Cloning of a carboxyl-terminal isoform of the prostanoid FP receptor.

An FP prostanoid receptor isoform, which appears to arise from alternative mRNA splicing, has been cloned from a mid-cycle ovine large cell corpus luteum library. The isoform, named the FPB receptor, is identical to the original isoform, the FPA, throughout the seven transmembrane domains, but diverges nine amino acids into the carboxyl terminus. In contrast to FPA, whose carboxyl terminus continues for another 46 amino acids beyond the nine shared residues, the FPB terminates after only one amino acid. The FPA isoform appears to arise by the failure to utilize a potential splice site, while a 3.2-kilobase pair intron is spliced out from the FP gene to generate the FPB isoform mRNA. The two isoforms have indistinguishable radioligand binding properties, but seem to differ in functional coupling to phosphatidylinositol hydrolysis. Thus, in COS-7 cells transiently transfected with either the FPA or the FPB receptor cDNAs, prostaglandin F2α stimulates inositol phosphate accumulation to the same absolute maximum, but the basal level of inositol phosphate accumulation is approximately 1.3-fold higher in cells transfected with the FPB as compared with cells transfected with the FPA isoform. Using the polymerase chain reaction, mRNA encoding the FPB isoform was identified in the ovine corpus luteum.

An FP prostanoid receptor isoform, which appears to arise from alternative mRNA splicing, has been cloned from a mid-cycle ovine large cell corpus luteum library. The isoform, named the FP B receptor, is identical to the original isoform, the FP A , throughout the seven transmembrane domains, but diverges nine amino acids into the carboxyl terminus. In contrast to FP A , whose carboxyl terminus continues for another 46 amino acids beyond the nine shared residues, the FP B terminates after only one amino acid. The FP A isoform appears to arise by the failure to utilize a potential splice site, while a 3.2-kilobase pair intron is spliced out from the FP gene to generate the FP B isoform mRNA. The two isoforms have indistinguishable radioligand binding properties, but seem to differ in functional coupling to phosphatidylinositol hydrolysis. Thus, in COS-7 cells transiently transfected with either the FP A or the FP B receptor cDNAs, prostaglandin F 2␣ stimulates inositol phosphate accumulation to the same absolute maximum, but the basal level of inositol phosphate accumulation is approximately 1.3-fold higher in cells transfected with the FP B as compared with cells transfected with the FP A isoform. Using the polymerase chain reaction, mRNA encoding the FP B isoform was identified in the ovine corpus luteum.
Prostanoids, autacoids formed from arachidonic acid by the actions of cyclooxygenases, exert diverse physiological effects throughout the body. For example, in sheep and in many other species, prostaglandin F 2␣ (PGF 2␣ ) 1 is the trigger that initiates luteolysis or regression of the corpus luteum in the absence of pregnancy (1). There are five primary prostanoids, prostaglandin D 2 (PGD 2 ), prostaglandin E 2 (PGE 2 ), PGF 2␣ , prostaglandin I 2 (PGI 2 ), and thromboxane A 2 (TXA 2 ). Unique Gprotein-coupled receptors have been cloned for each prostanoid, including four receptors specific for PGE 2 (EP 1 -EP 4 ) (2). The other cloned receptors are named DP, FP, IP, and TP and bind PGD 2 , PGF 2␣ , PGI 2 , and thromboxane A 2 , respectively. In ad-dition, alternative mRNA splice variants, have been cloned for the EP 3 receptor (3,4) and for the TP receptor (5,6). In each case, the splice variants are identical throughout the seventransmembrane domains, but diverge approximately 9 -12 amino acids into the carboxyl terminus. The significance of this carboxyl-terminal alternative splicing is not well understood, but for the EP 3 receptor, differences among the isoforms have been found in localization (7), receptor/G-protein coupling (3, 8 -10), and desensitization (11).
Phylogenetic analysis of the prostanoid receptors shows that the receptors segregate into two branches (12,13). One branch contains the DP, IP, EP 2 , and EP 4 receptors. The second branch contains the EP 1 , EP 3 , FP, and TP receptors. Since isoforms have been found for two of the four receptors in the second branch, the EP 3 and TP receptors, we hypothesized that isoforms might also exist for the FP receptor.
To examine this possibility, we screened an ovine mid-cycle large cell corpus luteum library. Previously, we used this library to clone the ovine homologue (14) of the human FP receptor (15) and found that the message for the FP receptor was highly abundant (ϳ0.1% of the total message). Using a combination of homology-based screening with the ovine FP receptor as a probe and PCR, we cloned a novel isoform of the FP receptor. This isoform, the FP B , is identical to the original (termed here the FP A isoform) throughout the seven-transmembrane domains, but diverges nine amino acids into the carboxyl terminus. Functionally, both isoforms are able to stimulate inositol phosphate (IP) accumulation to the same maximum, but the basal level of hydrolysis is 130% higher for the novel FP B isoform than for the original FP A isoform.

