Tissue-specific and Ubiquitous Promoters Direct the Expression of Alternatively Spliced Transcripts from the Calcitonin Receptor Gene*

The gene encoding the murine calcitonin receptor (mCTR) was isolated, and the exon/intron structure was determined. Analysis of transcripts revealed novel cDNA sequences, new alternative exon splicing in the 5 * untranslated region, and three putative promoters (P1, P2, and P3). The longest transcription unit is greater than 67 kilobase pairs, and the location of introns within the coding region of the mCTR gene (exons E3–E14) are iden-tical to those of the porcine and human CTR genes. We have identified novel cDNA sequences that form three new exons as well as others that add 512 base pairs to the 5 * side of the previously published cDNA, thereby extending exon E1 to 682 base pairs. Two of these novel exons are upstream of exon E2 and form a tripartite exon E2 (E2a, E2b, and E2c) in which E2a is utilized by promoter P2 with variable splicing of E2b. The third new exon (E3b * ) lies between E3a and E3b and is utilized by promoter P3. Analysis of mCTR mRNAs has revealed that the three alternative promoters give rise to at least seven mCTR isoforms in the 5 * region of the gene and generate 5 * untranslated regions of very different PCR conditions were 94 °C for 15 s, 60 °C for 15 s, and 72 °C for 30 s for 35 cycles, except for E2a-F2 1 E4-R, for which 50 °C was used instead of 60 °C. Southern blot analysis was performed sequentially or on parallel blots with [ g - 32 P]ATP end-labeled oligonucleotide probes E2a-F2, E2b-R, E2c-F, E3a-R, E3b 9 -F, and E3b-R. Construction of the P2 and P3 mCTR-pGL3 Luciferase Vectors— To test the putative P2 and P3 mCTR promoter activities, the blunt-ended mCTR DNA genomic fragments containing appropriate P2 and P3 regions were subcloned in both orientations into the Sma I site of the pGL3 basic luciferase vector (Promega), which does not contain a promoter or an enhancer. The initial P2 genomic region used spanned from the Bam HI site in E1 ( 2 1253 relative to E2a) to the Eco RI site at the end of E2c ( 1 398) ( 2 1253P2Bam-F and 2 1253P2Bam-R). The initial 859-bp P3 genomic region (starting at 2 797 relative to E3b 9 ) was synthesized by PCR using primers E3a-F and E3b 9 -R ( 2 797P3ab 9 -F and 2 797P3ab 9 -R). The identity and orientation of each construct was verified by sequencing. In addition, three P2 deletion constructs were generated using the Kpn I, Sac I, and Nhe I sites in the 5 9 polylinker region and in the P2 promoter DNA ( 2 806P2Kpn-F, 2 285P2Sac-F, 2 179P2Nhe-F, respectively) to drop successively larger fragments after recircularizing the construct. Further P2 deletions were derived from the 2 179P2Nhe-F plasmid as follows. The 2 30P2Afl-F and the 2 179/ 2 27P2NAf-F plasmids were

The calcitonin receptor (CTR), 1 which contains seven transmembrane domains, is a member of the class II G proteincoupled receptor family (1,2). The class II family, while structurally related, has little similarity at the amino acid level to the class I family (e.g. rhodopsin and ␤-adrenergic receptor). The CTR is coupled to multiple signal transduction pathways. Binding of the 32-amino acid peptide hormone calcitonin can stimulate activation of the following: the adenylate cyclase/ cAMP/protein kinase A pathway (3); the phosphoinositide-dependent phospholipase C pathway (which results in Ca 2ϩ mobilization (4) and protein kinase C activation (5)); and the phosphatidylcholine-dependent phospholipase D pathway (which also results in protein kinase C activation) (6). Calcitonin directly inhibits bone resorption by osteoclasts and enhances renal calcium excretion (7)(8)(9). It also has effects on the central nervous, cardiovascular, gastrointestinal, and reproductive systems, and CTRs have been identified on osteoclasts, certain kidney cells, some regions of the brain, testis, ovary, and spermatoza (for a review, see Ref. 10). Recently, it has been demonstrated that the human CTR (hCTR), when coexpressed with receptor activity-modifying proteins, is also a receptor for the 37-amino acid peptide hormone amylin (11)(12)(13). This peptide has effects on insulin release, glucose uptake, and glycogen synthesis in skeletal musculature (14).
The CTR gene has a complex structural organization with several CTR protein isoforms derived from alternative splicing of transcripts from a single gene (15). These isoforms, which are functionally distinct in terms of ligand binding specificity and/or signal transduction pathway utilization, are distributed both in a tissue-specific and species-specific pattern (16 -27). Furthermore, spliced mRNA products have been identified in hCTR that generate translation terminations shortly after transmembrane domain 1 (28,29), resulting in the expression of truncated CTR proteins. * This work was supported by National Institutes of Health Grants AR45421 (to D. L. G.) and DK46773 (to S. R. G.). 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 nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  Although some of the human (23) and porcine CTR (pCTR) (22) genomic sequences have been cloned, little is known about the mechanism of transcriptional regulation for the CTR gene in osteoclasts or in other tissues in which it is expressed. A 657-bp fragment of the pCTR promoter was demonstrated to drive expression of a luciferase reporter gene when transfected into the CTR-expressing porcine kidney epithelial cell line LLC-PK 1 (22). Recently, the function of a 2.1-kb fragment of the pCTR promoter has been assessed in a transgenic mouse by Jagger et al. (30). They found that although this region directed expression of the lacZ reporter in several embryonic and fetal tissues that express endogenous mCTR, it was not sufficient to direct transcription in the adult kidney or bone of the transgenic mice.
In this report, the gene encoding the murine CTR (mCTR) was isolated, and the exon/intron structure was determined. We have identified novel cDNA sequences that extend the beginning of the previously published mCTR cDNA (21) by 512 bp, thereby enlarging exon E1 to 682 bp. Further analysis of mCTR cDNAs has also revealed three new exons within the 5Ј-UTR, new alternative exon splicing in the 5Ј-UTR, and the presence of three putative promoters (P1, P2, and P3). Two of these novel exons are upstream of the original cDNA exon E2 and form a tripartite exon E2 (E2a, E2b, and E2c) in which E2a is utilized by promoter P2 with variable splicing of E2b. The third new exon (E3bЈ) lies between E3a and E3b and is utilized by promoter P3. Analysis of mCTR mRNAs reveals that the transcripts from the three promoters are spliced to yield seven different 5Ј-UTR structures. Analysis by both RT-PCR and transient transfection of promoter-luciferase reporter constructs shows that the P1 promoter (located upstream of an expanded exon E1) and the P2 promoter (located upstream of exon E2a) are utilized in osteoclasts, brain, and kidney, whereas the P3 promoter (located upstream of the novel exon E3bЈ) appears to be exclusively utilized in osteoclasts. The P2 promoter of mCTR is highly homologous to the promoter region previously defined for pCTR in kidney cells. These studies provide the first evidence that CTR is regulated in a tissuespecific manner by alternative promoter utilization and that there is a unique promoter (P3) that regulates CTR expression only in osteoclasts.

