Zebrafish Express the Common Parathyroid Hormone/Parathyroid Hormone-related Peptide Receptor (PTH1R) and a Novel Receptor (PTH3R) That Is Preferentially Activated by Mammalian and Fugufish Parathyroid Hormone-related Peptide*

To further explore the evolution of receptors for parathyroid hormone (PTH) and PTH-related peptide (PTHrP), we searched for zebrafish (z) homologs of the PTH/PTHrP receptor (PTH1R). In mammalian genes encoding this receptor, exons M6/7 and M7 are highly conserved and separated by 81–84 intronic nucleotides. Genomic polymerase chain reaction using degenerate primers based on these exons led to two distinct DNA fragments comprising portions of genes encoding the zPTH1R and the novel zPTH3R. Sequence comparison of both full-length teleost receptors revealed 69% similarity (61% identity), but less homology with zPTH2R. When compared with hPTH1R, zPTH1R showed 76% and zPTH3R 67% amino acid sequence similarity; similarity with hPTH2R was only 59% for both teleost receptors. When expressed in COS-7 cells, zPTH1R bound [Tyr34]hPTH-(1–34)-amide (hPTH), [Tyr36]hPTHrP-(1–36)-amide (hPTHrP), and [Ala29,Glu30,Ala34,Glu35, Tyr36]fugufish PTHrP-(1–36)-amide (fuguPTHrP) with a high apparent affinity (IC50: 1.2–3.5 nm), and was efficiently activated by all three peptides (EC50: 1.1–1.7 nm). In contrast, zPTH3R showed higher affinity for fuguPTHrP and hPTHrP (IC50: 2.1–11.1 nm) than for hPTH (IC50: 118.2–127.0 nm); cAMP accumulation was more efficiently stimulated by fugufish and human PTHrP (EC50: 0.47 ± 0.27 and 0.45 ± 0.16, respectively) than by hPTH (EC50: 9.95 ± 1.5 nm). Agonist-stimulated total inositol phosphate accumulation was observed with zPTH1R, but not zPTH3R.

PTHrP in mammals and frogs (1,2). Despite their limited structural homology, which is restricted to the amino-terminal region, both peptides bind with similar affinity to the PTH1R and activate this common receptor with indistinguishable efficacy. Due to this unusual ligand specificity, the PTH1R mediates the PTH-dependent endocrine regulation of mineral ion homeostasis (1,2) and the PTHrP-dependent autocrine/paracrine regulation of different developmental processes, including mammary gland development and endochondral bone formation (3)(4)(5)(6).
There is considerable pharmacological evidence for other receptors with specificity for amino-terminal PTH and PTHrP (1). For example, Ca 2ϩ /inositol 1,4,5-trisphosphate-coupled PTH/PTHrP receptors have been described for keratinocyte cultures, for squamous cell carcinoma and insulinoma cell lines, and for cells from the central nervous system (7)(8)(9). Other, more recent data indicate the presence of a PTHrPselective, cAMP-coupled receptor in the mammalian supraoptic nucleus, which may be involved in the regulation of vasopressin release (10,11). Although these findings suggested the presence of a distinct subfamily of PTH and/or PTHrP receptors, only one novel cDNA encoding the PTH2 receptor (PTH2R) has been isolated thus far (12)(13)(14).
To explore the molecular evolution of PTH receptors, we previously isolated DNAs encoding frog PTH1Rs (15) and teleost PTH2Rs (14). To extend these studies, we now isolated DNAs that encode the zebrafish homolog of the mammalian PTH1R and that of a novel, type 3 receptor (PTH3R). As with the hPTH1R, human PTH (hPTH), fugufish PTHrP (fuguPTHrP), and human PTHrP (hPTHrP) stimulated cAMP accumulation in cells expressing the zebrafish PTH1R (zPTH1R). In contrast, the zebrafish PTH3R (zPTH3R) was preferentially activated by fuguPTHrP and hPTHrP, and less efficiently by hPTH. Furthermore, only the zPTH1R, but not the zPTH3R, showed activation of the inositol 1,4,5-trisphosphate pathway.

