Ectodysplasin (or Eda), a soluble member of the tumor necrosis factor (TNF) family, its receptor Edar, and the death-domain containing adapter protein, Edaradd, are part of a signaling pathway that is required for the development of ectodermal organs (reviewed in Thesleff and Mikkola, 2002). Mutations in the genes encoding for these proteins cause in humans the X-linked and autosomal hypohidrotic ectodermal dysplasia syndromes (HED), in which the development of teeth, hair, and many glands, specifically the sweat glands, is defective. The corresponding mouse mutants with similar phenotypes are called Tabby, downless, and crinkled. The number and shape of teeth in these mice is affected (Mouse Genome Database [MGD], 2004). Wild-type mice have one incisor and three molars in each jaw quadrant, whereas in Tabby, downless, and crinkled (Ta/dl/cr) mice incisors and third molars are missing, albeit with an incomplete penetrance. However, they consistently have malformations in the shape of their molars with a reduced number of cusps. Surprisingly, some Tabby heterozygotes have an ectopic tooth mesial to the first molar.
Ectodermal organs (teeth, hair, feathers, scales, eccrine glands, etc.) share common developmental mechanisms in which two adjacent tissue layers, the epithelium and mesenchyme, signal to each other consecutively. An early signal to govern tooth morphogenesis comes from the oral epithelium. Local thickenings, dental placodes, are formed in the epithelium, and this formation is followed by condensation of mesenchymal cells underneath the placodes (embryonic day [E] 11). The epithelium then buds into the mesenchyme (E12). The odontogenic potential or the ability to instruct tooth morphogenesis shifts from the epithelium into the mesenchyme. Transient signaling centers form in the epithelium, first in the placodes and later in the primary and secondary enamel knots, and contribute to the shaping of the future tooth crown and its cusps (Jernvall and Thesleff, 2000). The association between enamel knot signaling and cusp patterning in rodent teeth has been demonstrated (Jernvall et al., 2000). Postnatally, specialized dental hard tissues, enamel and dentin, are formed by the epithelial ameloblasts and mesenchymal odontoblasts, respectively.
Several pieces of evidence indicate that ectodysplasin-Edar signaling is important for the patterning of ectodermal organs. First of all, the expression patterns of Eda and Edar suggest crosstalk between ectodermal compartments and regulation of signaling center activity. Eda and Edar are coexpressed throughout the simple ectoderm before the initiation of tooth and hair development (Tucker et al., 2000; Laurikkala et al., 2001, 2002). When the ectodermal placodes form, Edar becomes strictly confined to the placodes, whereas Eda shows complementary expression. In developing molar teeth, Edar is expressed reiteratively in the ectodermal signaling centers, first in the placode, then in the primary enamel knot, and later in the secondary enamel knots (Tucker et al., 2000; Laurikkala et al., 2001). Edar is localized also into mammary gland placodes (Pispa et al., 2003). Hence, there is a striking association between Edar expression and ectodermal signaling centers, suggesting functions in the patterning of ectodermal organs. Second, the mutant phenotypes support a role in patterning. The Tabby mouse molar phenotype has been associated with defective enamel knot signaling (Pispa et al., 1999). More recently, we overexpressed Eda in transgenic mice with the ectodermal promoter, keratin 14 (K14) (Mustonen et al., 2003). In these mice, Edar is expressed in its localized, endogenous pattern. Therefore, overexpression of its ligand, Eda, increases Edar activity only at the Edar-expressing sites. As a consequence of this increase, the formation of many ectodermal organs is stimulated. Most significantly, ectopic teeth and mammary glands are found. Hair follicle formation occurs continuously during embryogenesis in contrast to the sequential formation of hair follicles seen in wild-type embryos. In addition, the activity of many organs is enhanced, for example, nails are longer and sweat glands produce more sweat.
To further analyze the functions of ectodysplasin-Edar signaling, we have generated transgenic mice where the K14 promoter directs Edar expression to the cells of the epidermis, the developing hair follicle and the epithelium of the tooth germs. Unlike the localized expression of endogenous Edar, ectopic Edar is expressed throughout the developing epithelium in teeth and hairs. We show that this misexpression of Edar alters tooth shape and influences the cusp pattern. Low doses of transgenic Edar expression reduce the number of cusps, whereas high doses increase it. Furthermore, Edar misexpression disturbs enamel structure.
