The Calcium Sensing Receptor and Its Alternatively Spliced Form in Murine Epidermal Differentiation*

We have recently reported that human keratinocytes express both the full-length calcium sensing receptor (CaR) and an alternatively spliced form lacking exon 5, which were suggested to be involved in calcium induced keratinocyte differentiation. To understand further the role of these CaRs, we analyzed the structure of mouse CaRs, and investigated their role using a mouse model in which only the full-length CaR was disrupted. Our results show that both the full-length and the alternatively spliced variant lacking exon 5 encoding 77 amino acids of the extracellular domain were expressed in mouse epidermis. The deletion of the full-length CaR increased the production of the alternatively spliced form of CaR in mutant mice. The keratinocytes derived from these mutant mice did not respond to extracellular calcium, suggesting that the full-length CaR is required to mediate calcium signaling in the keratinocytes. The loss of the full-length CaR altered the morphologic appearance of the epidermis and resulted in a reduction of the mRNA and protein levels of the keratinocyte differentiation marker, loricrin. These results indicate that CaR is important in epidermal differentiation, and that the alternatively spliced form does not fully compensate for loss of the full-length CaR.

is important in epidermal differentiation in vivo (8,9). We identified the calcium sensing receptor (CaR) in human keratinocytes as a candidate molecule, which mediates calcium signaling during terminal keratinocyte differentiation (10,11). This receptor is a member of the superfamily of G proteincoupled membrane receptors through which signaling occurs via phospholipase C activation, with subsequent inositol triphosphate (IP 3 ) production and intracellular calcium ([Ca 2ϩ ] i ) release (12,13). In addition, we found that keratinocytes express an alternatively spliced form of CaR lacking exon 5. This spliced variant is unable to mediate the acute response to [Ca 2ϩ ] o to produce IP 3 . The transcript of the full-length CaR is predominantly expressed in undifferentiated keratinocytes, and its levels decrease as the cells differentiate, whereas the transcript of the spliced variant is expressed throughout the differentiation process. These molecular changes are consistent with the decrease of [Ca 2ϩ ] i and IP 3 response to [Ca 2ϩ ] o during differentiation. These results suggest a role for full-length CaR in human keratinocyte differentiation that is not filled by the alternatively spliced form (11).
Here, we investigated the expression of CaR in mouse epidermis. The complete nucleotide sequence of mouse CaR cDNA was determined. The function of the CaR was analyzed using a mouse model in which the CaR gene was disrupted. To our initial surprise, mutant mice continued to express CaR. However, we then determined that the mouse epidermis, like the human keratinocyte, produced both the full-length and alternatively spliced form of CaR lacking the human equivalent of exon 5 (originally reported as exon 4 (Ref. 14) but renamed exon 5 by a recent study of the genomic structure of CaR (Ref. 15)). The disruption of the CaR in this mouse model resulted in the sole production of the alternatively spliced variant because of the fortuitous insertion of the neomycin cassette into exon 5. This mouse model provided us the opportunity to examine the differentiation of mouse epidermis in which only the alternatively spliced form was expressed. This report contains the first complete nucleotide sequence of mouse CaR cDNA, and our initial description of the consequences to the keratinocyte of deletion of the full-length but not the alternatively spliced form of the CaR. These results indicate that the full-length CaR is important for normal epidermal differentiation in vivo presumably by mediating the calcium signaling required for keratinocyte terminal differentiation.

