Functional Characterization of the Alternatively Spliced, Placental Human Growth Hormone Receptor*

The human growth hormone family of peptide hor- mones is encoded by five genes, pituitary growth hormone (hGH-N), and four placentally expressed genes, growth hormone variant (hGH-V), chorionic somatomammotropin A and B (hCS-A, hCS-B), and prolactin (hPrl). As part of an effort to define the local effects of the placentally expressed members of the GH/Prl fam- ily of hormones on the placenta, we have identified an isoform (hGHRd3) of the growth hormone receptor expressed in the placental villi. hGHRd3 mRNA differs from the liver GHR mRNA by the deletion of a 66-base pair segment encoding exon 3. In this study we show that hGHRd3 mRNA encodes a stable and functional receptor. hGHRd3 mRNA is efficiently translated and processed in a rabbit reticulocyte lysate translation system as well as in an in vivo Xenopus laevis oocyte expression system. In Xenopus oocytes hGHRd3 is stably integrated into the cell membrane and binds and internalizes ligand as efficiently as hGHR. hGHRd3 binds all three of the placentally expressed members of the GH/Prl gene family (hGH-V, hCS, and Prl) as well as both the 22 and 20 kDa isoforms of the pituitary hGH-N. The results of the present study strongly support

hCS-A, hCS-B, and hGH-V are expressed exclusively by the syncytiotrophoblasts of the placental villi (Cooke et al., 1988a;Liebhaber, 1989;Cooke et al., 1991). In contrast to this highly restricted pattern of expression, hPrl is expressed in a number of tissues, most prominently lactotrope cells bf the anterior pituitary, endometrium, placental decidua, and lymphocytes (Cooke, 1993;Clements et al., 1983;and DiMattia et al., 1990). While the functions of hGH-N and hPrl are well established, the roles of the placentally expressed hCS, hGH-V, and hPrl are less well understood.
The functions of GH and Prl are mediated by specific binding to a set of cell surface receptors belonging to the cytokine receptor superfamily (Cosman et al., 1990). Receptors in this superfamily, which encompass receptors for a broad spectrum of growth factors and peptide hormones, are characterized by a large amino-terminal extracellular domain which contains four highly conserved cysteine residues, a single transmembrane domain, and a cytoplasmic domain of variable size. Although no kinase or other enzyme activities have been assigned to these receptors, several, including the erythropoietin and GH receptors, are phosphorylated in response to ligand binding (Foster et al., 1988;Taga and Kishimoto, 1992;and Linnekin et al., 1992). As a general rule these receptors dimerize upon ligand binding, then internalize as a ligand-receptor complex. The mode of signal transduction for this superfamily of receptors remains undefined.
The expression of certain GH/Prl genes by the placenta raises the question of whether they might be acting locally on this rapidly growing organ as well as on more distant tissues. hGH-V, the most recently characterized of the hormones in this group, binds to both somatogen and lactogen receptors with a binding profile that defines it as the most purely somatogenic human hormone (Ray et aL., 1990). Considering the placental expression of this potent somatogen as well as the two locally expressed lactogens, hCS and hPrl, it is necessary to consider the potential for autocrine and/or paracrine activities in addition to their effects on distant tissues. Such local activities would necessitate the presence of a corresponding set of receptors on the surface of placental cells.