Effects of Edar Misexpression Depend on the Amount of Edar Protein
Keratin 14 (K14) promoter has been used widely in mice to direct ectopic gene expression into the basal cell layer of the epithelium, into the outer root sheath (ORS) of hair follicles and to the dental epithelium (Wang et al., 1997; Dassule et al., 2000). To facilitate the analysis of the downstream effects of Edar signaling, the mouse Edar open reading frame (ORF), tagged C-terminally with myc and 6xHis epitopes, was cloned under the K14 promoter (Fig. 1A) and injected into the pronucleus of fertilized eggs. Nine founders were obtained, of which all, apart from one, had defects in tooth and/or hair development. Three of these were established as colonies. Of interest, there was remarkable phenotypic variation between founders (Table 1).
Additional lingual cusp in lower second molar (low penetrance)
Plicae digitales, nails, mammary glands
Additional lingual cusp in lower second molar
Additional lingual cusp in lower second molar
Missing third molars
Supernumerary tooth in maxilla, no hypoconulids in lower first molars
The transgene copy number in five different F0 animals (or their F1 progeny) was analyzed with Southern blotting and found to range from one copy to ca. 20 copies (data not shown). The expression levels of the Edar protein in the same transgenic lines were studied with Western blotting. Skin samples of adult mice were homogenized, run on an sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gel and blotted with antibodies recognizing either mouse Edar or the myc epitope. The protein detected in the blot by both antibodies ran as a single band of approximately 60 kDa (Fig. 1B). This corresponded to the molecular weight of Edar protein expressed in cultured cells. The difference from the 49 kDa predicted molecular mass of Edar may be due to glycosylation, as the Edar ORF contains potential N-glycosylation sites (Koppinen et al., 2001). As a control, an antibody recognizing keratin 10 was used to ensure the equal protein amount of different skin samples. Similar results were obtained with skin samples from 1-day postnatal pups (data not shown). The transgenic Edar protein was also visualized by immunohistochemistry in the skin epithelium of newborn pups by using the anti-myc antibody (data not shown). In situ hybridization for Edar confirmed the presence of Edar transcript throughout the tooth epithelium (Fig. 1C).
A correlation was found between the phenotype and the K14-Edar expression level. The eight founders with an aberrant phenotype were grouped into three different categories: (1) strong expression, (2) weak expression, and (3) intermediate phenotype (Table 1). The strong expression group consisted of three founders, of which one was established as a colony. All three founders had similar, although not identical, phenotypes. Southern and Western analyses of two of these lines showed that they had ca. 20 copies of the transgene and the highest levels of Edar expression (D42 and F77 in Fig. 1B). The weak expression group contained two founders whose phenotypes differed from each other only slightly. Both had four to six copies of the transgene, and they expressed Edar protein at levels not detectable with our Western blot analysis (F68 and F69 in Fig. 1B). However, Edar mRNA was shown by in situ hybridization to be misexpressed throughout the epithelium (data not shown). There was very little variation in the phenotype within a transgenic line (strong or weak), indicating that the penetrance for most phenotypic features was close to 100%. In addition, three founders had intermediate phenotypes that could not be classified into either the strong or weak expression group. It is likely that the Edar expression levels of these animals fall between those of the strong and weak expression groups or between the weak expression group and wild-type levels of Edar. However, we were not able to obtain transgenic progeny from these animals, and the expression levels were not analyzed in the founders.
In short, several ectodermal organs were affected by Edar misexpression, including teeth, hair, nails, plicae digitales, and mammary glands. In this study, we will describe the effect on tooth development; effects on other organs will be reported elsewhere.
Edar Misexpression Affects the Tooth Cusp Number and Shape
Edar misexpression disturbed tooth morphogenesis first by affecting the molar cusp pattern, which is normally well-conserved in mice. Wild-type lower first molars usually have seven cusps, second molars five, and third molars three. In the weak expression group, the cusp number was reduced. The distal cusps B3 and L3 (terminology according to Grüneberg, 1966) of the lower second molar were replaced by a single large cusp. In addition, the most distal cusps, 4s or the hypoconulids, of the lower first and second molars were missing (Fig. 2A,B). The cusp number in the lower third molars and the upper molars was normal. In the strong expression group, the cusp number had increased as an additional cusp had formed lingually between the L2 and L3 cusps in the lower second molar (Fig. 2C). This finding occurred either consistently or with a lower frequency, depending on transgenic founder. The cusp numbers in the lower first and third molars were normal. The upper third molars lacked the most distal cusp (data not shown). Of the intermediate group founders that could not be classified into either the weakly or strongly expressing groups, two had missing hypoconulids (Table 1).