EXPERIMENTAL PROCEDURES
cDNA Cloning-A cDNA containing the complete coding sequence of the ovine FP A receptor (14) was labeled with 32 P by nick translation (Life Technologies, Inc.) and an ovine large cell corpus luteum library screened as described previously (14). The library was plated at a density of ϳ3400 plaques/plate (15 cm), and five plates were screened. From a total of 24 positives, 18 were isolated and each was placed in 250 l of H 2 O.
To differentiate possible carboxyl-terminal splice variants from the original FP A isoform, two rounds of PCR were used. In the first round, sense (nt 258 -275) and antisense (nt 638 -654) primers upstream of the putative sixth transmembrane domain were used. These primers were predicted to be specific for the common region of all FP isoforms. In the second round of PCR, a sense primer (nt 793-810) upstream of the sixth transmembrane domain and an antisense primer (nt 1168 -1185) specific for the carboxyl terminus were used that could amplify the original FP A isoform but would be unlikely to yield products (at least of the predicted size) with carboxyl-terminal splice variants. The isolated positive plaques were vortexed for 15 s and lysed by three freeze/thaw cycles consisting of freezing in a dry ice/ethanol bath and thawing at 55°C for 5 min. For each 50 l PCR, 26 l of the phage lysate was added to a 50 l reaction containing final concentrations of 1 ϫ PCR buffer I (Perkin-Elmer), 10% dimethyl sulfoxide (Sigma), 200 M dNTPs (Perkin-Elmer), and 1 M each primer. The samples were heated to 95°C for * This work was supported by Allergan, Inc., with additional support provided by National Institutes of Health Grants EY09355 (to J. W. R.) and HD26778 and HD00907 (to P. B. H.). 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.
§ Supported by an individual fellowship from the National Science Foundation. To whom correspondence should be addressed: Dept. of Pharmacology and Toxicology, College of Pharmacy, University of Arizona, Tucson, AZ 85721. Tel.: 520-626-4367; Fax: 520-626-2466; E-mail pierce@tonic.pharm.arizona.edu. 1 The abbreviations used are: PG, prostaglandin; bp, base pair(s); IP, inositol phosphate; nt, nucleotide(s); RT, reverse transcription; PCR, polymerase chain reaction; PI, phosphatidylinositol; DMEM, Dulbecco's modified Eagle's medium. 5 min, chilled on ice, 2.5 units of Taq polymerase (Perkin-Elmer) were added, and 30 l of mineral oil overlaid. The PCR conditions were as follows: an initial denaturation step at 95°C for 2 min followed by 30 cycles at 95°C for 30 s, 42°C for 30 s, and 70°C for 30 s. The PCRs were held at 4°C until analyzed by agarose gel electrophoresis.
Clones that gave positive results in the first round of PCR, but negative results in the second round of PCR, were isolated using standard plaque purification techniques. The inserts were subcloned into pBluescript (Stratagene) and sequenced. One clone (KSϩ/FP B ) that was otherwise identical to the FP A but diverged in the carboxyl terminus was identified.
Construction of a Full-length FP B Isoform and Expression Vector-KSϩ/FP B lacked the first 69 bp of coding sequence, and a full-length clone was obtained by replacing a HindIII fragment from the original FP A isoform, KSϩ/FP (14), with a HindIII fragment from KSϩ/FP B . KSϩ/FP and KSϩ/FP B were restricted with HindIII which cuts at nt 638 ( Fig. 1, single underlining) and in the multiple cloning site of KSϩ downstream of the FP B coding sequence. The large fragment containing KSϩ plus the 5Ј end of the FP A receptor and the small fragment containing the 3Ј end of FP B from the FP B digestion were isolated using Geneclean II (BIO 101, Inc.), ligated, subcloned, and KSϩ/FP B /fulcod isolated. To make an expression vector, KSϩ/FP B /fulcod was restricted with NaeI, which cuts at nt 80, and NcoI, which cuts at nt 1213, filled in using Klenow and subcloned into the blunted BamHI/HindIII sites of pBC12BI (16) to form pBC/FP B /fulcod.