As part of an effort to define the local effects of the GH/ Prl family of hormones on the placenta, we recently screened a placental cDNA library to detect expression of the hGH receptor (hGHR) mRNA. We found that the placental villi are enriched for hGHR mRNA. Structural analysis of the placental villous hGHR mRNA revealed that it differed from the previously defined hepatic hGHR mRNA by the selective deletion of a 66 bp segment encoded by exon 3 (Urbanek et aL., 1992). This placental hGHR mRNA isoform, hGHRd3 mRNA, predicts the expression of a corresponding hGHRd3 isoform that differs from the hepatic hGHR by a deletion of 22 amino acids within the extracellular domain of the recep-

19026
Function of the Placental hGH Receptor (hGHRCl3) tor. While others have shown that there is no substantial effect of this region on hGH-N binding by bacterially expressed extracellular domains of the hGHR (Bass et d., 1991), the significance of this finding with respect to the biological activity of the full-length, processed, transmembrane receptor has not been addressed. Determining whether the hGHRd3 mRNA can encode a stable membrane-bound receptor and establishing its ligand binding profile is of central importance to understanding the potential roles of the placentally expressed hormones. The studies reported in the present paper demonstrate that the full-length, transmembrane hGHRd3 isoform encodes a stable and functional GH receptor. These results establish the potential for a GH-GHR axis within the placenta. Generation of cDNA Templates-Templates for in vitro synthesis of hGHR and hGHRd3 mRNAs were generated by joining together three overlapping cDNA segments by splice-overlap extension (SOE) polymerase chain reaction (PCR) (Horton et al., 1989(Horton et al., ,1990. Fragment A, common to both templates, contains the SP6 RNA polymerase promoter, the entire 5"nontranslated (5'-NTR) region and initiation codon of human @-globin, and the first 20 bp of the common hGHR/hGHRd3 coding sequence. The @-globin cDNA template for fragment A amplification, pSPk@cA, was amplified between a sense primer that contains a BamHI cloning site at its 5' end followed by the SP6 promoter sequence (primer 1: 5'-TATAGGATCCATTT-AGGTGACACTATA-3') and an antisense primer that contains the last 16 bases of the @-globin 5'-NTR, the adjacent initiation codon, and the first 17 bases of the coding sequence common to hGHR and hGHRd3 mRNA (primer 2: 5'-AGCAGCTGCCAGAGATCCATGG- . The unique middle fragments, B-R and B-Rd3, encompass the first 700 or 634 bp of the hGHR or hGHRd3 mRNA coding sequences, respectively. These fragments were amplified from either an hGHR (258~-13) or an hGHRd3 (258b-9) cDNA template. Both of these plasmids contain two separate single basepair deletions at nucleotides 719 and 1200 (the A of the AUG is designated +1) which lie outside the region of amplification. These amplifications utilize a sense primer which contains 16 bases of 5'-NTR and the AUG of @-globin mRNA and the first 17 bases of the common hGHR/hGHRd3 coding segment (primer 2': 5"ACCTC-AAACAGACACCATGGATCTCTGGCAGCTGCT-3') and an antisense primer which contains 18 bases of central coding region (bases 683-700 encoded in exon 7) common to hGHR and hGHRd3 (primer 3: 5'-GTTTGGATCTCACACGCA-3'). Fragment C, common to the two receptor mRNAs, extending from the central coding region through 85 bp of the 3'-NTR region was amplified from the hGHR cDNA plasmid 258-21 template. The sense primer corresponds to bases 612-631 of the coding region (primer 3': 5"GAAAATG-ATGGACCCTATAT-3') and the antisense primer is located in the 3'-NTR (primer 4: 5'-GCGGGATCCTTTAAACATTGTTTTGG- . Plasmid 258-21 contains three A to G point mutations at nucleotides 57, 153, and 351, all of which are excluded from the fragment prepared by PCR. Fragments were amplified from 20 ng of template plasmid for 25 cycles using Vent DNA polymerase (New England Biolabs). The cycles consisted of 1 min denaturation at 95 'C, 1 min annealing at 50 "C, and 2 min extension at 72 "C with the exception of a 3-min denaturation interval for the first cycle. Each reaction was resolved on a 1.0% agarose gel, and the amplified fragments were excised and purified. The @-globin 5'-NTR and two of the hGHR coding fragments (B-R and C, or B-Rd3 and C) were then joined by SOE/PCR to generate the final hGHR and hGHRd3 transcription templates, respectively. In each case 2% of each of the three fragments were mixed and amplified between primers 1 and 4. The PCR conditions were the same as described above except that the extension step was increased to 3 min. The full-length products generated by each of the two reactions were gel purified, and 2% of the final products were then used as template for reamplification between primers 1 and 4 for 25 cycles. This final amplification reaction was subsequently extracted with phenol-chloroform-isoamyl alcohol, ethanol precipitated, and used directly as a template for in vitro transcription.

Materials
I n Vitro Transcription and Translation-In vitro transcriptions were carried out using 1 pg of linearized plasmid DNA or 0.5 pg of cDNA fragment as template. The 2 0 4 transcription reaction was carried out in the presence of 0.10 p1 of [aSzP]CTP and 5 mM 'mG(5')ppp(5')G with all other reaction conditions as suggested by the manufacturer (MEGAscript SP6 RNA polymerase). After a 2-h incubation, transcripts were desalted on Sephadex G-50 columns to remove unincorporated cap analog and nucleotides. An aliquot of the final transcript was electrophoresed on a 6% acrylamide, 8 M urea gel and analyzed by autoradiography to document ita quantity and integrity.