The second effect of Edar misexpression was on the shape of the molars, which was analyzed by measuring the buccolingual width and mesiodistal length of each mandibular molar. In the weak expression group, all molars were narrower and shorter than the controls. In the strong expression group, the first molar was shorter and the second and third molars were wider than the controls. In both groups, the third molar was wider in relation to its length (Fig. 2D,E). In addition, all intermediate phenotype founders had abnormal molar shapes, which were different from each other and which did not correspond to the molar shapes of the weak or the strong expression group (data not shown). However, the age of the mice and subsequent wear of the molars hindered thorough analysis.
Third, Edar misexpression altered the tooth number, albeit with a low frequency. One intermediate founder had missing third molars and another had a supernumerary molar in the maxilla (Fig. 2F,G). This finding resembles the homozygous and heterozygous Tabby phenotypes, respectively (Grüneberg, 1966). Similar supernumerary molars were seen also in K14-Eda mice, although with a much higher frequency (Mustonen et al., 2003).
Enamel Formation Is Defective in K14-Edar Mice
The teeth of the strongly Edar-expressing transgenics wore faster than those of the control animals. In the strong expression group, two founders (A11 and D42) had one pair of incisors (upper or lower) that wore out completely, resulting in overgrowth of the opposite pair. At closer inspection, the crowns of molar teeth were also prematurely worn (Fig. 3A,B). In the third strongly expressing transgenic line (F77), the tooth wear was slower and preferentially affected the molars. Accelerated wear was not noted in the weak expression group, indicating that the severity of this phenotypic feature correlates with the level of transgene expression.
The enamel of one lower incisor of a strong expression group founder (D42) was studied histologically. In mouse incisors, enamel is formed only on the labial surface, whereas the lingual “root analogous” surface is enamel-free. With the proceeding enamel maturation, the proportion of the mineral phase increases at the expense of the organic matrix in the incisal direction. At cross-section, the transgenic incisor was thinner than the wild-type incisor, and its shape was oval throughout the tooth length (Fig. 3C,D). The amount of mineralized enamel and more basally, unmineralized enamel matrix was minimal. As seen in methacrylate-embedded ground sections, mineralized enamel was present as a thin layer only on part of the labial dentin surface (Fig. 3E,F). The crosswise prismatic pattern, characteristic of rodent incisors, was irregular. The outer enamel layer was absent as was the yellow–brown color of the outermost enamel, which is due to secretion of iron compounds by the ameloblasts before their degeneration. In proportion to tooth size, the thickness of dentin corresponded to normal and the tubular pattern was regular. In line F77, where the teeth wore out at a slower rate, enamel was present but it was thinner than in the wild-type teeth (data not shown).
As the methacrylate-embedding technique for mineralized tissues fails to preserve soft tissue morphology, the bone-embedded part of the transgenic incisor was demineralized with ethylenediaminetetraacetic acid (EDTA) and analyzed as paraffin sections. This procedure removes fully mineralized enamel but allows histological analysis of dentin and enamel matrix. In the control incisor, the scarce enamel matrix had disintegrated during the demineralization (Fig. 3G). This finding occurs frequently in wild-type samples and reflects the proportional abundance of fully mineralized enamel. On the transgenic labial dentin surface, a thin layer of enamel matrix was visible, suggesting a lower than normal level of mineralization (Fig. 3H). Polarized enamel-secreting ameloblasts were not visible.
Manipulating Transgene Copy Number in K14-Edar Mice Supports the Dose-Dependency of Edar Signaling
To further study the effect of the amount of Edar signaling on the tooth phenotype in K14-Edar mice, we crossed a strong expressing line (F77) into a weak expressing line (F69). Ten pups were born, and they were killed and analyzed at 6 weeks of age. Genotyping with Southern blotting showed that the transgene copy numbers could be grouped into four groups: none or one copy (three mice), low copy (one mouse), high copy (four mice), and the highest copy number (equal to the sum of the low and high copy numbers, two mice; data not shown).
Macroscopic analysis of the teeth showed that the differences in copy number corresponded to phenotypic differences (Fig. 4). Mice carrying no transgene had wild-type molars and incisors. The mouse with a low transgene copy number was similar to the weak expression group, having a reduced number of cusps and no enamel defects. Mice with a high copy number resembled the strong expression group. Of the eight mandible halves examined, five had a clear ectopic lingual cusp in the second molar. Enamel was not significantly reduced, most likely due to the young age of the mice. The two mice carrying transgene copies from both weak and strong expression lines had a tooth shape similar to the strong expression group. Ectopic lingual cusps were seen in all the lower second molars. The enamel was clearly reduced, resulting in chalky-white, malformed incisors.