PCR on Genomic DNA-PCR of sheep genomic DNA was performed using the Advantage Genomic PCR kit (Clontech). For both the FP A and the FP B receptor, the sense primer corresponded to nt 1018 -1044. For the FP A receptor, the antisense primer corresponded to nt 1291-1317, and for the FP B receptor, the antisense primer corresponded to nt 1078 -1104. Each 50-l reaction contained 2 l of sheep genomic DNA (ϳ200 ng total), 1 ϫ 10 Tth PCR buffer, 1.1 mM Mg(OAC) 2 , 1 M each primer, 200 M dNTPs, and 1 l of Advantage Tth Polymerase Mix (50 ϫ). 40 l of mineral oil was overlaid, and the PCR program used was as follows: 95°C for 1 min, 35 cycles at 95°C for 15 s, 68°C for 12 min, followed by a final extension at 68°C for 12 min. The reactions were held at 15°C until analyzed by agarose gel electrophoresis.
Reverse Transcription (RT)-PCR-Total RNA (ϳ5 g) from a day 10 ovine corpus luteum was reverse-transcribed in a 20-l reaction using reverse transcriptase (Superscript II, Life Technologies, Inc.) following the manufacturer's instructions using 200 ng of random hexamer primers. After the cDNA synthesis, RNA remaining in RNA-DNA duplexes was degraded by adding ϳ1 unit of RNase H and incubating for 30 min at 37°C. For the PCR, 2 l of the cDNA reaction was added to a 50-l reaction containing 10% dimethyl sulfoxide, 0.25 mM dNTPs (Perkin-Elmer), 1 M each primer, 1 ϫ PCR buffer, and 2.5 units of Taq polymerase (Perkin-Elmer). The PCR program was as follows: a 7-min denaturation step at 95°C, 36 cycles at 95°C for 1 min, 55°C for 1 min, and 72°C for 1 min, followed by 7 min at 72°C, and the reaction was held at 4°C.
IP Accumulation Assay-Receptor-mediated PI hydrolysis was determined by measuring the accumulation of IPs in cells that were preincubated with myo-[2-3 H]inositol (17-18 Ci/mmol, Amersham Corp.). COS-7 cells were plated in 10-cm dishes (10 6 cells/dish) in DMEM with 10% fetal bovine serum) and transfected the following day with the expression plasmids pBC/FP A or pBC/FP B /fulcod (4 g/dish), using Lipofectin (40 g/dish, Life Technologies, Inc.) following the manufacturer's instructions. On day 3, the cells were incubated with 3 C/ml myo-[2-3 H]inositol in DMEM, 10% fetal bovine serum. Cells were harvested on the following day, preincubated for 10 min in DMEM containing 25 mM HEPES buffer, 10 mM LiCl in order to block the dephosphorylation of IPs and aliquoted (200 l) with various concentrations of drug (50 l) in duplicate. After a 30-min, 37°C incubation, IPs were extracted with 750 l of chloroform/methanol/4 N HCl (100:200:2). An additional 250 l each of the chloroform and 0.5 N HCl were added, and the samples were vortexed. Unincorporated myo- [2-3 H]inositol was removed by adding 750 l of the resulting aqueous layer to a Dowex 1-X8 (formate, 100 -200 mesh, Bio-Rad) anion exchange column. Following three washes with 3 ml of 5 mM inositol, the IPs were eluted with 1.5 ml 1.3 M ammonium formate, 0.1 M formic acid and counts/min were determined by liquid scintillation counting.

RESULTS
We screened an ovine mid-cycle large cell luteal library with the previously cloned ovine FP receptor (14) and used PCR to identify possible carboxyl-terminal alternative mRNA splice variants. From a total of 17,000 plaques, one putative alterna-tive splice variant was identified. We have designated this clone FP B , in contrast to the original isoform which we now define as FP A . Fig. 1 shows the cDNA and deduced amino acid sequence of the ovine FP A and FP B isoforms. The FP A and FP B isoforms share the same sequence until they diverge nine amino acids into the carboxyl terminus. In the original isoform, the FP A , the carboxyl terminus continues for 46 amino acids beyond the splice site, while in the FP B isoform there is only one additional amino acid. Thus, the FP B is truncated relative to the FP A isoform. The immediate 3Ј-untranslated region of the mRNAs encoding these isoforms differ as well.
To determine how the two FP receptor isoforms arose, PCR was performed on sheep genomic DNA using primers that spanned the proposed splice site. The same sense primer (nt 1018 -1044) was used in both reactions in combination with an antisense primer specific to each isoform (FP A , nt 1291-1317; FP B , nt 1078 -1104). A picture of the agarose gel on which the reactions were electrophoresed is shown in Fig. 2. Lane 1 shows the size of the product obtained with the FP A -specific primer, and lane 2 shows the size of the product obtained with the FP B -specific primer. Thus, amplification with the FP A -specific primer yielded a product of ϳ300 bp, which is identical to the size predicted from the cDNA. Amplification with the FP Bspecific primer, however, yielded a 3200-bp product. This product is much larger than the 86-bp product that would be ex- pected from amplification of the FP B cDNA.