EXPERIMENTAL PROCEDURES
cDNA Cloning and Sequencing-A partial mouse CaR cDNA clone (mCaR-R) was isolated using RT-PCR from total RNA from the kidneys of either wild type (ϩ/ϩ) or mutant (Ϫ/Ϫ) mice. A set of primers flanking exon 5 of the mouse CaR was obtained from Dr. C. Ho (Harvard Medical School, Cambridge, MA): sense primer (mCaR1309F, 5Ј-CAA GGT CAT TGT CGT TTT CTC CAG C-3Ј), and antisense primer (mCaR2307R, 5Ј-GCA ATG CAG GAG GTG TGG TTC TCA T-3Ј). The latter includes 2 mismatches to the mouse CaR sequence subsequently * This work was supported by Grants R01-AR38386 and P01-AR39448 from the National Institutes of Health and by a Merit Review award from the Department of Veterans Affairs. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The found, but it was adequate for the initial experiments. The 3Ј clone of mouse CaR was obtained by 3Ј RACE. Briefly, total RNA from mouse kidney was reverse transcribed (Superscript II, Life Technologies, Inc.) using the poly(T) adaptor (5Ј-GAC TCG AGT CGA CAT CGA TTT TTT  TTT TTT TTT TT-3Ј). The 3Ј cDNA was amplified by PCR using the gene-specific sense primer (5Ј-ATT AAT TCT GTC CAC AAT GG-3Ј, CaR1018) and antisense adaptor primer (5Ј-GAC TCG AGT CGA CAT CGA T-3Ј). The 5Ј cDNA clone was isolated using 5Ј RACE (Life Technologies, Inc.) using the manufacturer's protocol. Briefly, the RNA was reverse transcribed using the gene specific antisense primer (5Ј-GGA AGG CAT TGG AGC TAT TGA G-3Ј, mCaR1175R). The reverse transcribed cDNA was conjugated with poly(C) on the 5Ј end using terminal deoxynucleotidyl transferase. The cDNA was amplified by PCR using an anchor primer with poly(G) (5Ј-CUA CUA CUA CUA GGC CAC GCG TCG ACT AGT ACG GGI IGG GII GGG IIG-3Ј) and a nested genespecific antisense primer (5Ј-CCA TTG TGG ACA GAC TTC CT-3Ј, CaR1018R). PCR amplification was performed using a mixed Pfu and Taq DNA polymerase (Stratagene) under optimized conditions. The appropriate PCR fragments were subcloned into a pCR II or pCR2.1 vector (Invitrogen). The isolated cDNA clones were sequenced on both strands using vector-specific and gene-specific primers using a Dye termination cycle sequencing kit (Applied Biosystems) on an Applied Biosystems 373A automated DNA sequencer. The DNA sequence was analyzed using the GCG DNA analysis package at the Computer Graphics Laboratory at the University of California, San Francisco.
CaR Disrupted Mice-The CaR disrupted mice were prepared by insertion of a neomycin-resistant gene into exon 5 (originally reported as exon 4 (Ref. 14), but which corresponds to exon 5 of the human CaR gene (Ref. 15)). The mice were given to us as a gift by Drs. C. Ho and J. Siedman (14). The genotype was determined by PCR analysis using the same primer set described in RT-PCR below.
RT-PCR Analysis of CaR Transcripts-The expression of the two forms of CaR was determined by RT-PCR using a set of primers that spanned the missing region (exon 5) of the spliced variant. Total RNA was isolated from the epidermis or kidney of wild type (ϩ/ϩ), heterozygous (ϩ/Ϫ), and homozygous (Ϫ/Ϫ) mutant mice. By reverse transcription using Superscript II (Life Technologies, Inc.), the region encompassing exon 5 was amplified with a sense primer (5Ј-CAA GGT CAT TGT CGT TTT CTC CAG C-3Ј, mCaR1309F) and an antisense primer (5Ј-GCA ATG CAG GAG GTG TGG TTC TCA T-3Ј, mCaR2307R). Each sample was also amplified using a pair of primers for human G3PDH (CLONTECH) as a control for cDNA loading in the PCR assay. Amplification was carried out using Taq polymerase Plus (Stratagene) for 35 cycles for CaR and 30 cycles for G3PDH. The PCR products were analyzed by 3% agarose gel electrophoresis. Molecular weights of the bands were estimated by X174/HaeIII fragments (Life Technologies, Inc.) as standards.
Immunostaining of CaR in CaR Mutant Mice-The expression of CaR protein in the skin was detected by immunostaining. The skin was dissected from 7-day-old mice. The skin was fixed with 4% paraformaldehyde at room temperature. The sections were incubated with an affinity-purified polyclonal anti-CaR antibody (1:500) raised against an epitope of amino acids 215-237 of bovine CaR, which is 100% conserved in mouse CaR. The sections were then washed and incubated with horseradish peroxidase-coupled goat anti-rabbit secondary antibody (1: 2000). The cells were stained with the Fast DAB substrate system (Sigma). Photomicrographs were taken with a 100ϫ objective.