In vitro translations were carried out in rabbit reticulocyte lysate in a 15-pl reaction volume at 30 "C for 2 h as detailed previously (Liebhaber et al., 1984). Where noted, translations were carried out in the presence of 1.5 pl of canine pancreatic microsomal membranes (CPMM) (Promega Corporation) and translated for 2.5 h. N-Linked carbohydrate residues were removed by incubating the translation reaction with 2 units of endoglycosidase H (endo H) at 37 "C for 6 h in 0.1% SDS and 0.1 M sodium citrate, pH 5.5. Aliquots of each reaction (1 or 5 pl) were analyzed directly or following immunoprecipitation on 8% SDS-PAGE.
X. lnevis Oocyte Injections-Experiments utilizing X. lneuis were reviewed and approved by the Institutional Animal Care and Use Committee of the University of Pennsylvania. Care of X. lnevis and procedures for oocyte harvesting followed standard protocol (Coleman, 1984;Gurdon and Wakefield, 1986;Huez and Marbaix, 1986). After anesthesia in 0.1% ethyl m-aminobenzoate (Tricaine, Sigma), ovaries from female X. lnevis were removed and treated with 2 mg/ ml collagenase until individual oocytes were free from their membra-Barth's saline (MBSH 88 mM NaCl, 1.0 mM KCI, 2.4 mM NaHC03, nous coverings. Released oocytes were thoroughly washed in modified 15 mM HEPES-NaOH, pH 7.6, 0.3 D M CaN03, 0.41 mM CaC12, 0.82 mM MgSO,). Individual oocytes were injected with approximately 40 nl of desalted transcription reaction. Oocytes were incubated overnight at room temperature in MBSH supplemented with 100 units/ ml penicillin and 100 mg/ml streptomycin sulfate (Life Technology Inc.) before being used in subsequent experiments. Oocytes which were nonviable by visual inspection were discarded prior to analysis.
Analysis of Injected mRNA-To assess the stability of the injected RNAs, transcripts were labeled to a high specific activity, injected, and incubated as detailed above. At the noted times oocytes were snap-frozen on dry ice and stored at -70 "C until processed. Oocytes were subsequently homogenized and incubated at room temperature for 45 min in 75 pl of RNA extraction buffer/oocyte (2% SDS, 20 mM NaCl, 5 mM MgCl2, 0.2 M Tris, pH 7.5, and 1 mg/ml proteinase K). NaCl (final concentration of 200 mM) and carrier tRNA were subsequently added to samples that were extracted twice with phenolchloroform-isoamyl alcohol, then precipitated with 2 volumes of ethanol. RNAs were analyzed by electrophoresis on a 6% acrylamide, 8 M urea gel with subsequent autoradiography.
Receptor Binding Assays-Oocytes injected with hGHR mRNA, hGHRd3 mRNA, or uninjected controls were incubated overnight at room temperature and washed with fresh MBSH. An experimental sample consisted of five identically injected oocytes incubated for an additional 4 h at 4 "C in 50 pl of MBSH containing 2 mg/ml BSA, approximately 25,000 counts/min '"I-hGH-N, and the indicated concentrations of unlabeled hormones. Unbound '"I-hGH was subsequently removed as described with a few modifications (Peacock et , 1988). Oocytes were rinsed briefly in ice-cold MBSH containing 2 mg/ml BSA, layered over 400 p1 of a 1:1.5 mixture of dibutyl phthalate-dioctyl phthalate in a 400-p1 microfuge tube, and spun for 10 s at 6,000 revolutions/min. The overlaying oil and aqueous layer were discarded, and the radioactivity from pelleted oocytes quantitated in a gamma counter. Each data point represents an average value from four separate experimental samples containing five oocytes each. Reported 50% maximal binding values represent the average of at least two independent determinations.