EXPERIMENTAL PROCEDURES
Isolation and Characterization of the Mouse CTR Gene-A murine genomic library made from the 129 mouse strain in Lambda FIX II (provided by Dr. Chuxia Deng, Bethesda, MD) was screened using probes from the murine CTR cDNA clone isolated by Yamin et al. (21). The probes were labeled with 32 P by random priming. The 13 positive phage clones were plaque-purified, and DNA was prepared utilizing standard procedures. NotI fragments containing the mCTR genomic DNA from these phage were subcloned into pBluescript KS (Stratagene) and analyzed by restriction enzyme site mapping and hybridization with region-specific CTR probes. In some cases, PCR between exons was used to determine the intron size. Note that the originally published cDNA (21) included 14 bp of adapter and the mCTR sequence actually starts at position 15 (GenBank TM accession number U18542). 2 Genomic regions of interest, such as exon/intron junctions, were sequenced using the dideoxy chain termination method (31) with Sequenase polymerase version 2 (U.S. Biochemical Corp.) and primers based on previously known or novel mCTR sequence as necessary. The exon/ intron junctions were established by comparison of genomic sequence with the cDNA sequence. Comparison with other CTR genomic sequences was done using both BLAST (32) and the GCG sequence analysis package (33).
RNA Preparation-Total RNA was isolated from either mouse organs (brain, liver, kidney), osteoclast cocultures, or cell lines by using Trizol Reagent (Life Technologies, Inc.) according to the manufacturer's instructions. Osteoclasts were generated by coculture of hematopoietic progenitors with the stromal cell line ST2 essentially as previously described (34). Bone marrow cells were collected by flushing the femurs and tibiae of 5-week-old male C57/Bl6 mice with ␣-minimal essential medium. The cells were cultured for 24 h in ␣-minimal essential medium containing 10% fetal calf serum, and then the nonadherent population was collected and erythrocytes were removed using a density gradient centrifugation on Ficoll-Hypaque (Sigma). The remaining hematopoietic progenitors were cocultured with ST2 cells at a 1:10 ratio in medium containing 1,25-(OH) 2 D 3 (10 Ϫ8 M) and dexamethasone (10 Ϫ7 M). After 9 days of coculture, the ST2 stromal cells were removed by incubation with collagenase, and then the remaining osteoclasts were harvested for RNA as previously described (35). mRNA was prepared using the poly(A) quick mRNA isolation kit (Stratagene). 2 The following sequences deposited into GenBank TM are referred to in this paper: the original mCTR cDNA sequence under GenBank TM accession number U18542 (21); human BAC GS1-438P6 under Gen-Bank TM accession number AC005024; human BAC GS1-117O10 under GenBank TM accession number AC003078; hCTR E2ac cDNA under GenBank TM accession number AB022177 (59); and pCTR genomic under GenBank TM accession number Z31356 (22).

Osteoclast-specific Calcitonin Receptor Gene Regulation
FIG. 1. Genomic structure of the mCTR gene. A, depiction of 13 mCTR genomic phage inserts (numbered thick lines) plus a portion of the gene obtained by PCR between exons E6 and E7 (CD-PCR). Phage inserts 13 and 15 do not overlap, so the intron between E3bЈ and E3b is Ͼ13.9 kb. The designations of the exons (E#) reflect the numbering system used for the pCTR coding exons (22). The filled boxes portray locations of the mCTR exons, and the intron lengths (kb) are denoted below each one. The novel exonic regions described in this report are included. Note that IVS10, between exons E10 and E11, is only 81 bp long, and consequently those two exons appear as one. The two putative translation starts (small arrows marked aa) in the mCTR cDNA are split between exons E3a and E3b. The three RNA starts described in this report are also denoted (large arrows). *, the exact position of the P1 RNA start is not yet known. Various restriction enzyme cut positions are indicated (B, BamHI; Bg, BglI; K, KpnI; P, PstI; X, XhoI). The order of the PstI sites marked with dots is not certain. B, correspondence between mCTR exons and putative protein domains. The exons of the mCTR cDNA published by Yamin et al. (21) plus the additional exon E1 sequences defined in this report are depicted in relation to the mCTR protein domains previously defined (21) presented schematically in the cellular membrane. The horizontal size of each protein domain represents the position and length of the exon(s) encoding it. The shaded protein domain at the N terminus is the additional 17 amino acids encoded when translation starts in E3a. The extracellular and intracellular protein domains are black, and the seven transmembrane domains are white.

TABLE II
Exon/intron junction sequences of the mCTR gene F The numbering scheme for the exons was adopted from that of the pCTR. * IVS2a is equivalent to E2b.

Characterization of the 5Ј-End of Exon E1 by Primer Extension and
Ribonuclease Protection Assay (RPA)-RNA samples (5-20 g) from various tissues or cell lines were reverse transcribed to generate primer extension products using the Ready-To-Go You-Prime First-Strand Beads kit (Amersham Pharmacia Biotech) and an mCTR-specific antisense primer located in exon E1 end-labeled with [␥-32 P]ATP using T4 polynucleotide kinase. The primer extension products were separated in a 7% polyacrylamide sequencing gel together with a sequencing reaction of the mCTR genomic clone 18.26 using the same primer used in the primer extension reaction.
To determine the 5Ј-end of exon E1, an RPA was performed. A 532-bp mCTR genomic DNA fragment spanning from a BsaBI site to a BamHI site encompassing the 5Ј-end region and part of exon E1 was blunted with Klenow and subcloned into the SmaI site of pBKS vector, linearized with BamHI, and used as a template for synthesizing a 635-bp riboprobe with [␣-32 P]CTP utilizing T3 RNA polymerase. Total RNA from mouse brain (200 g) and mRNA from MDCT209 cells (20 g) were used as templates for RPA using 2 ϫ 10 5 cpm of the riboprobe following the manufacturer's protocol (Ambion). In addition, mCTR cRNA sense strand derived from the same RPA construct was used as a positive control that would yield a 376-bp protected fragment. The RPA protected products were separated in a 5% sequencing gel and analyzed using a PhosphorImager (Molecular Dynamics, Inc., Sunnyvale, CA).
