithelium in culture and that integrin ␣6 is present at filopodia and colocalizes with F-actin of these cells. Antibodies against integrin ␣6␤4 inhibit spreading and filopodia formation of the dental epithelium on laminin-10/11. In tooth germ organ culture, anti-integrin ␣6 and, to a lesser extent, anti-integrin ␤4 antibody inhibit tooth germ development and cusp formation. The tooth germ that incubated with the antibody to integrin ␣6 fails to form a single cell layer of the polarized inner dental epithelium (data not shown) similar to tooth germ of laminin ␣5 knock-out mice. These results suggest that integrin ␣6␤4 mediates laminin-10/11-promoted cell spreading and filopodia formation of the dental epithelium in cell. It is also possible that ␣-dystroglycan is involved in these processes because ␣-dystroglycan binds laminin-10/11 (23,24) and is involved in controlling cell polarity (63).
Previous studies demonstrated that PI3K is activated by integrin-

TABLE 1 Inhibition of cusp formation and molar size in tooth germ organ culture
Tooth germs were dissected from E12.5, E13.5, and E14.5 embryos and cultured in the presence or absence of anti-integrin ␣6 antibody (Ab) and kinase inhibitor wortmannin for 1 week. The table shows quantitative data for tooth size and cusp number from the organ culture experiments such as shown in Fig. 9. The number of clearly formed cusps, such as shown in Fig. 9, were scored. Only 2 tooth germs of 12 from E12.5 embryos formed cusps when treated with anti-integrin ␣6 antibody. Similar inhibitory results were observed in tooth germ from E12.5 and 13.5 embryos by the wortmannin treatment. The largest width of tooth germ was measured and scored. About 40 -60% of tooth germs were reduced in size by treatment with anti-integrin ␣6 antibody and wortmannin. Other integrin antibodies and kinase inhibitors PD98059 and SB203580 showed no inhibition.  Integrin ␤4 antibody showed weak inhibition (data not shown and Table 1). Integrin antibodies to ␣3 and ␤1 and kinase inhibitors PD98059 and SB203580 showed no significant inhibition. mediated cell adhesion and is required for cell spreading and polarity of certain cell types (64 -66). Furthermore, ␣6␤4 integrin activates PI3K more effectively than other integrins (29,32), and ␣6␤4 signaling strongly enhances migration and invasion of certain tumor cells (30,62,67,68). Activation of PI3K by integrin ␣6␤4 may stimulate the function of other integrins, especially ␣3␤1, which is also important for epithelial adhesion and migration (68). In dental epithelial cells cultured on laminin-10/11, PI3K inhibitors wortmannin and LY294002 specifically inhibit filopodia formation and cell spreading. In tooth organ culture, they also inhibit cusp formation in a stage-specific manner, similar to the inhibition observed with anti-integrin ␣6 antibody. Akt, a downstream molecule of PI3K, is activated when the dental epithelium is plated on a laminin-10/11-coated dish, and this activation is inhibited by wortmannin and anti-integrin ␣6 antibody (data not shown). These results suggest that the interaction between laminin-10/11 and integrin ␣6␤4 is necessary for activation of the PI3K/Akt pathway. The interaction of laminin-10/11 and dystroglycan may also regulate the activation because dystroglycan was reported to mediate the PI3K/Akt pathway (69). However, expression of dystroglycan in dental epithelium is unknown.

Cusp formation
It has been shown that Rac1 and Cdc42 are regulators of cell spreading and lamellipodia and filopodia formation (70 -73). The ␣6␤4 integrin promotes activation of Rac and RhoA GTPases (29,67). Activation of Rac is linked to the integrin ␣6␤4-mediated stimulation of PI3K (29). We also found that Rac1 and Cdc42 regulate laminin-10/11-mediated cell spreading and filopodia formation of the dental epithelium. However, Rac1 affects cell spreading more effectively than filopodia formation when cells are plated on laminin-10/11, whereas Cdc42 affects filopodia formation more effectively than cell spreading. This may reflect the distinct role of Rac1 and Cdc42 in cytoskeletal reorganization (73). A recent study shows that orientation of apical polarity in developing Madin-Darby canine kidney epithelial cysts requires Rac1 and laminin. Dominant negative Rac1 alters the assembly of endogenous Madin-Darby canine kidney laminin and causes a striking inversion of apical polarity. These results suggest the important role of Rac1/Cdc42 and laminin in cell polarity, which is consistent with our findings with the dental epithelium.