Preparation of Conditioned Medin-The generation of stably transfected mouse fibroblast C127 cell lines which express hGH-N (Cooke et al., 1988b), 22 or 20-kDa hGH-N (Estes et al., 1992), or hGH-V (Cooke et al., 1988b) has been previously described. Each of these cell lines was grown to 90% confluence in minimal essential media (JRH Biosciences, Lenexa, KS) supplemented with 10% fetal bovine serum, 2 mM glutamine (Sigma), 100 unita/ml penicillin, and 100 mg/ml streptomycin sulfate, rinsed with PBS, and incubated overnight with serum-free minimal essential media. These conditioned media were concentrated using Amicon Centriprep-10 concentrators (Amicon Division, W. R. Grace & Co, Beverly, MA), and the amount of GH was quantitated using an enzyme-linked immunosorbant assay prior to use. For enzyme-linked immunosorbant assay, serial dilutions (1:40, 1230, 1:160, 1:320, and 1:640) of each sample were made using media conditioned by nontransfected C127 cells as diluent. Each sample volume was then expanded with PBS to 1.25 ml and applied to a single well of a 96-well Linbro Titertek plate and incubated overnight at 4 'C. The fluid in each well was subsequently removed and the wells blocked with 400 pl of PBS, 0.5% gelatin for 1 h at room temperature. Wells were emptied and washed twice with PBS, 0.05% Tween 20, 50 pl of 1:lOO dilution of rabbit anti-hGH antibody (Liebhaber et al., 1986) in PBS, 0.5% gelatin, 0.05% Tween 20 were added to each well, and incubated 1 h at room temperature. Plates were then washed three times with PBS, 0.05% Tween 20 followed by the addition of 50 p1 of a 1:2000 dilution of peroxidase-conjugated goat anti-rabbit IgG in PBS, 0.5% gelatin, 0.05% Tween 20. After 1 h of incubation at room temperature, the secondary antibody was removed, wells were washed five times with PBS, 0.05% Tween 20, 100 pl of fresh OPD color reagent was added (0.4 mg/ml o-phenylene diamine and 0.4 pl of HzOZ in citrate buffer), and the absorbance at 450 nm was measured in a Titertek Multiskan Plus (ICN Flow, Irvine, CA).
Western Blotting-The integrity of hormones in the conditioned media was determined by Western immunoblotting. The volume of conditioned media which contained 8 ng of hormone as determined by enzyme-linked immunosorbant assay was assayed on an 8% SDS-PAGE (Sambrook et al., 1989). Protein was electrotransferred from the gel to a nitrocellulose sheet for 2 h at 0.4 amps in Tris-glycine transfer buffer (39 mM glycine, 48 mM Tris base, and 20% methanol). The hormones were detected using a rabbit anti-hGH antibody (Liebhaber et al., 1986) and the antigen-antibody complexes were visualized by an enhanced chemiluminescence (ECL) Western blotting detection system (Amersham Corp.). Primary antibody was used at a 1:lOO dilution in 10 ml of blocking buffer (PBS, 0.05% (v/v) Tween 20, and 5% (w/v) nonfat dry milk), and the second antibody was used at a 1:500 dilution in 10 ml of blocking buffer.
Analysis of Receptor Internalization-Oocytes injected with receptor mRNAs were incubated overnight at room temperature in MBSH. Intact oocytes were incubated for 2 additional h at 4 "C with approximately 25,000 counts/min "'1-hGH-N with or without an excess of cold hGH-N (2 mg/ml). The oocytes were then washed twice with ice-cold MBSH containing 2 mg/ml BSA and rapidly brought to room temperature to allow internalization of the bound hormone. At the indicated times three samples of five oocytes each were removed and assessed for ligand internalization. Surface-bound hormone was released by incubating the ice-cold oocytes twice for 5 min each with 1 ml of glycine buffer (50 mM glycine, 100 mM NaC1, pH 2.5) (Rodriguez et al., 1992). These acid washes were pooled and designated the surface-bound fraction. The radioactivity associated with the acidtreated oocytes was designated as the internalized fraction. Specific binding to the injected oocytes was defined as total binding in the absence of unlabeled hGH-N competitor minus total binding in the presence of an excess of unlabeled competitor.

RESULTS
Synthesis of hGHR and hGHRd3 mRNA-We have previously demonstrated that hGHR gene transcripts are alterna-tively spliced in the placental villi to exclude exon 3. This hGHRd3 mRNA predicts the expression of an hGH receptor isoform with a 22-amino-acid deletion in its extracellular domain. In the present study we attempt to express this putative hGHRd3 receptor in a membrane-associated form and determine whether it can bind GH or GH-related hormones. These studies were carried out by microinjecting hGHR and hGHRd3 mRNAs into Xenopus oocytes. Initial attempts to clone full-length hGHR and hGHRd3 cDNAs for use as templates for in vitro transcription were unsuccessful. Numerous full-length GHR cDNAs were isolated either by screening a placental cDNA library (Urbanek et al., 1992) or by RT-PCR cloning from placental RNA. In each case fulllength inserts contained one or more rearrangements, missense mutations, or nonsense mutations (data not shown). After screening a large number of such clones, we concluded that the bacterial strains that we were using (Escherichia coli HBlOl and DH5a) could not support the expression of an intact hGHR cDNA. To circumvent this problem full-length transcription templates were generated by direct transcription of amplified full-length cDNAs using a SOE/PCR strategy (see "Experimental Procedures"). In these templates the 5'-NTR of the hGHR cDNA was replaced by the full 5'-NTR of the human @globin cDNA to maximize translational efficiency in the Xenopus oocyte.2 For convenience these mRNAs are referred to as hGHR and hGHRd3 mRNAs with the understanding that the 5'-NTRs are @-globin in origin. The receptors encoded by each chimeric mRNA contain only hGHR sequences. The full-length hGHR and hGHRd3 transcripts generated from the SOE/PCR cDNA templates contain only trace amounts of prematurely terminated transcripts (Fig. 1A). The expected 66 base difference in their sizes can be clearly discerned on agarose gel analysis (data not shown).