These results indicate that there is an inverse relationship between the number of Edar transgenes and consequently the amount of Edar protein, and the amount of enamel. They also show that the reduction in cusp number seen in the weak expression group can be overcome by increasing the amount of Edar signaling.
Misexpressed Edar Signaling May Affect Tooth Shape by Interfering With Enamel Knot Signaling and Patterning Of Cusps
Edar misexpression throughout the dental epithelium affected the overall shape of the tooth crown and the number of tooth cusps. In addition, it had a minor impact on the number of teeth. In principle, these changes may occur from increased Edar signaling at locations where endogenous Edar is expressed, i.e., either in the oral epithelium, the placode or the enamel knots. The changes may also result from Edar signaling outside the signaling centers. Edar signaling in K14-Eda and K14-Edar mice is presumably similar in the oral epithelium before placode formation, both endogenous Eda and Edar being expressed there. However, K14-Eda and K14-Edar mice have a different dental phenotype. K14-Eda mice have an ectopic mesial tooth. The molars are wider, but the cusp pattern is relatively normal (Mustonen et al., 2003). Only one K14-Edar founder had an ectopic tooth, and the cusp number was either reduced or increased depending on dose. This finding suggests that increased Edar signaling before placode formation is not a likely cause for the disturbances seen in K14-Edar tooth shape and cusp number. In principle, increased Edar signaling could initiate novel tooth placode formation in the dental lamina, and result in the ectopic tooth seen in one K14-Edar and most K14-Eda mice. We have studied the early mandible in K14-Eda mice and see no evidence of ectopic placodes in the dental lamina (Mustonen et al., manuscript in preparation). Instead, supernumerary teeth seem to be formed by maintaining an existing placode that normally regresses in wild-type mandible. Thus, it seems that Edar signaling before placode formation is not sufficient to change dental fate.
The link between enamel knot signaling and cusp patterning has been demonstrated. In two different rodent species, the size of the primary enamel knot is linked to the timing and location of successive secondary enamel knots, which then determine the position of the cusps (Jernvall et al., 2000). In Tabby molars, a small primary enamel knot results in fused and fewer secondary enamel knots and, consequently, in a decreased cusp number (Pispa et al., 1999). Increased signaling in the primary enamel knot may have caused the additional cusp seen in the K14-Edar strong expression group. However, similar increased signaling in K14-Eda mice does not cause similar cusp effects, arguing against this possibility. We find that the likeliest explanation for the changes in K14-Edar mice cusp number and tooth shape is ectopic signaling outside the signaling centers. It is possible that Edar signaling throughout the dental epithelium affected the function of the primary enamel knot. As a consequence, the secondary enamel knots may have been defined irregularly. Obviously further research is needed to analyze the size and positions of the primary and secondary enamel knots in the K14-Edar mice.
The effects of different doses of Edar signaling were practically opposite on tooth patterning. Why different amounts of Edar had such effects on tooth shape is an intriguing question. The similarity of the weak expression group phenotype to the Tabby tooth phenotype raises the possibility that the transgenic receptor is acting as a dominant negative molecule. The receptor was tagged with myc and 6xHis epitopes, but we do not believe that the tags affect the signaling capability. In cultured cells, the same transgenic construct has been shown to activate the transcription factor NF-κB, a common response of TNF receptors (Koppinen et al., 2001). Sequestering of ligand by the ectopic receptor to outside of the signaling centers has been suggested to cause a lack-of-function phenotype in mice misexpressing the tyrosine kinase Ret in kidney (Srinivas et al., 1999). A similar mechanism can of course account for the reduction in cusp number seen in the weak expressing K14-Edar mice. In wild-type molars, ectodysplasin is expressed in the outer enamel epithelium, where it is cleaved from the membrane. Diffusion or transport of ectodysplasin into the Edar-expressing enamel knot is required in order for signaling to occur. We cannot rule out the possibility that, in our transgenic weak expression group, ectopic receptors in the dental epithelium sequester the ligand before its transport into the enamel knot. As no molecular markers have been found that can detect a lack of Edar signaling (Pispa et al., 1999; Laurikkala et al., 2001), we have no means to detect the level of signaling in these teeth. However, a lack-of-function phenotype does not seem likely given that increasing the dose of receptor in the strong expression group reverses this effect by increasing the cusp number. We do not think that this reversion could be explained by increased ligand-independent Edar signaling, because, to our knowledge, there are no reports of this type of nonmodified TNF receptor activity in vivo. Furthermore, the effect of transgene integration site was ruled out, as the weak expression group phenotype was found in two separate founders.