Protein Analysis of CaR by Western Blot-Crude plasma membranes were isolated from mouse kidney as described (11). Membrane fractions were also prepared from HEK293 cells transfected by human cDNA of full-length or spliced variant CaR (11). Briefly, the tissues or cells were sonicated, and the membrane fractions were extracted with radioimmune precipitation buffer containing 1% deoxycholate, 1% Triton X-100, 0.1% SDS. Protein concentration in these membrane preparations was determined using the Pierce BCA protein assay (Pierce). The CaR proteins were analyzed by Western blot as described (11). Briefly, the membrane protein samples were electrophoresed through polyacrylamide gels and electroblotted onto nylon membranes. After block- ing, the blot was incubated with monoclonal antibody against human CaR at a dilution of 1:32,000 (0.1 g/ml) (ADD antibody; gift from Dr. E. F. Nemeth, NPS Pharmaceuticals) overnight at 4°C. As a control for specificity, the antibody was preabsorbed with the synthetic peptide against which the antibody was raised. Subsequently, the samples were incubated with secondary antibody conjugated with horseradish peroxidase (anti-mouse Ig, sheep antibody, Amersham Pharmacia Biotech) for 1 h at room temperature. The bound antibody was visualized using an enhanced chemiluminescence system (ECL, Amersham Pharmacia Biotech).
Measurement of [Ca 2ϩ ] i -Mouse keratinocytes were obtained from 2-4-day-old newborn mouse epidermis by trypsinization as described (15). Briefly, cells were grown in serum-free keratinocyte growth medium (KGM, Clonetics) with 0.05 mM Ca 2ϩ (5). The [Ca 2ϩ ] i response to elevated [Ca 2ϩ ] o was measured as described (11). Briefly, preconfluent keratinocytes grown on a coverslip were loaded with 7.5 M Fura-2 AM and 0.05 mM calcium in buffer A (20 mM Hepes, 120 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 1 mg/ml pyruvate, 1 mg/ml glucose). Cells were then exposed to 2.0 mM calcium. The fluorescence (ratio of F 340 nm /F 380 nm ) from single cells was recorded by an image analyzer. The signals from more than 20 single cells for each preparation were recorded, and an average profile was calculated. Each sample was calibrated by the addition of ionomycin (10 M final concentration) (F max ) followed by 0.1% Triton X-100 and 20 mM EGTA/Tris, pH 8.3 (F min ). [Ca 2ϩ ] i was calculated from the following formula: Microscopy, In Situ Hybridization, and Immunohistochemistry-Dorsal skin samples of 3-day-old wild type (ϩ/ϩ) and mutant (Ϫ/Ϫ) mice were fixed in 4% formaldehyde. Tissue processing, in situ hybridization, and immunohistochemistry were performed as described previously (16). Microscopic anatomy of the tissues was analyzed on toluidine blue-or hematoxylin-eosin-stained sections. A digoxigenin-labeled antisense loricrin RNA probe was made from linearized cDNA templates (gift from S. H. Yuspa, National Institutes of Health, Bethesda, MD), using enzymes and reagents supplied by Roche Molecular Biochemicals. Loricrin mRNA was detected by in situ hybridization using 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium substrate (Roche Molecular Biochemicals). Sense control RNA was used to establish the specificity of the hybridization. The affinity-purified rabbit antibody against loricrin (BABCO, Berkeley, CA). (17) was used to stain the loricrin protein with biotinylated goat anti-rabbit IgG followed by ABC-peroxidase (both from Vector, Burlingame, CA). Peroxidase activity was revealed by DAB substrate (Vector).

RESULTS
Complete Nucleotide Sequence of Mouse CaR-To initiate the sequencing of mouse CaR, both the full-length and anticipated alternatively spliced form, we first generated partial cDNA clones by RT-PCR using primers spanning exon 5 (14). The RNA from the wild type (ϩ/ϩ) mice produced a 0.8-kilobase pair DNA of mouse CaR (mCaR-R). Mutant mice (Ϫ/Ϫ) generated a smaller DNA fragment (mCaR-S). Sequencing demonstrated that the clone mCaR-S has an in frame deletion of 231 bp of exon 5, which would cause deletion of 77 amino acids, equivalent to the naturally produced CaR spliced variant in human keratinocytes. The complete nucleotide sequence of mouse CaR was then determined using 5Ј and 3Ј RACE. The cDNA of mouse full-length CaR contains 4550 bp including an open reading frame of 3236 bp (525-3761) having a translation initiation codon assigned to the first in frame ATG with Kozak sequence (Fig. 1) (GenBank accession no. AF110178). The spliced variant includes 4319 bp lacking 231 bp of exon 5 (AF110179). The deduced amino acid sequence of full-length CaR consists of 1079 amino acids (M r ϳ120,000), and the spliced variant includes 1002 amino acids (M r ϳ100,000). The full-length mouse CaR consists of three major domains: a large hydrophilic extracellular domain (EC), a transmembrane (TM) domain, and an intracellular domain (Fig. 2). The mouse CaR is highly homologous to the CaR from the rat (98% amino acid identity, respectively) (18), human (95%) (19), and bovine (93%) (20), and it has less homology with puffy fish CaR (21) (83%) which has a much shorter intracellular domain (Fig. 2). The position of exon 5, which is deleted in the spliced variant, is