In Vitro Translation of hGHR and hGHRd3 Transcr@ts-To establish that the SOE/PCR-generated transcripts can be translated, the synthetic hGHR and hGHRd3 mRNAs were incubated in rabbit reticulocyte lysate in the presence of [3sS] methionine. Incubation duration was prolonged to 2 h to obtain maximal levels of synthesis (Fig. 1B). An 84-kDa product was specifically generated by translation of the hGHRd3 and hGHR mRNAs. Although the predicted size for the unprocessed GHR is 70 kDa, previous studies indicate that deglycosylated GHR from IM9 cells migrates at 95 kDa on SDS-PAGE gels, significantly slower than predicted (Kelly et al., 1991). To demonstrate appropriate processing of the encoded proteins by the endoplasmic reticulum, a step critical to membrane receptor biogenesis, the in vitro translations were carried out in the presence of CPMM (Fig. 1C). hGHR and hGHRd3 have identical NH2 termini encompassing their signal peptide cleavage sites and they both contain N-linked glycosylation consensus sites (four and five sites, respectively). Translation of the hGHRd3 mRNA in the presence of CPMM results in an apparent net increase of 10 kDa (84-94 kDa). A parallel increase is seen in the hGHR mRNA translated under identical conditions (Fig. IC). Treatment with endo H, which cleaves the unmodified N-linked sugar residues, decreases the size of the translation products to the unprocessed preproteins. These results are consistent with in vitro expression and processing of the predicted hGHR and hGHRd3 by their respective synthetic RNAs.
Stability of the hGHRd3 and hGHR mRNAs in X. laevis Oocytes-To study the receptors on an intact cell surface, the two synthetic mRNAs were separately injected into the cytoplasm of Xenopus oocytes. To assess the utility of this A. B. system for expression and comparison of the hGHRd3 relative to the hGHR mRNA and to optimize the incubation time, the relative stabilities of these two mRNAs were assessed postinjection. Internally 32P-labeled transcripts were injected into Xenopus oocytes, and mRNA was extracted 0, 10,20, and 30 h post-injection and directly analyzed by gel electrophoresis (data not shown). Each transcript was coinjected with the highly stable human P-globin mRNA as an internal control for the efficiency of RNA harvest. The absolute and relative levels of hGHR and hGHRd3 mRNAs did not change substantially over 30 h of incubation. The stability of the hGHR and hGHRd3 mRNAs suggested that an overnight incubation of the injected oocytes would be a reasonable time for their expression and the assembly of their encoded receptors on the oocyte membrane.

X . laeuis Oocyte Binding Assays-A competition binding
assay was established to measure the interaction of ligands with the hGH receptors assembled on the surface of intact microinjected oocytes. The sensitivity and accuracy of the assay was established by determining the binding of recombinant GH to oocytes injected with the mRNA for the hGHR the binding affinity for this interaction is well established in a number of systems (Lesniak et al., 1977;Hocquette et al., 1989;Fuh et al., 1990;Spencer et al., 1990;and Dusquesnoy et al., 1991). Oocytes injected with hGHR mRNA and incubated overnight to allow expression of the hGHR were subsequently incubated with lz5I-hGH-N along with known amounts of competing unlabelled hGH. The relationship between lZ5I-hGH binding to the amount of cold competitor establishes the competition binding curve (Fig. 2). 100% binding, defined as the amount of binding which occurs in the absence of cold competitor, is between 5-10% of the total 1251-hGH-N added to the incubation, and the amount of binding to uninjected oocytes is generally between 5-10% of the maximal specific binding seen in oocytes injected with receptor mRNA. 50% maximal binding or EC, of hGH for the hGHR expressed on the surface of the injected oocytes is 14 Function of the Placental hGH Receptor (hGHRd3)
Having established that the Xenopus oocyte system is a suitable system in which to study the hGH-N and hGHR interaction, we next investigated whether hGH-N would show specific binding to oocytes injected with hGHRd3 mRNA. AS seen in Fig. 2.4, the competition binding profile of hGH-N to oocytes injected with hGHRd3 mRNA is indistinguishable from the profile of oocytes injected with the hGHR mRNA; EC5o of the hGHRd3 injected oocytes is 19 f 7 X 1 0 " ' M, essentially identical to the hGHR value of 14 f 9 X 10"' M.