EXPERIMENTAL PROCEDURES
Materials-Oligonucleotides and peptides were synthesized as described (16) by the Massachusetts General Hospital Polymer Core Facility. Zebrafish (Danio rerio) for DNA and RNA extraction were purchased from a local aquarium. PCRs were performed on a MJ Research Thermal Cycler (Watertown, MA) using 2.5-5 units of Taq polymerase (Life Technologies, Inc.) or KlenTaq (CLONTECH). Reverse transcription (RT) was performed at 42°C with Superscript II RNase H Ϫ (Life Technologies, Inc.). Restriction enzymes were from New England Biolabs (Beverly, MA); T4 DNA ligase, pGEM-T, and pGEM-3Z from (Promega, Madison, WI); DH5␣ E. coli cells from Life Technologies, Inc.; and Top 10 FЈ E. coli cells and pcDNAI/Amp from Invitrogen. Plasmid DNA from single colonies was purified and sequenced as described (14), and all nucleotide sequences were analyzed using the GCG package (University of Wisconsin, Madison, WI).
Isolation of Full-length cDNAs Encoding the zPTH1R-Using a 5Ј-RACE kit (Life Technologies, Inc.) and total zebrafish RNA, the 5Ј end of the cDNA encoding the zPTH1R was amplified as described (14). zPTH1R Rev 6(2) was used for RT, and the first PCR was performed with the AUAP (Abridged Universal Amplification Primer; Life Technologies, Inc.) and Rev 6(3) (5Ј-GAAGACTATGTAGTGAACACCGAA-3Ј) using the same PCR profile as above, except that annealing was at 55°C (1 min) and that polymerization was at 72°C (2 min) for the first seven cycles, followed by 28 cycles with annealing at 64°C. 0.05 l of the resulting PCR product was reamplified, Rev TM5 (5Ј-ATATTGTT-GTCTGGTGTCACATCT-3Ј) for the first nested PCR, and Rev 4x (5Ј-CGCATTTGTTTCTCGAAGTTTTGTTGC-3Ј) for the second nested PCR. The PCR profiles were identical, except for annealing at 62°C for 1 min and polymerization at 72°C for 2 min (35 cycles; final extension 10 min at 72°C). The products from the second nested PCR were purified and ligated into pGEM-Teasy (Promega) to yield zPTH1R(5Ј)/pGEMTeasy.
RT for 3Ј-RACE was performed according to the manufacturer's instructions using an oligo(dT)-containing adapter primer. PCR was performed with For TM3 (5Ј-ATCTTCATGACCTTCTTCTCAGAC-3Ј) and UAP (Universal Adaptor Prime, Life Technologies, Inc.), using the same profile as before, except annealing at 64°C for 1 min and polymerization at 72°C for 3 min (35 cycles; final extension of 10 min at 72°C). Nested PCR using For 4(1) (5Ј-AGGAAGTACCTCTGGGGCT-TCA-3Ј) and cloning of the resulting products was performed as described above to yield zPTH1R(3Ј)/pGEMTeasy.
Transfection of COS-7 Cells, Radioreceptor Assays, and Agonist-stimulated cAMP and IP Accumulation-COS-7 African green monkey kidney cells (approximately 200,000 cells/well) were cultured in 24-well plates and transfected with plasmid DNA encoding zPTH1R, zPTH3R, or hPTH1R (200 ng/well) as described (14). After transfection, cells were cultured for 72 h at 37°C with daily exchanges of medium, followed by an additional 24 h at 33°C (16) until functional evaluation after 96 h.
Radioreceptor assays with cells expressing either receptor were performed as described (15)  Open boxes depict exons, and filled boxes depict transmembrane helices; approximate locations of recognition sites for restriction enzymes are shown by double vertical lines. Arrows indicate the approximate location of primers used for RT-PCR, 5Ј-RACE, and 3Ј-RACE. Primers in the tail region that are receptor-specific were used to PCR-generate 32 P-labeled probes for the Southern analysis. Double-line arrows (f), degenerate primers; single line arrows with straight heads, zPTH1R-specific; single line arrows with curved heads, zPTH3Rspecific; , conserved sites for potential N-linked glycosylation.
Intracellular accumulation of cAMP in the absence or presence of different agonists was determined as described using transfected COS-7 cells (15,16). To assess total IP accumulation, COS-7 cells (approximately 800,000 cells/six-well plate) were transfected with plasmid DNA (1 g/well) encoding either zPTH1R, zPTH3R, or hPTH1R, respectively (18,19), and cultured for 3 days in DMEM, 7% fetal bovine serum at 37°C with daily exchanges of medium. The cells were then preloaded with 3 Ci/ml myo-[ 3 H]inositol (NEN Life Science Products) in inositol-free DMEM (Life Technologies, Inc.), 7% fetal bovine serum (33°C for 18 h). The following day, plates were rinsed and then incubated with 10 Ϫ6 M hPTH or hPTHrP in DMEM, 0.1% bovine serum albumin, or with DMEM, 0.1% bovine serum albumin alone (40 min at 37°C), in the presence of 30 mM LiCl. Total IP was isolated by anion exchange column chromatography as described previously (18,19), and 1 ml of the eluate (1/8th of total) was counted in a liquid scintillation counter (model LS 6000IC; Beckman, Fullerton, CA). All points represent mean Ϯ S.E. of two to three replicates from two or more independent experiments.