5Ј-Rapid Amplification of cDNA Ends (5Ј-RACE)-To identify and verify the sequence of the 5Ј-ends of mCTR transcripts, mouse brain, kidney, and osteoclast RNA samples were used to generate adapterligated double-stranded cDNA using the Marathon cDNA amplification kit (CLONTECH). For each PCR reaction, 2 l of the double-stranded cDNA adapter ligation solution diluted to 1:10 was added into a total volume of 50 l of PCR buffer solution containing 10 mM Tris-HCl, pH 8.3, 50 mM KCl, 2.5 mM MgCl 2, 1 M each of the 5Ј and 3Ј primers, 2 mM dNTPs, and 2.5 units of Taq DNA polymerase (Fisher). Several 5Ј-RACE experiments were performed using different primer sets with two rounds of PCR. The first round PCR was performed using the sense adapter primer AP1 and two different mCTR antisense primers, either mPE1 or E6-R (see Table I). The AP1 ϩ mPE1 PCR products were used for a second nested round of PCR using the sense adapter primer AP2 and the mCTR primer mPE3. The AP1 ϩ E6-R PCR products were used for several nested PCRs using sense primer AP2 and either mCTR antisense primer E4-R, E5-R, or E6-R. The PCR conditions for both PCR rounds were 94°C for 15 s, 60°C for 15 s, and 68°C for 15 s for 30 cycles using the Gene Amp PCR system 9600 (PerkinElmer Life Sciences). Southern blot analysis was performed using [␥-32 P]ATP-endlabeled oligonucleotide probes located internal to the expected PCR products for each experiment, such as mPE4 and E3b-R. The mCTR positive 5Ј-RACE products were subcloned into the TA 2.1 vector (Invitrogen). The resultant cDNAs were characterized by sequencing from both orientations using T7 and M13-reverse primers.
Tissue Distribution of mCTR Isoforms Derived from Three Promoters Using RT-PCR-For reverse transcription reactions, 2 g of total RNA samples from mouse liver, brain, kidney, and osteoclasts derived from bone marrow cocultures were used to generate cDNAs with 0.5 g of oligo(dT) using the reverse transcription Ready-To-Go You-Prime First-Strand Beads kit (Amersham Pharmacia Biotech). Different sets of mCTR-specific primer pairs were used to analyze the variable structure of the mCTR mRNAs, and a pair of primers specific for GAPDH was used to assess the quality of the RNA samples. Additionally, a pair of primers (mTM5-F and mTM7-R) from a mCTR region without variable splicing was used to establish the presence or absence of mCTR mRNA in each sample. For all PCRs except GAPDH and plasmid controls, equal amounts of each cDNA (2 l of a 30-l RT reaction) were added to a total volume of 50 l of PCR solution similar to that previously described in the 5Ј-RACE method. For the GAPDH PCR, only 1 l of the RT reaction was added to the PCR. PCR of relevant mCTR cDNA plasmids (1 ng) generated by the 5Ј-RACE (P1.1, P2.1, P2.3, and P3.1) were used to prepare control PCR products. Also, for all PCR reactions, a negative control containing H 2 O instead of cDNA was run. The PCR conditions for the 110-bp GAPDH PCR product were 95°C for 2 min and then 30 cycles of 95°C for 30 s and 60°C for 30 s, ending with 72°C for 7 min. To analyze the mCTR mRNA splice isoforms generated from the P1, P2, and P3 transcripts, either sense primer E1-F, E2a-F, or E3bЈ-F was used, respectively, together with antisense primer E4-R.

FIG. 2. Sequence of upstream mCTR genomic and exon E1.
A, the uppercase letters denote the genomic sequences defined as cDNA sequences by either 5Ј-RACE or RPA (see Fig. 2B), and the lowercase letters indicate genomic sequence not found in mCTR cDNAs. *, ϩ, and represent the upstream end of 5Ј-RACE products derived from brain, kidney, and osteoclast RNAs, respectively. The boldface G denotes the first nucleotide of the previously published mCTR cDNA (GenBank TM accession number U18542) (21) and the underlined C denotes the extra C found in both our genomic and cDNA sequences. The underlined ag is a putative 3Ј splice acceptor. Several reverse primers used in primer extension and 5Ј-RACE experiments are denoted by boxed sequences. B, identification of the 5Ј-end of exon E1 by RPA. A schematic representation of the mCTR probe used in the RPA is presented above the gel. A 532-bp mCTR genomic DNA fragment spanning from a BsaBI site to a BamHI site encompassing the 5Ј-end region and part of exon E1 was blunted with Klenow and subcloned into a SmaI site of pBKS vector, linearized with BamHI, and used as a template for synthesizing a 635-bp riboprobe with [␣- 32  PCR conditions were 94°C for 15 s, 60°C for 15 s, and 72°C for 30 s for 35 cycles, except for E2a-F2 ϩ E4-R, for which 50°C was used instead of 60°C. Southern blot analysis was performed sequentially or on parallel blots with [␥-32 P]ATP end-labeled oligonucleotide probes E2a-F2, E2b-R, E2c-F, E3a-R, E3bЈ-F, and E3b-R.
Construction of the P2 and P3 mCTR-pGL3 Luciferase Vectors-To test the putative P2 and P3 mCTR promoter activities, the blunt-ended mCTR DNA genomic fragments containing appropriate P2 and P3 regions were subcloned in both orientations into the SmaI site of the pGL3 basic luciferase vector (Promega), which does not contain a promoter or an enhancer. The initial P2 genomic region used spanned from the BamHI site in E1 (Ϫ1253 relative to E2a) to the EcoRI site at the end of E2c (ϩ398) (Ϫ1253P2Bam-F and Ϫ1253P2Bam-R). The initial 859-bp P3 genomic region (starting at Ϫ797 relative to E3bЈ) was synthesized by PCR using primers E3a-F and E3bЈ-R (Ϫ797P3abЈ-F and Ϫ797P3abЈ-R). The identity and orientation of each construct was verified by sequencing. In addition, three P2 deletion constructs were generated using the KpnI, SacI, and NheI sites in the 5Ј polylinker region and in the P2 promoter DNA (Ϫ806P2Kpn-F, Ϫ285P2Sac-F, Ϫ179P2Nhe-F, respectively) to drop successively larger fragments after recircularizing the construct. Further P2 deletions were derived from the Ϫ179P2Nhe-F plasmid as follows. The Ϫ30P2Afl-F and the Ϫ179/ Ϫ27P2NAf-F plasmids were generated by dropping the 148-bp NheI/ AflII and the 453-bp AflII/HindIII fragments, respectively, blunting, and recircularizing, while the Ϫ178/ϩ16P2NNs-F was generated by subcloning the 194-bp NheI/NspBII fragment into SmaI-cut pGL3basic. The Ϫ797P3abЈ-F construct was used to generate P3 deletions to Ϫ319 and Ϫ94 by digesting the plasmid with NheI plus XhoI, which cut in the polylinker, isolating the mCTR fragment, and redigesting it with BsrI or MslI. The BsrI/XhoI and MslI/XhoI mCTR fragments were then blunted and cloned back into the SmaI site of pGL3basic to make Ϫ319P3Bsr-F and Ϫ94P3Msl-F, respectively.