indicated as a region having 77 amino acids with 2 N-linked glycosylation sites and 10 acidic amino acids in the position of 460 -537. The region of exon 5 was highly conserved among FIG. 1-continued species, having only 1 amino acid difference from rat and human CaR among 77 amino acids, suggesting that the mouse alternatively spliced variant of CaR would show the same loss of acute calcium signaling as the human variant (11). The mouse CaR is also homologous to metabotropic glutamate receptors, which belong to G protein-coupled receptor family 3, and showed the typical structural features conserved in this family. The hydrophobic segment in the N-terminal side of the EC domain (aa 148 -176) was identified, which may contribute to the ligand binding pocket (20). The 17 conserved Cys and the 6-Cys cluster in the C-terminal part (560 -585) of the extracellular domain may be involved in disulfide bonds. The 7-TM domain includes a conserved cytoplasmic loop (795-805) having a PKC site (646) between the 5th and 6th TM. The Cterminal cytoplasmic domain (217 amino acids) contains 6 potential phosphorylation sites for PKC at 794, 888, 895, 915, 993, and 1059 in addition to two sites in the TM domain at 646, 699. Potential phosphorylation sites for cAMP-dependent protein kinase were present at aa 899 and 900, suggesting that cAMP also may modulate the function of CaR. The nucleotide sequence of mouse CaR reported here is 100% identical to a partial 500-bp CaR cDNA isolated from the mouse osteoblast (GenBank accession no. AF002015) (22).
CaR mRNA Expression in Mutant Mice-To determine whether the expression of the alternatively spliced form of CaR would be affected by targeted disruption of the full-length CaR, we next compared CaR mRNA expression in wild type (ϩ/ϩ), heterozygous (ϩ/Ϫ), and homozygous (Ϫ/Ϫ) CaR mutant mice. Both full-length and the spliced variant of CaR were detected by RT-PCR using a primer set encompassing exon 5. Both the 1007-bp full-length CaR (upper band) and the 777 bp spliced variant (lower band) were expressed in the epidermis of (ϩ/ϩ) and (ϩ/Ϫ) mice (Fig. 3B, lanes 1 and 2) , lanes 4 -6). These results indicate that these two forms of CaR are expressed in the mouse epidermis even in the wild type animal, and that the alternatively spliced variant is expressed in other tissues such as the kidney when the full-length form is deleted. Whether expression of this mRNA leads to production of protein CaR is the next question we addressed.

Protein CaR Expression in CaR Mutant Mice-The
CaR mutant mice (Ϫ/Ϫ) were originally reported not to produce CaR protein (14). However, the mRNA data suggested that at least in the skin and kidney, the alternatively spliced CaR protein was likely to be produced in mutant (Ϫ/Ϫ) mice. In order to verify this prediction, we immunostained the skin from (ϩ/ϩ) and (Ϫ/Ϫ) mice using a polyclonal antibody against CaR (11). The CaR was detected throughout the upper layers of epidermis of the 7-day-old wild type (ϩ/ϩ) mice (Fig. 4A). The epithelial cells around the hair follicle also stained strongly for CaR (A). Similar staining was observed in the skin of the CaR mutant (Ϫ/Ϫ) mice (B). The staining with pre-immune IgG did not show significant signal (C). When the antibody was preadsorbed with excess peptide against which the antibody was raised, the staining was markedly reduced (D). These results show that CaR is present in the epidermis on both wild type (ϩ/ϩ) and mutant (Ϫ/Ϫ) mice. We next analyzed CaR proteins by Western blot. The membrane fraction from the mouse kidney was prepared. The wild type (ϩ/ϩ) and heterozygous (ϩ/Ϫ) mice expressed two major bands of CaR, estimated as ϳ140 and ϳ160 kDa (Fig. 5, lanes 1 and 2). These two bands correspond in size of the two differently glycosylated forms of human CaR expressed in HEK293 cells transfected with cDNA for the full-length CaR (lane 4) (11). In contrast, the homozygous mutant (Ϫ/Ϫ) mice had only a ϳ130-kDa band (lane 3), which corresponds to the single band of CaR transfected with the human spliced variant cDNA (lane 5) (11). The size of the ϳ130-kDa band is consistent with the spliced variant lacking 77 amino acids (lane 3) having altered glycosylation. The bands at ϳ130, ϳ140, and ϳ160 kDa were removed by pre-adsorption of antibody (data not shown). The lower band around 100 kDa was nonspecific (lanes 1-3). The CaR protein was less readily detected in epidermal samples using Western analysis unlike immunolocalization. These results indicate that the full-length CaR is produced in the wild type (ϩ/ϩ) mice, although the spliced variant of CaR protein is still expressed in knockout mice (Ϫ/Ϫ).   1-3). The same RNA was amplified by a primer set for G3PDH as a control (A and B, lanes 4 -6). Two independent PCR analysis showed the same results.   (Fig. 6). In contrast, keratinocytes from mutant mice (Ϫ/Ϫ) showed higher basal [Ca 2ϩ ] i (283 nM), and increased their [Ca 2ϩ ] i only slightly to 333 nM following the calcium switch (Fig. 6). These results indicate that the full-length CaR is required for normal acute [Ca 2ϩ ] i response to [Ca 2ϩ ] o , and this function can not be replaced with the alternatively spliced CaR.