Southern Blot Analysis of Zebrafish Genomic DNA-Approximately 16 g of zebrafish genomic DNA was digested to completion with either BamHI, EcoRI, or HindIII, split into two equal aliquots for electrophoresis through a 0.8% agarose gel containing ethidium bromide, and transferred onto a nitrocellulose membrane (MSI, Westborough, MA). After baking in vacuo for 2 h at 80°C, the blots were hybridized in 50% formamide (42°C for 18 h) with a PCR-generated 32 P-labeled probes (14) encoding either the carboxyl-terminal tail of zPTH1R or zPTH3R (zPTH1R/tail and zPTH3R/tail, 240 and 335 bp, respectively). Washes, for 30 min each, were performed at room temperature and 42°C in 1ϫ SSC, 0.1% SDS, and at 50°C in 0.5ϫ SSC, 0.1% SDS followed by autoradiography at Ϫ70°C for 7 days with intensifying screen (DuPont Cronex) and Kodak XAR film.
Phylogenetic and Structural Analyses of All Known PTHRs-Alignment of all known PTH1Rs, PTH2Rs, zebrafish PTH3R, goldfish VIP receptor, and the human CRF receptor sequences was performed as described previously (14). Sequences were subsequently aligned within MacClade 3.0 (20), and gaps were entered to maximize the homology of the native proteins. Each amino acid was treated as an unweighted character when analyzed using the branch-and-bound search option of PAUP 3.1 (21). A bootstrapping analysis using the heuristic option on 250 replicates (22) was performed, and only groups that were compatible with the 50% majority-rule consensus were retained (23). Further analysis within MacClade was performed to determine residues that are likely to be specific for each receptor subtype (14,20).

Isolation of Genomic DNAs Encoding Zebrafish PTH/PTHrP
Receptors-Using zebrafish genomic DNA and several degenerate forward and reverse primers (Fig. 1A), we obtained, under relatively stringent conditions, two distinct products of approximately 200 and 840 bp. The 5Ј and 3Ј ends of these products showed significant nucleotide sequence identity with each other (73%), with the hPTH1R (78% and 73%, respectively), and with the hPTH2R (64% and 70%, respectively), and are referred to as zPTH1R(TM6/7) and zPTH3R(TM6/7). While zPTH1R(TM6/7) contained 84 bp of intronic sequence (compared with 81 bp of the hPTH1R) (24), zPTH3R(TM6/7) contained an intron of approximately 750 bp. Taken together, the findings indicated that portions of two distinct genes had been isolated which both share higher homology with the PTH1R than with the PTH2R.
Isolation of the cDNA Encoding the zPTH1R-RT-PCR on total zebrafish RNA using the For 3 and primers specific for zPTH1R(TM6/7) (Fig. 1B) produced a 450-bp cDNA fragment, zPTH1R(TM3/6), which encodes a region corresponding to TM3 through TM6 of the mammalian PTH1R. Subsequently, 5Ј-and 3Ј-RACE reactions were performed to generate overlapping sequences, and a full-length zPTH1R clone was constructed using a unique MfeI restriction site. The amino acid sequence (536 residues) encoded by the full-length zPTH1R cDNA (Fig.  2) showed highest sequence homology to the frog, fugufish, and mammalian PTH1Rs (15,17). 3 The overall amino acid sequence homology with the hPTH1R was 76%, but only 68% when compared with the hPTH2R, similar to the mammalian PTH1Rs (17,24). The 3Ј-non-coding regions of zPTH1R and the zPTH3R did not contain typical polyadenylation signals, but several imperfect sequences were found upstream of the poly(A) n tails. 5Ј-RACE generated two PCR products (5Ј-RACE#29, 149 bp; 5Ј-RACE#25, 391 bp), which contained an identical Kozak sequence and coding region (including the putative signal sequence) but varied in the length of the 3 2. Alignment of the amino acid sequences of the zPTH1R, zPTH2R, and zPTH3R. Conserved consensus sites for potential N-linked glycosylation are identified by #; 17 residues that are lacking in one splice variant of the zPTH2R are boxed; the residues that are predicted to comprise the signal peptide are outlined with a dotted box; residues that appear to be PTH3R-specific are boxed. The predicted locations of the transmembrane helices are underlined.

5Ј-untranslated region.