To verify that the message for the FP B was expressed in corpus luteum, RT-PCR was performed on mRNA obtained from a mid-cycle corpus luteum using primers flanking the coding region of the FP B isoform. The sense primer corresponded to nt 88 -111, and included the codon for the initiator methionine, and the antisense primer corresponded to nt 1081-1098 of the FP B receptor, downstream of the stop codon. Fig. 3 shows a photograph of the ethidium-stained agarose gel with the products of the PCR. Lane 1 is the product obtained from the ovine corpus luteum cDNA, lane 2 shows a positive control in which KSϩ/FP B /fulcod plasmid DNA was amplified with the same pair of primers, and lane 3 is the negative control. For both the ovine corpus luteum cDNA and the plasmid positive control, a product of the expected size, 1010 bp, was obtained. To verify that this 1010-bp product encoded the full-length FP B isoform, the bands from lanes 1 and 2 were isolated and digested with HindIII, which yielded products of the expected size, 459 and 551 bp. based on the presence of a HindIII site at nt 638 (data not shown).
To examine the pharmacology of the FP B clone, radioligand binding was performed using membranes from COS-7 cells that were transiently transfected with pBC/FP B /fulcod. Fig. 4 shows radioligand binding competition curves using 17-[ 3 H]phenyltrinor-PGF 2␣ as the radioligand. The ability of a series of natural and synthetic prostanoids to displace 17-[ 3 H]phenyl-trinor-PGF 2␣ binding was measured, and the rank order of potencies of the competitors was as follows: 17-phenyl-trinor-PGF 2␣ Ͼ PGF 2␣ Ͼ fluprostenol Ͼ PGD 2 ϭ PGE 2 Ͼ Ͼ 8-epi-PGF 2␣ . Thus, the synthetic prostanoid, 17-phenyl-trinor-PGF 2␣ , is the most potent with an EC 50 of ϳ10 nM, while PGF 2␣ is the most potent natural prostanoid, with an EC 50 of ϳ40 nM, although both PGD 2 and PGE 2 were also able to displace 17-[ 3 H]phenyl-trinor-PGF 2␣ binding. These results are nearly indistinguishable from those obtained for the FP A under similar conditions (14).
Since the FP receptor cloned from other species activates PI hydrolysis (17,18), we wanted to determine whether the ovine FP A and FP B isoforms could also stimulate IP accumulation and whether there were any differences between the isoforms.  1 shows the products obtained from PCR using a common sense primer and an antisense primer specific for the FP A isoform. Lane 2 shows the products obtained from PCR using the same sense primer with an antisense primer specific for the FP B isoform. The size of the products obtained is indicated to the right of the gel. PGF 2␣ was able to elicit the same maximal level of IP accumulation, 220% over basal, and the EC 50 was similar for the two isoforms, 8 nM for the FP A and 11 nM for the FP B isoform. However, the basal level of IP accumulation was ϳ130% higher in cells expressing the FP B isoform than in cells expressing the FP A , which in turn was ϳ140% higher than the basal level of hydrolysis in untransfected cells (data not shown). Based on radioligand binding competition studies, the level of receptor expression was higher for cells expressing the FP A receptor than for cells expressing the FP B receptor. In addition, similar experiments were repeated in COS-P cells, with essentially the same findings. DISCUSSION Previously, we and others reported the cloning of the FP prostanoid receptor from bovine (17), human (15), mouse (18), and ovine (14). Here we report the cloning of the FP B , an isoform of the ovine FP receptor. Isoforms have been reported for the EP 3 and TP prostanoid receptors, but this is the first report of isoforms of the FP receptor. Like the EP 3 and TP receptor isoforms, the FP receptor isoforms diverge in the carboxyl termini. Thus far, two isoforms of the human TP receptor (6) have been cloned, while a total of 13 different isoforms of the EP 3 receptor have been cloned among bovine, human, rabbit, and mouse (2).