We favor the possibility that ectopic Edar signaling affected the primary enamel knot function differentially, depending on the Edar dose. Different downstream targets of Edar signaling may require different thresholds of Edar protein for their activation. Evidence from several developmental contexts has suggested that high and low concentrations of signaling factors may have different and even opposite effects. For example, studies on feather and neural tube development in chick embryos indicate that high concentrations of bone morphogenic proteins (BMPs) may activate target genes contrary to the repression seen at lower concentrations (Jung et al., 1998; Timmer et al., 2002). Recently, a model of cusp pattern formation was presented in which BMP acts as an activator of enamel knot formation, and its inhibition is required for lateral inhibition (Salazar-Ciudad and Jernvall, 2002). Of interest, the localization of the additional lingual cusp seen in the strong Edar expression group is in line with the prediction of this model if the activity of the activator is increased (unpublished results). Therefore, the balance of signaling by different, possibly counteracting, downstream targets of Edar is likely to determine the final patterning of teeth.
Enamel Formation Is Disturbed by Ectopic Edar Signaling
Misexpression of Edar disrupted enamel formation severely in the K14-Edar strong expression group. Histological analysis showed that the enamel matrix was either absent or poorly mineralized. Enamel matrix is composed of several unique enamel proteins and of proteolytic enzymes degrading the matrix during its final mineralization, and it is deposited by ameloblasts differentiating from the dental epithelium (Ten Cate, 1994). Hypoplastic enamel may result from defects in quality or quantity of enamel proteins (Lagerström et al., 1991) or from weakened integrity of the ameloblast cell layer or weakened contacts between ameloblasts and the deposited matrix (Ryan et al., 1999). The epithelial cells neighboring the ameloblast layer regulate ameloblast functions, and recently defective SHH signaling from these epithelial cells was implicated as a cause of deficient enamel formation (Dassule et al., 2000). Keratin 14 is expressed in the entire dental epithelium in postnatal teeth (Dassule et al., 2000). Misexpressed Edar, thus, is present in ameloblasts but also in the neighboring epithelial cells, and it is not clear how amelogenesis is affected in K14-Edar mice. K14-Eda mice have similar defects on enamel formation (Mustonen et al., 2003). It has also been recently shown that mice expressing Wnt3a under the K14 promoter have defective enamel (Millar et al., 2003). Because Wnt signaling regulates expression of ectodysplasin (Laurikkala et al., 2001, 2002; Durmowicz et al., 2002), it is possible that the effects of increased Wnt signaling are mediated by increased Edar signaling. No major enamel defects have been reported in the teeth of HED patients or Ta/dl/cr mice. Although this finding suggests that ectodysplasin-Edar signaling is not necessary for enamel development, it is possible that there are compensatory pathways.
In conclusion, we propose that Edar activity throughout the dental epithelium in K14-Edar mice, specifically outside the signaling centers, disturbs dental patterning and cusp formation. Analysis of transgenic lines indicates that the precise localization and amount of Edar is important in specifying the cusp pattern. During evolution, differential localization of Edar signaling may have contributed to the changes in tooth shapes. In addition, ectopic Edar signaling can disturb enamel differentiation in a dose-dependent manner. Edar is known to activate the NF-κB pathway (Yan et al., 2000; Koppinen et al., 2001; Kumar et al., 2001), but no Edar target genes have been identified. The K14-Edar mice will provide a tool for analysis of these downstream targets.
Generation and Genotyping of Transgenic Mice
Mouse Edar ORF, followed by sequences coding for C-terminal myc and 6xhistidine tags and a stop codon, was excised from pEF1-dlwt1 (Koppinen et al., 2001) with EcoRI and PmeI. The blunt-ended fragment was cloned into the BamHI site of K14-BG (Gat et al., 1998) producing the K14-BG-DL plasmid. The transgene was released from K14-BG-DL by digestion with EcoRI and HindIII and microinjected into the pronucleus of FVB mouse fertilized eggs. Transgenic animals were identified by polymerase chain reaction (PCR) using primers 5′-ACATCCTGGTCATCATCCTGCC-3′ and 5′-ACAGGACATGTAGGGCTCCTCC-3′. For transgene copy number analysis genomic DNA was digested with EcoRI, Southern blotted and probed with a 220-bp PCR fragment from mouse Edar cDNA. (PCR primers: 5′-CCTTGCCAAGAGCTTTGG-3′ and 5′-GATCCTCGAGTCAGGACGCAGCTGGGG-3′.) The blot was normalized using as probe a mouse genomic DNA fragment from the Tabby promoter.