Morphological Analysis of the Skin from Knockout Mice-To determine whether the loss of the full-length CaR is accompanied by altered epidermal differentiation, we analyzed the skin of the CaR mutant (Ϫ/Ϫ) mice. By 3 days after birth, the skin from mutant mice was shinier, less pigmented, and appeared thinner. This difference was less apparent after 7 days. The histopathology of 3-day-old mutant mice (Ϫ/Ϫ) showed a modest reduction in the number of nucleated epidermal cell layers, with a disordered differentiation sequence, demonstrated by abnormal polarity and flattening from the basal through suprabasal cells in comparison to wild type epidermis (Fig. 7). In addition, both the stratum granulosum and the density of ker-atohyalin granules appeared to be reduced in the epidermis of mutant mice (Fig. 7). The same results were obtained in comparison with wild type littermates and with wild type animals from different litters. These results indicate that the CaR plays an important role in vivo during epidermal morphogenesis.
Loricrin Expression in the Epidermis of CaR Mutant Mice-To examine the effect of deletion of the full-length CaR on differentiation, we next compared the expression of the terminal differentiation marker, loricrin, in 3-day-old mutant (Ϫ/Ϫ) mice (Fig. 8, B, D, and F) with wild type (ϩ/ϩ) mice (A, C, and E). The histologic appearance (A and B), the loricrin mRNA (C and D), and protein levels of loricrin (E and F) are shown in the same skin samples. Whereas loricrin mRNA and protein levels and localization in wild type epidermis were as expected (C and E), the epidermis of the mutant (Ϫ/Ϫ) mice was altered (D and F). In wild type epidermis, loricrin mRNA was found in the stratum granulosum (C), where it was very abundant. In contrast, an extremely low (almost undetectable) signal for loricrin mRNA was present across the suprabasal layers of the epidermis in mutant mice (D), and most cells in the stratum granulosum did not show any accumulation of loricrin message. Even those occasional cells of the stratum granulosum which had a stronger signal, stained much less intensely than granular cells in the wild type epidermis. This expression pattern of the loricrin gene was paralleled in the localization of its protein product. In wild type epidermis, loricrin was localized in a sharply demarcated manner to the stratum granulosum (E). In mutant epidermis, however, a weak, diffuse immunostaining was seen (F) across the suprabasal epidermis, with lack of increased staining in the stratum granulosum. Therefore, in the absence of the full-length CaR, loricrin expression and localization are markedly reduced. These results strongly suggest that the full-length CaR is required for epidermal differentiation.

DISCUSSION
The identification of the CaR provided an important step toward our understanding of how calcium regulates cell function. The ability of calcium to regulate keratinocyte growth and differentiation is well established (1,2). The discovery that the CaR, which was originally identified in the parathyroid gland and kidney, was also found in keratinocytes (10) suggested a common mechanism for calcium signaling in these different cell types. Deletions of the full-length CaR by homologous recombination in which the mutant gene contained a neomycin cassette in the mouse equivalent of exon 5 resulted in a phenotype comparable to neonatal hyperparathyroidism (14). However, when we examine the epidermis of these mice using an antibody to the CaR, we found abundant protein CaR. Since we had previously observed that human keratinocytes produce an alternatively spliced form of CaR in which same exon 5 was deleted, we asked whether the mouse epidermis did likewise, and if so whether this would explain the finding that the CaR (Ϫ/Ϫ) mice continued to produce immunologically detectable CaR.