RESULTS
Structure of the mCTR Gene-A murine genomic library was screened using probes derived from a previously isolated mCTR cDNA clone (21). We obtained 13 positive phage clones (Fig. 1A) encompassing most of the transcription unit (Ͼ67 kb) with the exception of part of one very long intron (IVS3bЈ) that is greater than 13.9 kb. In addition, the genomic DNA corresponding to a gap between the sequences contained in phages 3 and 15 was isolated by using PCR between exons E6 and E7 employing mouse strain 129 genomic DNA as template (CD-PCR in Fig.  1A). The identified phage clones include 15 kb of the putative 5Ј-flank and 8 kb of the 3Ј-flanking regions. Characterization of all the exon/intron junctions, determination of intron sizes, and mapping of sites for multiple restriction enzymes is shown in Fig. 1A. Sequence analysis confirms the presence of appropriate donor-acceptor consensus sequences (GT-AG) (39,40) at the ends of each intron (Table II). A schematic of the relationship between the exon borders and the protein regions is depicted in Fig. 1B. Of interest, the location of the exon/intron junctions of the coding exons in the mCTR gene are exactly the same as in the porcine and human CTR genes (22,23). This includes the two putative translation start sites (Fig. 1A, labeled aa) that are split between two exons in the murine (E3a and E3b) and human (71-bp insert and E3) CTR genes. Use of the upstream translational start would add 17 amino acids to the mCTR protein (Fig. 1B, shaded box at the N terminus of the protein). Exon E8b is a rodent-specific 111-bp coding exon that adds 37 amino acids to the extracellular domain 2 of mCTR and is alternatively spliced in mCTR mRNAs (18,21,41,42). A number of sequence conflicts were found between our genomic sequence and our originally published mCTR cDNA (21) in exons E1, E2c, and E8b. The newly isolated cDNAs described below contain the same E1, E2c, and E8b sequences as the genomic sequence.
Identification of the 5Ј-End of Exon E1-In order to determine the transcription start site of the mCTR gene, primer extension and 5Ј-RACE analyses were performed using RNA from mouse tissues known to express CTR and in which CT exhibits biological activities (murine brain, kidney, and osteoclasts). The primer extension analysis was first carried out with a 32 P-end-labeled mCTR-specific antisense primer (mPE1; see Fig. 2A) with its 3Ј-end located 75 nt downstream from the reported 5Ј-end of mCTR (21). The results showed that for all the tissues there were several primer extension products distributed more than 300 bp beyond the original 5Ј-end (data not shown). Since the primer extension results indicated that the 5Ј-end of the mCTR mRNA is located upstream of the 5Ј-end previously reported, 5Ј-RACE was performed to isolate the predicted additional cDNA sequences. The mCTR-specific primer used for the final PCR step was mPE3 (see Table I and Fig. 2A). The subcloned products containing mCTR inserts were identified by hybridization with 32 P end-labeled mPE4 probe. Representative 5Ј-RACE clones with varying insert sizes from each of the RNAs were sequenced and compared with mCTR genomic DNA as shown in Fig. 2A. This reveals that all of the newly identified 5Ј-cDNA sequences are contiguous with the previously defined exon E1 in the mCTR genomic DNA, thereby only extending E1 without forming a new exon. The clones generated from mouse brain RNA contained an additional 56, 66, 240, and 372 nt ( Fig. 2A, *); those from kidney RNA contained an additional 10 nt ( Fig. 2A, ϩ); and those from osteoclast RNA contained an additional 56, 66, and 84 nt (Fig. 2A, ).
To clarify whether any of these new cDNA sequences include the true 5Ј-end of mCTR mRNA, another primer extension analysis was performed using antisense primer mPE6 (see Fig.  2A) with its 3Ј-end located 156 nt downstream of the new putative mCTR 5Ј-end. The results from this primer extension analysis of RNAs from murine brain, kidney, and osteoclasts revealed several products longer than the new putative 5Ј-end identified above by ϳ150, 170, 300, and Ͼ300 nt (data not shown). Even with repeated 5Ј-RACE, we have been unable to recover the additional 5Ј-cDNA sequences revealed by mPE6 primer extension. Therefore, to determine the 5Ј-end of exon E1, an RPA was performed. A subcloned 532-bp mCTR genomic DNA fragment containing 430 bp of putative 5Ј-genomic sequence and 102 bp of exon E1 (as defined by 5Ј-RACE) DNA was used as a template for synthesizing a 635-bp [␣-32 P]CTPlabeled riboprobe, which was used to analyze mouse brain total RNA and MDCT209 mRNA (Fig. 2B). The RPA product observed with both RNA samples was about 245 nt (Fig. 2B, solid  arrow), thereby indicating that the 5Ј-end of exon E1 is located ϳ143 bp further upstream. The additional size of exon E1 revealed by the RPA is too small to explain the primer exten-sion results discussed earlier. These data therefore indicate that there must be another exon upstream of exon E1 and separated from it by an intron of unknown length. The sequence in the region identified as the 5Ј-end of exon E1 (derived from the RPA) is 5Ј-AAGGTG-3Ј. The presence of a possible splice acceptor junction, aag/GTG, suggests that the 5Ј-end of E1 is precisely 140 bp upstream of the 5Ј-end of the most upstream 5Ј-RACE cDNA clone (see Fig. 2A). Between the FIG. 3. Novel mCTR cDNAs indicate the presence of additional promoters P2 and P3 and demonstrate alternate splicing. A, the sequence of the novel exons E2a and E2b identified through 5Ј-RACE and of exon E2c, which contains many conflicts with the same region in the GenBank TM accession number U18542 file. The structures of the cDNAs found starting with E2a are depicted schematically below the sequence. *, the 5Ј-ends of E2a found through 5Ј-RACE. The splice consensus sequences at intron ends found in exon E2b are in boldface type and underlined. There is extensive homology between the mCTR P2-E2abc region and similar regions in hCTR and pCTR (see Fig. 9A), and we have elected to call the mCTR nucleotide aligned with the hCTR and pCTR start ϩ1 (here and in Figs. 4, 8, and 9A). B, the sequence of the novel exon E3bЈ identified through 5Ј-RACE. The structure of the cDNA found starting with E3bЈ is depicted schematically below the sequence.