Circulating pituitary hGH is present in two isoforms, 90% as a 22-kDa protein and the remaining 10% as a 20-kDa protein lacking an internal 15-amino-acid segment. Previous studies by others have revealed that the 20 kDa form has a unique receptor binding profile (McCarter et al., 1990;Kostyo et al., 1986;Smal et al., 1985Smal et al., ,1987Baumann and Shaw, 1990;Siege1 et al., 1981;Spencer et al., 1981;Mosier and Lewis, 1982;and Closset et al., 1983). We analyzed the ability of recombinant 20-kDa hGH-N to bind to the hGHR and hGHRd3 receptors (Fig. 2B). The 20-kDa hGH-N was generated by a stably transformed cell line expressing only this alternatively spliced product (Estes et al., 1992). The 20-kDa hGH-N isoform has a %fold weaker affinity for hGHR and hGHRd3 than does the 22-kDa hGH-N isoform, and ECW occurs at 39 f 15 x 10"' M for hGHRd3 and 64 f 9 X 10"' M for hGHR. Although the hGHRd3 has a slightly higher affinity for 20-kDa hGH-N than does hGHR in each of three independent binding experiments, the degree of variability in our system makes it difficult to state with certainty whether this difference is significant. The binding studies in Fig. 2 clearly demonstrate that the hGHRd3 mRNA encodes a stable receptor protein in Xenopus oocytes that binds the two isoforms of pituitary hGH-N. The hGHRd3 therefore has the potential to act as a functional hGH receptor.
Placental Hormone Binding Assays-The alternatively spliced hGHRd3 mRNA is selectively expressed in placental syncytial cells. To assess its ability to bind and potentially mediate the action of placentally expressed hormones of the GH gene cluster, we carried out a series of competition binding studies with each of the placentally expressed members of the family (hCS, hGH-V, and Prl). In these studies the binding was measured by the ability of the placental hormones to compete for binding with '251-hGH-N. In each case the hGHR and the hGHRd3 were compared. These studies are summarized in Fig. 3. Conditioned media from a mouse fibroblast cell line stably transfected with the hGH-V gene was used as the source of hGH-V. The results of the binding studies using the hGH-V media demonstrate that hGH-V binds hGHRd3 and hGHR with the same affinity (Fig. 3A). The ECso occurs at 10 f 1 X 10"' M for hGHRd3 and 12 f 2 X 10"' M for hGHR. Thus the binding affinities of hGH-V for the two receptors are quite similar to the affinities of hGH-N.
We next assessed the ability of the most abundant protein expressed by the placental villi, hCS, to bind to hGHRd3. Fig.  3B is a competition binding curve in which hCS competes with labeled hGH-N for binding to hGHR or hGHRd3 expressed in Xenopus oocytes. hCS binds equally to hGHR and hGHRd3, but its affinity is approximately 1000-fold lower than that of the 22-kDa hGH-N. ECw occurs at 11 f 1.7 x The prolactin gene is expressed in both the pituitary and the placental decidua. To further assess the ability of lacto-M for hGHRd3 and 17 2 4.2 X M for hGHR.
gens to bind to the hGHRs, oPrl was used as a prolactin source due to its availability in pure form. As seen in Fig. 3C, oPrl competes effectively with 22-kDa hGH-N for binding to hGHR or hGHRd3. There is no difference in the ability of oPrl to bind to the two forms of the hGHR. ECm occurs at 214 f 162 x 10"' M for hGHRd3 and 237 e 4 X 10"' for hGHR. The affinity of oPrl for the hGHRs is, therefore, 10-20-fold weaker than that of hGH-N.