The existence of FP receptor isoforms was hypothesized based on previously cloned EP 3 and TP receptor isoforms and relationships established by the phylogenetic analysis of the prostanoid receptors. This analysis places the DP, EP 2 , EP 4 , and IP receptors in one branch and the remaining EP 1 , EP 3 , FP, and TP receptors in the second. With the cloning of the FP B , isoforms have now been found for three of the four receptors in the second branch. Isoforms have not yet been reported for the EP 1 receptor, but the phylogenetic relationships suggest that they, too, may exist. Interestingly, the existence of isoforms also seems to correlate with second messenger coupling. Thus, the receptors in the second branch where isoforms have been reported are all primarily coupled to either PI hydrolysis or inhibition of adenylyl cyclase, while isoforms have not been obtained for the receptors in the first branch, which are all coupled to stimulation of adenylyl cyclase.
The TP and EP 3 receptor isoforms arise by alternative mRNA splicing at a location that is homologous to the site at which the FP isoforms diverge, suggesting that the FP receptor isoforms are also generated by alternative mRNA splicing. In the case of the TP receptor isoforms, the isoforms arise by the failure to utilize a potential splice site (6). Thus, the sequence of the carboxyl terminus of the endothelial variant is identical to the genomic sequence, while this sequence is spliced out to form the placental variant. It appears that the same mechanism generates the two FP receptor isoforms. Using the primers specific for the FP A receptor, the PCR product obtained from the genomic DNA is identical to the size of the product predicted from the cDNA sequence, indicating that no sequence is spliced out to generate this isoform. However, when primers specific for the FP B are utilized in a PCR on genomic DNA, the product obtained is much larger than predicted from the cDNA. Thus it appears that a ϳ3.2-kilobase pair intron is spliced out to generate the FP B receptor message. Consistent with the mechanism for the generation of these two isoforms, the highly conserved dinucleotide (GT) from the splice donor site is retained in the sequence of the FP A isoform (nt 1048 -1049) (19).
The functional significance of the carboxyl-terminal isoforms has been most well characterized for the ␣, ␤, and ␥ isoforms of the mouse EP 3 receptor. Agonist binding among the isoforms is virtually indistinguishable (8), but differences among the isoforms have been found for receptor/G-protein coupling. Tradi-tionally, the mouse EP 3 isoforms are coupled to inhibition of adenylyl cyclase, but some of the isoforms have been found to stimulate cAMP formation and/or activate PI hydrolysis (3,9,20). In addition, two isoforms, the EP 3␣ and EP 3␥ , and a recombinant mutant that is truncated at the splice site have been shown to have agonist-independent constitutive activity (21). Thus, in the absence of agonist, the level of forskolin-stimulated cAMP accumulation is greater in cells expressing the EP 3␤ isoform than in cells expressing either the EP 3␣ or EP 3␥ isoforms or the truncation mutant. However, maximal inhibition of forskolin-stimulated cAMP formation in the presence of agonist is the same for all three isoforms and for the truncation mutant. For the EP 3␣ isoform, there is agonist-independent constitutive activity and agonist-dependent inhibition of cAMP formation, while for the truncation mutant, there is only agonist-independent inhibition of cAMP formation. Data obtained with the EP 3␥ isoform suggest that there is agonist-independent stimulation of the G-protein, G i , but agonist-dependent stimulation of the G-protein, G s (22).
Our data with the FP receptor isoforms indicate that the receptor/G-protein coupling for the FP receptor isoforms are analogous to the data found for the mouse EP 3␣ and EP 3␤ isoforms. There is agonist-independent constitutive IP accumulation in cells expressing either the FP A or FP B receptor isoforms, but this level is approximately 130% higher for cells expressing the truncated FP B isoform than for cells expressing the FP A receptor isoform. In addition, for both the FP A and FP B isoforms, there is agonist-stimulated IP accumulation which reaches the same maximal level. This suggests that the FP B isoform is a naturally occurring truncated receptor that shows agonist-independent constitutive activity, but is still responsive to agonist. This is in contrast to the truncation mutant of the EP 3 receptor, which shows only agonist-independent constitutive activity and is unresponsive to agonist. The data with both the EP 3 and FP receptor isoforms are consistent with mutations made in other G-protein coupled receptors, including the avian ␤-adrenergic receptor, which suggests that sequences in the carboxyl-terminal tail may suppress G-protein interactions in the absence of ligand (23,24).
More work will be needed to establish if additional biochemical differences exist between the FP receptor isoforms. Of particular interest are possible differences in desensitization, localization, and ability to activate other downstream effectors. The physiological significance of the FP B isoform, and of other prostanoid receptor alternative splice variants, is also intriguing. In the mid-cycle large cell corpus luteum library from which the FP B was cloned, the FP A is far more abundant than the FP B . This, however, may not be true throughout the luteal cycle and in other tissues that are also responsive to PGF 2␣ .