FIG. 4. Genomic sequence containing promoter P2 and exons E2a, E2b, and E2c.
Shown is a schematic of the mCTR genomic region containing exons E1 through E3bЈ with the sequenced region shown as indicated below. Restriction sites used to generate 5Ј deletions in the P2 mCTR-pGL3 reporters used in Fig.  8 are denoted: A, AflII; K, KpnI; N, NdeI; Ns, NspBII; S, SacI. Note that exons E2a, E2b, and E2c are contiguous in the genomic sequence. Several putative transcription factor binding sites found by computer analysis, using Transfac (75) and other transcription data bases, are underlined.

FIG. 5. Genomic sequence containing promoter P3 and exons E3a and
E3b. Shown is a schematic of the mCTR genomic region containing exons E1 through E3bЈ with the sequenced region shown as indicated below. The primers used to generate by PCR a genomic fragment containing the P3 promoter that was utilized to create the P3abЈ-pGL3 reporters used in Fig. 8 as well as the ATG in E3a are boxed. Restriction sites used to generate 5Ј deletions in the P3 mCTR-pGL3 reporters used in Fig. 8  5Ј-RACE and the RPA results, exon E1 has been extended 512 bp upstream of the previously published cDNA, and the fulllength of exon E1 is 682 bp (including the 1-bp insert within the 3Ј-half of E1 observed in our cDNAs and genomic clones). The exact position of the putative promoter P1 is yet to be determined.
Isolation of Novel mCTR cDNAs Indicating the Presence of Two Novel Primary Transcripts-Analyses of the paralogous genes coding for mouse and rat PTH/PTHrP-R have revealed that there are multiple transcription initiation sites regulated by different tissue-specific promoters (43)(44)(45)(46). To search for the existence of splicing isoforms of the 5Ј-end of the mCTR gene, additional 5Ј-RACE was performed in which the first PCR step employed primers AP1 and E6-R (an antisense primer located in exon E6) (see Table I). The product of this reaction was analyzed by Southern hybridization, cloning, and sequencing. It was also used in a second round of PCRs with three different primer pairs: AP2 ϩ E4-R, AP2 ϩ E5-R, and E1-F ϩ E5-R (see Table I). Sequence analysis of representative cDNA clones generated from osteoclast RNA revealed four different mCTR cDNAs. These are apparently derived from two primary transcripts that initiate from two novel exons designated E2a and E3bЈ and do not include exon E1 (Fig. 3). The primary transcript initiating from exon E2a is spliced to form mRNAs, with three different 5Ј-UTR structures designated as P2.1, P2.2, and P2.3 in Fig. 3A. Exon E2a was observed in the 5Ј-RACE products to contain an additional 10 -12 bp that are 5Ј to the exon E2 sequence (redesignated E2c) found in the previously reported mCTR cDNA. The P2.1 and P2.2 mRNA isoforms differ by the presence or absence of exon E3a, which contains the upstream putative translation start. In the mCTR P2.3 mRNA isoform, an additional 182-bp exon (designated E2b) was identified between exons E2a and E2c. This isoform was found to contain the variably spliced exon E3a. The newly identified exons E2a and E2b do not contain an ATG translational start codon and are probably 5Ј-UTR sequences. Therefore, their inclusion in a mRNA does not change the mCTR protein structure.
The primary transcript initiating from the novel exon E3bЈ is spliced to form a mRNA, with a single 5Ј-UTR structure designated as P3.1 in Fig. 3B. Exon E3bЈ was found by 5Ј-RACE to contain 65 bp that are located upstream of exon E3b sequences (Fig. 3B). Exon E3a, containing the upstream translation initiation codon, is not present in the P3.1 type mRNA. Also, exon E3bЈ does not contain an ATG codon. Therefore, translation from this mRNA probably starts in exon E3b.
Analysis of genomic DNA by PCR between exons along with Southern blotting with region-specific oligonucleotide probes revealed that the novel exons are located within the mCTR genomic region defined with the earlier cDNAs (see Fig. 1). Genomic regions around the new exons were sequenced, including the putative promoter regions P2 (Fig. 4) and P3 (Fig. 5) upstream of exons E2a and E3bЈ, respectively (also see Table  II). Exons E2a and E2b were found to be contiguous in the genomic DNA with exon E2c (Fig. 4). Therefore, exon E2b serves both as an exon in the P2.3 type mRNA and as an intron in the P2.1 and P2.2 type mRNAs. The exon E2b sequence contains the GT-AG splice 5Ј-donor-3Ј-acceptor consensus sequences required for it to be processed as an intron between E2a and E2c, and it therefore may also be considered to be a "retained intron." Intron retention is another form of alternative splicing that has been observed to occur in a number of different genes such as Drosophila male-specific lethal 2 (47), bovine and human growth hormone, murine vitamin D receptor, and rat renal outer medulla K ϩ channel (see Ref. 48 and references therein). Exon E3bЈ is located 690 bp downstream of exon E3a and an unknown distance (Ͼ12.9 kb) upstream of exon E3b (Figs. 1A and 5). The sequences upstream of the putative 5Ј-ends of exons E2a and E3bЈ do not contain a 3Јsplice acceptor consensus sequence, suggesting that these exons are probably used to initiate mCTR transcripts. In contrast, the sequences downstream of exons E2a and E3bЈ contain 5Ј-splice donor consensus sequences (see Table II and Figs. 4 and 5). Primer extension and RPA were attempted to further define the 5Ј-ends of the P2 and P3 transcripts. However, due to the small sizes of exon E2a and E3bЈ, the usage of E2c and E3b in mRNAs initiated upstream of E2a and E3bЈ, and the very low abundance of mCTR RNA, we have not been able to establish the definitive 5Ј-ends. However, in light of the sequence homology within the region around E2a between mCTR, pCTR, and hCTR (see Fig. 9) and for the sake of consistency across the species, we have assigned the ϩ1 of E2a to be 21 bp upstream of the E2a start detected by 5Ј-RACE (Figs. 3 and 4).