Internalization of Hormone Receptor Complexes-Ligand binding to the GHR is followed by its internalization (Roupas and Herrington, 1989). To determine whether this post-binding internalization also occurs after ligand binding to hGHRd3, we established an internalization assay using the Xenopus oocytes based on protocols of previous investigators (Rodriguez et al., Peacock et al., 1988). Oocytes were injected with hGHRd3 or hGHR mRNA and incubated overnight to permit receptor synthesis and assembly on the cell surface. The receptors were then loaded with '*'I-hGH-N at 4 "C, and internalization was measured over time by determining the proportion of radioactive tracer that remained on the surface of the oocyte after shifting the cells to RT. Oocytes injected with hGHRd3 or hGHR mRNAs were studied in parallel. The rates of internalization of labeled ligand-bound to the hGHR and hGHRd3 are shown in Fig. 4. The ligandbound hGHRd3 internalized with the same kinetics as the hGHR. The internalization of the receptor-ligand complex as assayed by the increasing resistance of hormone to release with acid started almost immediately after shifting the oocytes to RT and was complete after approximately 50 min. Essentially the same results were generated in two additional experiments. These data suggest that the initial post-ligand binding steps mediating receptor internalization are identical for hGHRd3 and hGHR.

DISCUSSION
The importance of the exon 3-encoded region of the hGHR is unknown. All sequenced GHRs in mammalian species conserve this region while it is specifically absent in Prl receptor mRNAs (Boutin et al., 1988;Davis and Linzer, 1989;Smith et al., 1989;Adams et al., 1990;Cioffi et al., 1990;Edery et al., 1989;Hauser et al., 1990;and Urbanek et al., 1992). This pattern of evolutionary conservation suggests that the loss or retention of exon 3 by alternative splicing could have a major impact on receptor expression and/or function. In addition to affecting hGH-N ligand binding, deletion of exon 3-encoded sequences might affect receptor processing, transport, stability, binding to other related ligands, and/or signal transduction.
In this study we show that the exon 3-deleted isoform of hGHR encodes a stable and functional receptor. We demonstrate that hGHRd3 mRNA can be efficiently translated in an in uitro rabbit reticulocyte lysate translation system as well as in an in uiuo X. h u i s oocyte expression system. Furthermore, the latter system is demonstrated to be appropriate for ligand binding studies. In Xenopus oocytes we demonstrate that hGHRd3 is stably integrated into the cell membrane and efficiently binds and internalizes ligand. Our data indicate that hGHRd3 is as efficient as hGHR at the level of ligand binding or internalization. hGHRd3 binds all three of the placentally expressed members of the GH/Prl gene family, hGH-V, hCS, and Prl, as well as the 22-and 20-kDa isoforms of the pituitary hGH-N.
Of special significance is the finding that hGHRd3 is expressed in placental tissue as are two potential ligands, hGH-V and hCS. The high affinity of hGH-V for the receptor  Oocytes injected with hGHR or hGHRd3 mRNA were incubated at 4 "C with '%I-hGH. The oocytes were then brought to room temperature and the receptor-hormone complex allowed to internalize for increasing amounts of time at room temperature. The ratio of acid releasable (surface-bound) lZ61-hGH-N to the total 'T-hGH-N bound expressed as a percent is shown as a function of time. Each point represents the average and standard deviation of three data points utilizing five oocytes each. Internalization of the hGHRd3 complex is indicated by closed boxes and that of hGHR by open boxes. makes its potential for local interaction quite evident. Although hCS has only 1/1000 the affinity for hGHRd3 compared to hGH-V, a case can also be made for its potential significance as a ligand since hCS is found in the sera of pregnant women during the second and third trimester at levels that are approximately 1000-fold that of normal serum hGH-N and 300-fold that of hGH-V at term (Frankenne et al., 1988;Phillips and Vnencak-Jones, 1989;Walker et al., 1991). The fact that hGHRd3 may serve as a receptor for both hGH-V and hCS supports their potential to act in an autocrine fashion in the syncytial cells of the placental villi. What effect this autocrine stimulation might mediate is open to speculation, but the rapid growth of the placenta presents an obvious possibility.