To initiate this investigation, we first determined the complete nucleotide sequence of mouse CaR. The mouse CaR is highly homologous to CaR from other species including the rat (18), human (19), bovine (20), and puffy fish (21). The structural conservation of EC and TM domains were high (Ͼ90% amino acid identity) including fish CaR, suggesting the evolutionary conservation of these regions. The intracellular domain has more variation in amino acids and in length, suggesting that the signaling functions of this receptor may vary from fish to mammals.
As in human keratinocytes, the mouse epidermis makes both the full-length and the spliced variant lacking exon 5. As shown first in human keratinocytes, and now in mouse keratinocytes from CaR mutant mice, the region of exon 5 is critical to the function of CaR at least with respect to the initial response to calcium shown by IP 3 production and [Ca 2ϩ ] i response. The mechanism of functional abolishment by exon 5 is not well understood, although several possibilities were considered. A change in binding affinity to calcium may be caused by the deletion of 10 acidic amino acids in this region of CaR. In addition, the glycosylation of the two N-linked glycosylation sites may be critical for correct folding or transport of CaR protein to the membrane; the spliced variant has an altered glycosylation pattern. The full-length CaR is expressed in normal cells as two bands: the ϳ160and ϳ140-kDa bands. From the analysis of human CaR, we predict that the ϳ160-kDa band is the mature functional form of CaR, and that the ϳ140-kDa band may be an intermediate product of CaR because it has endoglycosidase H-sensitive high mannose type oligosaccha-ride (11). In contrast, the spliced variant is expressed as a single band of ϳ130 kDa, having only the high mannose type glycosylation.
The availability of a mouse model in which the full-length but not the spliced variant CaR was "knocked out" gave us the opportunity to determine the degree to which the alternatively spliced form could compensate for loss of the full-length form in terms of epidermal differentiation. Previous studies with this model (11) clearly showed loss of normal calcium homeostasis in that these animals were severely hypercalcemic and died within a few weeks of birth. The keratinocytes from the CaR mutant mice did not respond to elevated [Ca 2ϩ ] o with an increase in [Ca 2ϩ ] i , clearly showing that the full-length CaR is required to mediate calcium signaling. However, we also observed that the CaR mutant keratinocytes maintained a higher basal [Ca 2ϩ ] i level compared with normal cells, suggesting additional alterations in calcium regulation by CaR mutant keratinocytes. Higher resting [Ca 2ϩ ] i levels were also observed in calreticulin-deficient mice, in which the [Ca 2ϩ ] i response to ATP or bradykinin was abolished (23). Therefore, a rise in [Ca 2ϩ ] i under circumstances in which normal calcium signaling is disrupted may result in increased [Ca 2ϩ ] i by way of compensation. The in vivo significance of this altered calcium signaling may be reflected in the altered differentiation pattern of the epidermis, as demonstrated by both the changes in morphology and expression of loricrin. The epidermis of the CaR mutant mice showed disordered differentiation with reduction of the number of epidermal layers, especially within the first few days of the birth. These differences were not observed after day 7. However, the differentiation marker loricrin still was decreased in mutant mice at this stage.
These alterations in calcium response, skin morphology, and loricrin expression are not likely to result secondarily from the hypercalcemia of these animals. The calcium response was measured in cultured keratinocytes grown in low (0.03 mM) calcium conditions, which should obviate differences in calcium concentrations in vivo. The changes in morphology were only observed shortly after birth, and they were normalized with time, whereas the rise in serum calcium was progressive. Loricrin expression ought to be increased by hypercalcemia, not decreased. Therefore, the observed changes are more consistent with an intrinsic defect in calcium signaling by the CaR mutant (Ϫ/Ϫ) keratinocytes than a secondary response to hypercalcemia.
In summary, we have reported the complete structure of mouse CaR and its alternatively spliced variant. Both are produced by the epidermis, but when the full-length CaR is deleted, the spliced variant is produced in other tissues that normally express only the full-length CaR. Keratinocytes lacking the full-length CaR do not respond to calcium normally, indicating that the full-length CaR is required for normal calcium signaling. The skin lacking the full-length CaR showed thinner epidermis with disordered differentiation, and decreased loricrin expression. These results indicate that CaR is required for normal epidermal differentiation in vivo, presumably by mediating the calcium signaling required for normal keratinocyte differentiation.