mCTR mRNA Isoform Tissue Distribution-To assess the presence of P1, P2, and P3 mCTR transcripts in mouse osteoclasts, kidney, liver, and brain, semiquantitative RT-PCR was performed using primers E1-F, E2a-F, or E3bЈ-F plus E4-R to separately detect all of the mRNA products derived from each of the primary transcripts through exon E4 (Fig. 6). The RT-PCR cDNA products were hybridized sequentially or in parallel with 32 P-labeled oligonucleotides from exons E2a, E2b, E2c, E3a, E3bЈ, and E3b in order to detect and confirm all of the isoforms. Additional PCRs with primers from a region of mCTR included in all mRNAs (mTM5-F ϩ mTM7-R; see Table I) and primers for GAPDH were used, respectively, to assess the presence of mCTR mRNA and the general quality of the mRNA preparations and RT reactions. The results of these analyses are shown in Figs. 6 and 7. The data observed in Fig. 6, A and B, demonstrate that liver does not express CTR. As observed earlier with the P2 mRNAs, exon E3a is alternately spliced in P1 mRNAs (Figs. 6C and 7). In osteoclast, brain, and kidney, the predominant P1 transcript splice isoform appears to be the P1.1, which contains exon E3a (ϩE3a). Although the P1.2 (ϪE3a) isoform was most clearly observed in the osteoclast, it is very weakly present in kidney and brain. Comparison of the relative amount of mCTR PCR products generated from the 3 mCTR-producing tissues observed in Fig. 6, A, C, and D, indicates that the P1 promoter does not appear to be a major contributor to mCTR transcripts in kidney. It should be noted that exons E2a, E2b, and E3bЈ were not detected in cDNA products derived from P1 transcripts. This suggests that these exons are always spliced out of P1 mRNAs, thereby adding weight to the argument that exons E2a and E3bЈ are only used to initiate mCTR transcripts P2 and P3, respectively. The P2.1 cDNA, that includes both exons E2a and E3a, but not E2b (ϪE2b, ϩE3a), seems to be the predominant P2 transcript isoform observed in all three CTR-positive tissues (Figs. 6D  and 7). The P2.2 cDNA (ϪE2b, ϪE3a), although clearly observed in the osteoclast, was very weakly detected in the kidney and not detected in the brain. A fourth novel cDNA isoform, P2.4, derived from the P2 transcript was detected most clearly by the E2b probe and includes E2b but not E3a (ϩE2b, ϪE3a). The two E2b-containing cDNAs, P2.3 (ϩE2b, ϩE3a) and P2.4, were observed in osteoclast and brain but not in kidney. These RT-PCR products were also hybridized with an oligonucleotide probe from exon E3bЈ, and no bands were detected (not shown). This indicates that exon E3bЈ is always spliced out of P2 mRNAs. Usage of exon E3bЈ was only detected in the osteoclast RNA (Figs. 6E and 7), suggesting that the putative P3 promoter may be osteoclast-specific. These data demonstrate that the three primary mCTR transcripts are both expressed in a tissue-specific manner and alternatively spliced within the 5Ј-UTR region to form seven different cDNA structures in a tissue-specific manner (Fig. 7).
Function of the Putative P2 and P3 mCTR Promoters-In order to establish that the putative mCTR P2 and P3 promoters are functional, luciferase reporter constructs containing these regions were transfected into both the mouse kidney cell line MDCT209 and the chicken osteoclast-like cell line HD-11EM (Fig. 8A). The full-length mCTR P2 region containing 1253 bp of sequence 5Ј to exon E2a (including a portion of exon E1) as well as exons E2a, E2b, and E2c (ϩ398) was cloned into the pGL3basic vector (which lacks both heterologous enhancer and promoter) in both the forward (Ϫ1253P2Bam-F) and reverse (Ϫ1253P2Bam-R) orientations. The forward and reverse fulllength mCTR P3 constructs contained 797 bp of sequence upstream of E3bЈ (including exon E3a) as well as most of E3bЈ (Ϫ797P3abЈ-F and Ϫ797P3abЈ-R). Several 5Ј deletions of P2 and P3 were also constructed: Ϫ806P2Kpn-F, Ϫ285P2Sac-F, and Ϫ179P2Nhe-F as well as Ϫ319P3Bsr-F and Ϫ94P3Msl-F. Consistent with the RT-PCR data indicating the lack of P3 transcripts in kidney (Figs. 6 and 7), only the mCTR P2 promoter constructs generated luciferase activity in the MDCT cells (Fig. 8A, solid bars). Also consistent with the RT-PCR data demonstrating the presence of both P2 and P3 transcripts in osteoclasts (Figs. 6 and 7), the forward-oriented P3 promoter construct (Ϫ797P3abЈ-F) was as active as the forward-oriented full-length P2 construct (Ϫ1253P2Bam-F) in the HD-11EM cells (Fig. 8A, hatched bars). None of the reverse constructs were active in either cell type, indicating that although the P2 and P3 genomic regions are likely to contain some enhancer function, they are contributing a key promoter function to the pGL3basic vector. Although relative to the pGL3basic vector, the mCTR P2-promoter constructs were ϳ10-fold more active in the HD-11EM cells than in the MDCT cells, the P2 promoter constructs of different lengths demonstrated the same relative activity to each other in both cell lines. The truncation from Ϫ1253 to Ϫ806 approximately doubled the P2 promoter activity, and the truncation from Ϫ806 to either Ϫ285 or Ϫ179 once again doubled the activity. In contrast, a truncation of the P3 promoter from Ϫ797 to Ϫ319 did not change the P3 promoter activity in either cell type (Fig. 8A), while further deletion to Ϫ94 halved the activity. While the P3 region between Ϫ319 and Ϫ94 is clearly important for full activity, it cannot act as a promoter-independent enhancer when added to the c-Fos promoter (not shown).