The validity of the Xenopus oocyte expression system for the study of GHR and ligand interactions is supported by several lines of evidence. The ECso values that we obtained for hGHR using this approach correspond quite closely with those obtained in other systems (Lesniak et al., 1977;Hocquette et al., 1989;Fuh et al., 1990;Spencer et al., 1990;Dusquesnoy et al., 1991). In our system, only oPrl has a significantly different binding affinity for the GHR than has been reported in other studies. Monkey kidney COS cells transfected with the rabbit GHR have an affinity for oPrl that is approximately 50-fold weaker than that for hGH-N (Spenser et al., 1990). In IM9 cells, a human lymphoblast cell line that has hGH binding activity, oPrl competes approximately 1000-fold less efficiently for receptor binding than does hGH-N (Lesniak et al., 1977). The reasons for discrepancy among these results are not clear. Our results are reproducible with several different batches of hormone. Each oPrl sample that we used contained less than 0.1% contamination with other anterior pituitary hormones as determined by radioimmunoassay (see "Experimental Procedures"). The strong competition observed with oPrl cannot, therefore, be due to contamination with oGH. The differences in results may reflect details of each experimental model. For example, IM9 lymphocytes may express a multitude of yet uncharacterized receptors that could compete for oPrl binding and shift the binding curve to the right. Lymphocytes have recently been characterized as possessing Prl receptors, and the structure of the rat Nb2 lymphoma cell Prl receptor is unique, containing a short cytoplasmic tail (Pellegrini et al., 1992;Ali et al., 1992). In the COS cell experimental model, competition studies were carried out by binding to the rabbit GHR which may have a different affinity for hGH than does the hGHR.
Our findings using the full-length, transmembrane hGHR to study ligand binding complements several other studies. First, using a bacterially expressed soluble hGH-binding protein, Bass et al. (1991) demonstrated that this extracellular fragment of the hGHR binds to hGH equally well with and without the exon 3-encoded segment. Second, the recently established crystal structure of the hGH/hGH-binding protein complex has revealed that one molecule of hormone binds to two molecules of hormone-binding protein (De VOS et al., 1992). Therefore, sequences encoded by exon 3 are not implicated in either the ligand-receptor or receptor-receptor interaction. However, although hGHR and hGHRd3 have the same ligand binding and internalization profile, the tissue-specific splicing patterns of hGHRd3 and hGHR imply that hGHR and hGHRd3 may be biologically distinct. Therefore, if there are any functional differences between hGHR and hGHRd3, they would have to be at the level of signal transduction. Alternative splicing has been shown to impart different signal transduction activities to receptors (Pin et al., 1992;Nakamura et al., 1992;Doherty et al., 1992;Sokka et al., 1992;Danoff et ul., 1991;and Sommer et aL, 1990); however, in all these cases the alternatively spliced domain has been localized to the intracellular region. More recently, studies by Chiba et al. (1993) using chimeric cytokine receptors showed that the specificity of the signal transduction pathway which was activated in response to ligand binding was determined by the extracellular domain of the receptor. It is believed that the activation of the signaling pathway may be mediated via an accessory protein which recognizes specific extracellular domains. It is therefore possible that the presence or absence of exon 3 may allow the interaction of different accessory proteins which may activate different signal transduction pathways.
Another possible difference between the biological activity of hGHRd3 and hGHR in vivo may be at the level of expression via translational efficiency. In this study we used the 6globin 5'-NTR instead of the endogenous GH receptor 5'-NTR and therefore did not address the significance of different endogenous 5'-NTRs of the hGHR. Evidence suggests that there is extensive variation in the 5'-NTR of the hGHR mRNA (Leung et al., 1987;Godowski et al., 1989). Such differences, possibly reflecting utilization of alternative promoters, might affect the subsequent splicing as well as translational efficiency of the processed mRNA (Moldave, 1985;Ratner et al., 1987;Kozak, 1987;London et al., 1987;Kozak, 1988).
Finally, immunohistochemical studies directly support the expression of GHRs in the fetus and placenta. These studies have shown that GHRs are expressed at significant levels in fetal liver, pancreas, cerebral cortex, and epidermis as well as second trimester placenta (Hill et al., 1992). Immunoreactive GHR was also detected in the syncytial cells of the placental villi as early as 8 weeks of gestation, and strong staining for the GHR was detected at term (Hill et al., 1992). Decidual cells also stain for the GHR while cytotrophoblast, chorionic trophoblast, and amnion do not. This would indicate that hGHRd3 protein, which our earlier studies have shown is the only GHR mRNA isoform expressed in the placental villi (Urbanek et al., 1992), is translated at significant levels in the placenta. Taken together, the results of the present studies in the context of these previous reports strongly support the expression of a functional hGHRd3 isoreceptor in the placenta. The function(s) that this receptor may serve in autocrine, paracrine, and/or endocrine activation can now be addressed.