Isolation of the cDNA Encoding a Novel PTHrP-selective Receptor (PTH3R)-Genomic clone zPTH3R(TM6/7) contained, at the 5Ј and 3Ј end, nucleotide sequences that were similar to but distinct from zPTH1R and zPTH2R (14), and contained approximately 750 bp, rather than 84 bp, of intronic sequence. This information indicated that portions of a novel gene, subsequently referred to as zPTH3R, had been isolated. RT-PCR using the For 3 and primers specific for zPTH3R(TM6/7) (Fig.  1B) produced a 450-bp cDNA clone, zPTH3R(TM3/6), which encodes TM3 through TM6 of zPTH3R.
To isolate a full-length clone, we screened an adult zebrafish phage cDNA library (1.5 ϫ 10 6 plaque-forming units), and isolated a single clone that was subcloned to yield zeb3-3Ј/pcDNAI/Amp, and was shown to extend from the region corresponding to the mammalian exon E1 to the carboxyl-terminal region encoded by exon T (17,24), but lacked the portions encoding the amino-terminal, extracellular domain, and the termination codon including the 3Ј-untranslated region. 3Ј-RACE on total zebrafish RNA produced a single product that was subcloned into the xhoI/sphI site of zeb3-3Ј/pcDNAI/Amp. 5Ј-RACE using total zebrafish RNA revealed the presence of several putative splice variants, which is similar to the findings with the zPTH1R (see above) and the zPTH2R (14), and with several mammalian PTH1Rs (26)(27)(28). Seven of 10 isolated zeb3-5Ј clones contained cDNA sequences that were similar to the mammalian exons E1 and E3 (17,24). These clones also contained a Kozak sequence and a nucleotide sequence with homology to the signal peptide sequence found in the mammalian PTH1Rs (17,24), and were therefore ligated into the ApaLI site of zeb3-3Ј/pcDNAI/ Amp to yield the full-length zPTH3R (see Fig. 1).  The remaining three zeb3-5Ј clones also contained the E1 and E3 equivalent, but differed at the 5Ј end and could therefore represent putative splice variants. Similar to the zPTH2R (14) and the human PTH1R (26,28), one of these three putative splice variants of the zPTH3R lacked a signal peptide sequence but did contain a potential initiator AUG, which is located two codons upstream of the exon E1 equivalent. The second putative splice variant lacked a Kozak sequence, an initiator AUG, and contained a charged sequence upstream of the equivalent of E1, which is unlikely to represent a signal peptide. Because of these structural features, both latter clones were not characterized functionally.
Overall, the amino acid sequence comprising the full-length zPTH3R (542 residues, see Fig. 2) shared 69% similarity and 61% identity with zPTH1R, but only 57% similarity and 48% identity with the zPTH2R (14). Similar to the frog PTH1Rs (15) and all known PTH2Rs (12)(13)(14), zPTH1R and zPTH3R lacked the equivalent of the cDNA encoded by exon E2, suggesting that the appearance of this apparently non-essential exon (29) represents a mammalian evolutionary innovation. In contrast to this mammalian apomorphy regarding exon E2, zPTH1R and zPTH3R contained numerous "signature residues" that are characteristic of this family of G protein-coupled receptors; only two consensus sequences for potential N-linked glycosylation were found to be conserved for all zebrafish PTH receptors (see Fig. 2). Amino acid sequence comparisons of the intracellular carboxyl-terminal tails showed that zPTH1R and hPTH1R share 56% similarity, which is similar to the 58% similarity between zPTH2R and hPTH2R (Table I). In contrast, the amino acid sequence homology between the tail regions of zPTH3R and hPTH1R is only 38%, and only 32% when compared with hPTH2R. The zPTH3R was therefore identified as a novel member within the PTH/PTHrP receptor family.
Functional Characterization of zPTH1R and zPTH3R-COS-7 cells expressing either zPTH1R#25 or zPTH1R#29 showed almost indistinguishable results with regard to binding of either radioligand, and with regard to agonist-stimulated cAMP and IP accumulation; therefore, only results obtained with zPTH1R#29 are presented in detail, and referred to as zPTH1R.