We further investigated which regions of the P2 promoter are important for activity by deleting both from the 5Ј and the 3Ј sides of the most active P2 mCTR-pGL3basic construct (Ϫ179/ϩ398P2Nhe-F) to generate Ϫ30/ϩ398P2Afl-F, Ϫ179/ Ϫ27P2NAf-F, and Ϫ178/ϩ16P2NNs-F (Fig. 8B). Transfections into both cell types yielded similar results. Deletion from the 5Ј side to Ϫ30 (Ϫ30/ϩ398P2Afl-F) and deletion from the 3Ј side to Ϫ27 (Ϫ179/Ϫ27P2NAf-F) yielded constructs with very little activity. However, in both cell types, the construct with a 3Ј deletion to ϩ16 (Ϫ178/ϩ16P2NNs-F) retained about half of the activity as compared with the Ϫ179/ϩ398P2Nhe-F construct. Therefore, while there may be some positive regulatory elements between ϩ16 and ϩ398, the region between Ϫ179 and ϩ16 is required for promoter activity. DISCUSSION In the present study, we cloned and characterized the mCTR gene. It contains 19 exons distributed over a region Ͼ67 kb long. This report establishes that the mCTR gene is transcribed from three different promoters in a tissue-specific fashion and that these three primary transcripts are spliced within the 5Ј-UTR to generate seven mRNA isoforms (see Fig. 7). Of note, a RNA transcript of ϳ4.2 kb was identified in RNA from mouse kidney and brain using Northern blot analysis (21), and the original mCTR cDNA (which is a type P1.1 5Ј-UTR and contains E8b) is only 3736 bp. With the additional novel 512 bp of exon E1 added, this isoform of the mCTR mRNA is then expected to be 4247 bp. However, primer extension results indicate that this mRNA isoform is even longer and that there is another exon upstream of exon E1. RT-PCR products synthesized from exon E1 to downstream exons such as E2c, E3, E4, etc. have never been found to include exon E2a, E2b, or E3bЈ (see Fig. 6), thereby adding weight to the argument that exons E2a and E3bЈ are only used to initiate mCTR transcripts P2 and P3, respectively, and are always spliced out of primary transcripts that start upstream. Rodent CTRs contain an additional variably spliced exon (111-bp E8b) within the coding sequence whose presence adds 37 amino acids to the extracellular domain 2 and alters the ligand specificity (18,21,41,42). We have not established the configuration of the seven 5Ј-UTR splice forms with the presence or absence of exon E8b. However, exon E8b is primarily retained in mCTR mRNA expressed in the brain. Therefore, mCTR mRNAs with 5Ј-UTR types P1.1, P2.1, P2.3, and P2.4 are the only isoforms likely to contain exon E8b.
mCTR Genomic Structure-Both the P2 and the P3 promoters lack many well known transcription initiation site consensus sequences. These include a TATA box in the Ϫ30 region, an Inr element (YYANWYY) at ϩ1 (49), a DPE site (RGWCGTG) downstream near ϩ30 (50,51), and a BRE site (SSRCGCC) at approximately Ϫ38 relative to the start of transcription (52). Both promoters possess a possible YY1 binding site (VKHCAT-NWB) at the putative transcription start that could be involved in recruiting TFIIB (53,54). The mCTR P2 promoter between the NheI site (Ϫ179) and the transcription start is relatively GC-rich (70% GC) and contains 18 CpG (which represents 10.1 CpG/100 bp) and 21 GpC to give a ratio of CpG/GpC ϭ 0.86. These ratios are the hallmark of a CpG island (55). Promoters containing CpG islands have been proposed to be associated with replication origins and with transcriptional activity during embryogenesis (56). On the other hand, the P3 promoter region between E3a and E3bЈ is only 36% GC, and the Ϫ319 region is only 41% GC.
There are a number of potential transcription factor binding sites that can be identified within both promoters, including many possible sites for C 2 -H 2 zinc finger type transcription factors in P2 (57,58). Nishikawa et al. (59) recently isolated an hCTR cDNA containing the equivalent of E2a-E2c by 5Ј-RACE using a human mammary tumor cell line, MCF-7. Comparison of the Ϫ179 mCTR P2 region with the pCTR promoter sequence upstream from "exon 1" (22)  ber AC005024) revealed a high degree of homology and conserved sequence motifs for several transcription factors (Fig.  9A). The homology between the three species was very high in pairwise comparisons (ϳ70%) for more than 2 kb further upstream (not shown). Typical for many TATA-less myeloid promoters, the proximal P2 promoter regions of the CTRs of all three species contain two putative Sp1 sites (60), although their positions are not identical. Among the more intriguing putative binding sites in the Ϫ179 P2 promoter (indicated in Figs. 4 and 9A) are consensus sequences for several transcription factors known to be important for myeloid gene expression: Spi-1/PU.1, myeloid zinc finger-1 (MZF-1), and MBF, which overlaps a good consensus half-site for either AP1 or CREB (for a review, see Ref. 61). Fitting with the location of the P2 promoter within a CpG island, Jagger et al. (30) found that a 2.1-kb fragment of the pCTR promoter that resembles mCTR P2 was only able to direct expression of the lacZ reporter in several embryonic and fetal tissues that express mCTR but not in the adult kidney or bone of the transgenic mice. Similarly, among the more intriguing putative binding sites (indicated in Figs. 5 and 9B) in the Ϫ319 P3 promoter are putative sites for NFAT ϩ AP1, E-box binding proteins, STATs (GAS sites), MZF-1, GATA, and Spi-1/PU.1. Comparison of the mCTR P3 region with the human BAC GS1-117O10 (GenBank TM accession number AC003078), which contains part of hCTR, revealed a region of high homology (ϳ73%) between 172939 and 171751 (Fig. 9B, only the Ϫ319 E3bЈ region is shown). Using 5Ј-RACE on RNA from human osteoclasts, Nishikawa et al. (59) identified a 288-bp osteoclast-specific exon (located at positions 168711-168422 in GenBank TM accession number AC003078) spliced to exon E3. This region lies between the P3 homology region and E3. Interestingly, unlike the mCTR E3a (which is upstream of P3), the hCTR variably spliced exon (71-bp insert) containing the upstream ATG is located downstream of both the P3 homology region and the osteoclast-specific exon identified by Nishikawa et al. (59).