For the production of the anti-Edar antibody (DL3), the sequence coding for the intracellular domain of mouse Edar (amino acids 211-448) was excised from mouse Edar cDNA (Koppinen et al., 2001) and cloned into a bacterial expression vector pTAT (Mikkola et al., 1999). The resulting 6xhis-tagged thioredoxin fusion protein was expressed as previously described (Mikkola et al., 1999) and purified under denaturing conditions using TALON matrix as recommended by the manufacturer (Clontech). The purified protein was dialyzed against phosphate buffered saline, and two rabbits were immunized with the antigen. The same fusion protein was used to affinity purify the antisera (Mikkola et al., 1999).
Skin samples were placed directly into 10 volumes of sample buffer (5% β-mercaptoethanol, 2% SDS, 20% glycerol, 5 mM EDTA, 60 mM Tris-HCl pH 6.8), then boiled for 5 min, vortexed, and homogenized by using syringe and needle, boiled again for 2 min, and vortexed, followed by centrifugation for 2 min at 13,000 rpm. Protein concentrations were determined with the Bio-Rad protein assay. Sixty micrograms of proteins were separated by 10% SDS-PAGE, transferred onto a Hybond-C-extra membrane (Amersham), and blots were developed by using enhanced chemiluminescence (Amersham). Edar was detected with the rabbit antibody, DL3, and with an anti-myc tag antibody (Upstate Biotechnology). Keratin 10 was detected by using a polyclonal antibody against MK10 (BAbCO).
Analysis of the Adult Teeth
Skeletal preparations of jaws were made as described previously (Pispa et al., 1999). Measurements of the maximum length and width of each lower molar were made by photographing the teeth coronally with a digital camera and analyzing the pictures with NIH Image software. The measurements were visualized as box plots, the height of the box showing 50% of the results, median shown as a horizontal line inside the box, and 90% of data points were inside the bars stretching from the box. Number of measured individuals for each molar type was 15 for the wild-type group, 8 for the weak expression group, and 5 for the strong expression group.
For histology of the adult incisors, mandibles were sagittally bisected, and the incisal part of half was dissected at the level of the mesial surface of the first molar. This part was further cut into an incisal and a basal block. To preserve any mineralized enamel, the incisal block was processed to ground sections. The specimens were fixed with 10% formalin, dehydrated, and embedded in liquid methylmethacrylate monomer. After the completion of polymerization of the monomer, facilitated by benzoyl peroxide at 37°C, the tooth block was serially sawed into cross-sections, 120–150 μm thick, and mounted unstained. The basal block was demineralized with 0.33 M EDTA, conventionally embedded in paraffin and processed to serial cross-sections, 5 μm thick, and stained with hematoxylin and eosin. Because the mouse tissues were preboiled for making the skeletal preparations, the quality of the paraffin sections was not optimal.
Histology, In Situ Hybridization, and Immunohistochemistry
Embryonic mouse tissue was fixed in 4% paraformaldehyde over 2 nights, dehydrated and embedded in paraffin for sectioning by using standard procedures. In situ hybridization with 35S-UTP (Amersham) labeled riboprobes was performed as described in Wilkinson and Green (1990) with a 1,415-bp fragment of murine Edar (Laurikkala et al., 2001). The slides from the radioactive in situ hybridization were photographed with an Olympus Provis microscope equipped with an Olympus DP70 CCD camera. The bright- and darkfield images were combined in Adobe Photoshop. Anti-myc tag antibody (Upstate Biotechnology) was used at a 1:500 dilution on paraffin-embedded skin sections.
We thank Elaine Fuchs and Uri Gat for the K14-BG plasmid; Heikki Rauvala and Raija Ikonen for generation of the transgenic mice; Heidi Kettunen, Riikka Santalahti, and Sari Suomi for assistance with the genotyping and analysis of mice; Merja Mäkinen, Lydmila Razzkozova, and Marjatta Kivekäs for assistance with histology; Martyn James for help with Western blotting; and Mark Tümmers for help with the figures.