COS-7 cells expressing zPTH1R, zPTH3R, or hPTH1R showed similar basal cAMP accumulation (4.3 Ϯ 0.8, 7.2 Ϯ 1.1, and 8.4 Ϯ 1.9 pmol/well, respectively). Significant differences were observed with regard to maximal agonist-stimulated cAMP accumulation (91.5 Ϯ 8.0, 229.6 Ϯ 26.7, and 357.7 Ϯ 30.5 pmol/well), but all three peptides, hPTH, hPTHrP, and fuguPTHrP, showed indistinguishable efficacies with either receptor. The zPTH3R was most efficiently activated by hPTHrP and fuguPTHrP, while hPTH was approximately 22-fold less potent (Fig. 3A, Table II). Interestingly, the zPTH3R showed significant activation already at 10 Ϫ11 M human or fugufish PTHrP. In contrast, the zPTH1R was activated with similar efficacy by hPTH, hPTHrP, and fuguPTHrP (Fig. 3B, Table II), and fuguPTHrP was less potent than hPTH or hPTHrP when tested with the hPTH1R (Fig. 3C, Table II). These findings indicated that the zPTH3R interacts preferentially with PTHrP, while the zPTH1R showed characteristics that are similar to those of the human receptor homolog. FuguPTHrP, which shares more amino acid sequence homology with hPTHrP than with hPTH, was surprisingly active when tested with either zebrafish or human receptors, indicating that its functional characteristics have been preserved throughout evolution.
Consistent with the above findings, i.e. higher efficiency of cAMP accumulation, cells expressing the zPTH3R showed higher specific binding of radiolabeled hPTHrP than of rPTH (20.6 Ϯ 0.6% and 13.3 Ϯ 1.38%, respectively). Furthermore, regardless of the tracer that was used for radioreceptor experiments, apparent binding affinities were higher for fuguPTHrP and hPTHrP, than for hPTH (Fig. 4, A and C, Table II). In contrast, cells expressing the zPTH1R showed, irrespective of the radioligand, similar apparent binding affinities for fuguPTHrP, hPTHrP, and hPTH (Fig. 4, B and D, Table II). Consistent with the less efficient activation of the hPTH1R by fuguPTHrP, the human receptor homolog showed an approximately 2-fold lower apparent binding affinity for the teleost analog, than for mammalian PTH and PTHrP (data not shown). These findings confirmed that the novel zPTH3R interacts preferentially with PTHrP, while the zPTH1R and the hPTH1R interact almost equivalently with all three tested peptides.
Comparable to the mammalian PTH1Rs, COS-7 cells expressing the zPTH1R showed an equivalent increase of IP accumulation (2-fold) when stimulated with either hPTH or hPTHrP. In contrast, despite higher expression levels, no IP accumulation was detectable when cells expressing zPTH3R were challenged with either ligand (Fig. 5). It remains to be determined which portion(s) in the zPTH3R confer this alteration in signaling, and whether any of the amino acid residues that were previously identified in the rat PTH1R as being important for IP signaling could provide an explanation for the differences in phospholipase C activation (19).
Southern Blot Analysis Using cDNA Probes That Are Specific for the zPTH1R or the zPTH3R-To confirm that zPTH1R and zPTH3R are encoded by distinct genes, three infrequently cutting restriction endonucleases were utilized to digest zebrafish genomic DNA to completion (Fig. 6). Initial Southern blot data using probes corresponding to TM3 through TM6 of either receptor showed multiple hybridizing DNA species, indicating that these probes cross-hybridized with each other's gene or with closely related genes (data not shown). To increase the specificity for either subtype, hybridizations were performed with radiolabeled probes comprising only the tail region of each receptor, zPTH1R/tail or zPTH3R/tail, which showed the highest rates of sequence variation between PTH receptor subtypes. For each digest the tail probes hybridized, under stringent conditions, to single but different genomic DNA fragments, indicating that distinct genes encode zPTH1R and zPTH3R.
Phylogenetic and Structural Analyses of the zPTH1R and the zPTH3R-The single most parsimonious Bootstrap consensus tree revealed two clades, the PTH2R clade and the PTH1R/ PTH3R clade. Within the PTH1R/PTH3R clade, the PTH3R grouped in a significantly different manner from the PTH1Rs, and the terminal branches for the latter group were found to be congruent with morphology-based phylogenies (25) (Fig. 7). Although additional data are required, multivariate analysis suggested that some amino acid residues are PTH3R-specific, and some of these could explain the functional differences for both receptors, particularly with regard to phospholipase C activation.
In summary, two cDNAs were isolated, which encode related but distinct PTHRs. Despite the considerable phylogenetic distance between teleosts and mammals, the PTH1Rs from these divergent species were found to be sufficiently alike to allow an equivalent interaction with mammalian G proteins, and efficient activation by PTH and PTHrP analogs. PTH1R and PTH2R from all known species thus appear to have similar structural and functional properties, making it likely that the mammalian homolog of the zPTH3R will show a similar ligand selectivity as the teleost receptor.