5Ј-UTR Lengths and Translation-The utilization of three alternative promoters that give rise to at least seven mCTR isoforms in the 5Ј region of the gene with 5Ј-UTRs of very different lengths (see Fig. 7) raises questions concerning the functional significance of these heterogeneous mCTR transcripts. Determination of the exon/intron structure of the mCTR gene reveals that the two putative in-frame translation start sites are localized to two separate exons (E3a and E3b). Both of these exons were included in the originally published mCTR cDNA (21), but alternative splicing of exon E3a is evident from the data presented in this report. The mCTR mRNA types P1.1, P2.1, and P2.3 contain E3a, whereas it is spliced out of types P1.2, P2.2, P2.4, and P3.1. The translation start site used when both exons E3a and E3b are present has not been established. However, both agree well with the Kozak consensus (RXXAUGR) (62) for translation starts, and therefore, it is reasonable to presume that when exon E3a is present, its AUG is used to initiate translation (63). The translation product initiated in E3a is 17 amino acids longer than that initiated in E3b, and these 17 amino acids are quite hydrophilic. However, the 17 amino acids encoded by exon E3b, when translation starts at that AUG, are very hydrophobic and probably consti- FIG. 9. Comparison between the murine, human, and porcine CTR promoter regions. A, three-way comparison of the P2 promoter region between mCTR, hCTR, and pCTR. The Ϫ179 NheI P2 mCTR promoter region (GenBank TM accession number AF333472) was aligned with hCTR (1578 -1343 in GenBank TM accession number AC005024, which was located using the hCTR E2ac cDNA described by Nishikawa et al. (59) (Gen-Bank TM accession number AB022177) as the query sequence with BLAST) and pCTR 2231-2460 E1 genomic region (GenBank TM accession number Z31356) (22) using MACAW. The ϩ1 denotes the reported 5Ј-ends for the hCTR and pCTR and the putative assignment for mCTR.
Boxed regions indicate regions of identity between all three species. B, homology between the mCTR P3 promoter region and a region of hCTR. The Ϫ319 P3 promoter region of mCTR (GenBank TM accession number AF333473) was aligned with the hCTR homology region (Gen-Bank TM accession number AC003078) found using mCTR P3 as the query sequence in BLAST. Boxed regions indicate regions of identity. In both A and B, the putative transcription factor binding sites in mCTR denoted in Figs. 4 and 5 are marked above the appropriate sequences. Additionally, in B, the NFAT/AP1 sites found in hCTR using the algorithm designed by Kel et al. (76) are marked below the sequence by thick bars (none were found in the P2 region shown in A). Various restriction enzyme cut positions are indicated (A, AflII; Bs, BsrI; M, MslI; N, NheI; Ns, NspBII). tute a "signal peptide." When translation initiates in E3a, it is possible that the N terminus of mCTR is cytoplasmic, with the adjacent sequence of hydrophobic residues forming an additional eighth transmembrane domain. Like mCTR, the hCTR mRNA has also been reported to contain two putative translation start sites (17), and the upstream start is contained within a variably spliced 71-bp exon (24). The translation product initiated from the hCTR upstream start site is 18 amino acids longer than if initiated from the downstream start site. The hCTR 71-bp insert is located 8896 bp upstream of hCTR exon E3 in BAC GS1-117O10 (GenBank TM accession number AC003078) at positions 104875-104804.
The P1 mRNAs have very long 5Ј-UTRs of Ͼ955 and Ͼ898 nt (see Fig. 7) that are slightly GC-rich (53% GC) and contain seven AUGs before the AUG in E3a. Most have a pyrimidine at Ϫ3, which makes them poor translation candidates, although one such upstream open reading frame is 51 amino acids long (AUG at ϩ252 in E1). One open reading frame (which encodes 14 amino acids) has an AUG (at ϩ248 in E1) in a good Kozak context for translation. The occurrence of upstream AUG codons nearly always reduces the efficiency of initiation from downstream AUGs (64). The P2 5Ј-UTRs have only one upstream open reading frame with its AUG in a poor Kozak context, range in size from 249 to 487 nt, and are all GC-rich (55-60% GC). The P3 5Ј-UTR has no upstream AUGs, is only 93 nt long, and is slightly AT-rich (48% GC). One possible purpose of the generation of multiple mCTR 5Ј-UTRs is that mRNAs with different untranslated exons can differ in their stability and translational potential. When the translatability of mRNAs from the same gene with both a long and a short 5Ј-UTR has been compared, the short 5Ј-UTR usually is more efficiently translated (65). Indeed, in some instances, the effect of the 5Ј-UTR on translation can be so profound that a minor transcript from certain genes appears to be the major functional mRNA (66,67). It is therefore possible that translation of mCTR protein from mRNAs with a short 5Ј-UTR is more efficient than from mRNAs with a long 5Ј-UTR. The GC-rich untranslated regions could have mRNA secondary structure that may interfere with the translational process (68 -70). It has been shown that deletion of GC-rich untranslated sequences improves translation efficiency of the guanylate cyclase gene (71) and the fibroblast growth factor-related oncoprotein INT-2 (72). Differences in the length of the 5Ј-UTR might also affect the rate of mRNA degradation. For instance, platelet-derived growth factor B/c-sis mRNA possessing a truncated 5Ј-UTR was more resistant to degradation in response to cycloheximide and anisomysin than the mRNA with a long 5Ј-UTR (73). Additionally, the various 5Ј-UTRs could differentially affect mRNA compartmentalization, targeting mRNA to be either translated immediately or stored for later use (74). Therefore, it is possible that translation of the seven mCTR cDNAs is differentially regulated, and the relative abundance of a particular mRNA isoform may not correlate with its contribution to the translated product.
Osteoclasts express all seven 5Ј-UTR mCTR isoforms. However, the mRNA isoforms containing E3a are more abundant than their counterparts lacking E3a. It has not been possible to quantitate the relative promoter usage. On the other hand, in kidney and brain, promoter P3 is not utilized, and it appears that the kidney does not use exon E2b. The presence of alternative promoter usage and splicing, localized to the 5Ј-end of the mCTR gene, may thus provide a mechanism for regulating the expression of this gene at both the translational and transcriptional level. The existence of an mCTR promoter (P3) that is osteoclast-specific is, perhaps, not surprising due to the fact that CTR is expressed in a restricted spectrum of tissues with developmental regulation. Upon proper stimulation, the osteoclast precursor, which is of monocyte/macrophage lineage, undergoes a program of cellular differentiation in which a distinct profile of genes are induced, including, for example, cathepsin K, ␤ 3 integrin, acid phosphatase, and CTR. These genes encode protein products that confer upon the osteoclast the unique functional activities required for attachment and resorption of the mineralized bone matrix. The CTR gene appears to be induced during the terminal stages of osteoclast differentiation coincident with the acquisition of bone resorbing capacity (35). Characterization of molecular regulation of the CTR gene in osteoclasts could lead to novel approaches in treating osteoporosis, periodontal disease, inflammatory arthritis, and related bone disorders.