Isolation and Characterization of a Novel Promoter for the Bovine Growth Hormone Receptor Gene*

The use of alternative promoters represents an important mechanism for the regulation of growth hormone receptor (GHR) gene expression. Two promoters have been isolated previously for the GHR gene: the P1 promoter that drives liver-specific expression, and the P2 promoter that drives ubiquitous expression. In the present study, we isolated a third GHR promoter termed P3. The P3 promoter was GC-rich and TATA-less. The P3 promoter was able to drive the expression of a luciferase reporter gene in cell lines Hep G2, PLC/PRF/5, and BHK-21. In vivo, the P3 promoter initiated transcription from two major sites in exon 1C of the GHR gene in many tissues. In the adult bovine liver, the P3-transcribed GHR mRNA represented only 10% of the total GHR mRNA pool. In non-hepatic tissues such as kidney, skeletal muscle, mammary gland, and uterus, P3-transcribed GHR mRNA represented 30–40% of the total GHR mRNA pool. Within the bovine GHR gene, the P3 promoter was located immediately downstream from the P2 promoter. In transfected cells, the P2 promoter served as an enhancer for the P3 promoter. Existence and co-regulation of two ubiquitous promoters may be a mechanism for achieving a high level of expression of the GHR gene in multiple tissues.

mediate the local effect of GH on both proliferation and differentiation at the cellular level, through IGF-I-dependent or -independent mechanisms (8 -11). Although the expression of GHR increases dramatically during postnatal life (5), functional GHR is also found in the early embryo (12) and fetus (13,14), suggesting a role of GH/GHR in early embryogenesis and fetal development.
Little is known about the mechanism regulating the ontogeny of the GHR gene. GHR mRNA is transcribed from a singlecopy gene (15,16), from which GH-binding protein, a shortened GHR lacking transmembrane and intracellular domains, is also generated by alternative splicing (2,4,17). The cDNA sequence for GHR has been determined in various species (2)(3)(4)(5)(17)(18)(19)(20). The structural organization of the GHR gene, however, is only known for the human. The human GHR gene is composed of 10 exons and spans 87 kb from exon 2 to exon 10. The translation start codon (AUG) is located in exon 2, and transcription is initiated from exon 1 (15). Cloning of GHR cDNAs in various species has revealed that the GHR cDNA are heterogeneous in their 5Ј-untranslated regions (5Ј-UTRs) (18,21,23,24) and that alternative 5Ј-UTRs are spliced onto a common splice site 11 bp upstream from the translation initiating codon AUG in exon 2. This has led to the hypothesis that the transcription of the GHR gene is controlled by multiple promoters in exon 1. Consistent with this hypothesis, a liver-specific GHR promoter (P1) has been isolated in various species including sheep (25), mouse (26), and human (27). The P1 promoter drives the expression of GHR in the liver of postnatal animals. A second GHR promoter (P2) has also been isolated in sheep and appears to drive the expression of GHR in various tissues including liver (13). Thus, transcription from multiple promoters probably represents an important mechanism controlling the amount and tissue specificity of GHR gene expression.
In this paper, we report the isolation and characterization of a novel promoter for the bovine GHR gene. Initially, we identified 5Ј-end sequences for the bovine GHR mRNA in the uterus in an attempt to isolate GHR with an NH 2 -terminal extended extracellular domain that acts as a unique receptor for bovine placental lactogen. This strategy was implemented because the bovine uterus has high affinity binding sites for bovine placental lactogen, and these binding sites are antigenically similar to the GHR (28). Using rapid amplification of cDNA ends (RACE), we isolated three new 5Ј-UTRs for the bovine GHR cDNA from uterus. However, RNase protection analysis (RPA) showed that the new 5Ј-UTRs were expressed in many other tissues including uterus. Cloning and transient transfection analysis of the genomic region corresponding to these 5Ј-UTRs have defined a third promoter (designated P3) for the bovine GHR gene.

EXPERIMENTAL PROCEDURES
RNA Isolation-Various bovine tissues were collected at slaughter. Tissues were immediately frozen in liquid nitrogen and stored at Ϫ80°C until used. Extraction of total RNA was carried out by using Trizol TM reagent (Life Technologies, Inc.) according to the manufactur-* This work was supported in part by National Research Initiative Competitive Grant USDA CSREES 95-37205-2312 (to M. C. L.). This is contribution 12837 from the Missouri Agricultural Experiment Station Journal Series. 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 /EBI Data Bank with accession number(s) AF085281 (1C3 5Ј-UTR).
‡ Supported by a Food for the 21st Century postdoctoral fellowship from the University of Missouri-Columbia.
er's instructions. The RNA samples were checked for integrity by gel electrophoresis. Concentrations of RNA samples were determined from absorbance of duplicate samples at 260 nm. Quantitation was confirmed by gel electrophoresis.
5Ј-RACE-Amplification of the 5Ј-end sequences of the bovine GHR cDNA in uterus was performed using 5Ј-RACE (Life Technologies, Inc.). 2 g of total uterine RNA was reverse transcribed into first-strand cDNA using SuperScript II reverse transcriptase and the reverse primer E4R (Table I), which was specific for the exon 4 region of the bovine GHR cDNA. After degrading the original RNA template with RNase H, the first-strand cDNA was extended from its 3Ј-end using terminal transferase to form a poly(dCTP) tail. The poly(dCTP)-tailed cDNA was then used as a template in 30 cycles of polymerase chain reaction (PCR) of 94°C for 30 s, 55°C for 45 s, and 72°C for 1 min with the reverse primer E3R (Table I) and the forward 5Ј-RACE abridged anchor primer 1 (AAP) ( Table I). The product of this PCR amplification was subjected to a second PCR amplification with abridged universal amplification primer (AUAP) and primer E2R (Table I) for 30 cycles with an annealing temperature at 58°C. The PCR components other than primers and template were included in Ready To Go PCR beads (Amersham Pharmacia Biotech). The PCR product of the second amplification was separated by gel electrophoresis in a 2% agarose gel, and bands of interest were gel purified by using Qiaquick (Qiagen, Valencia, CA). Purified PCR fragments were cloned into the PCR 2.1 vector (Invitrogen, Carlsbad, CA) and sequenced on an Applied Biosynthesis 377A DNA Sequencer (University of Missouri DNA Core, Columbia, MO). Sequence comparison was made by using the GCG program. Primers except AAP and AUAP were synthesized by the University of Missouri DNA Core. All enzymes were purchased from Promega (Madison, WI) unless otherwise indicated.
RPA-The RPA was used to determine the expression of 5Ј-UTR for GHR cDNA in bovine tissues and to map the transcription start site. The PCR 2.1 plasmid containing the longest 5Ј-UTR, named 1C3, was lineralized with SpeI. An antisense probe was generated by in vitro transcription using [ 32 P]CTP (800 mCi/mmol, NEN Life Science Products) and T7 RNA polymerase (Stratagene, La Jolla, CA). The specific activity of the probe was approximately 5 ϫ 10 8 cpm/g. The RPA was performed using the RPA II TM kit according to the manufacturer's instructions (Ambion, Austin, TX). Briefly, 5 ϫ 10 4 cpm of probe was hybridized with 20 g of total RNA in 80% formamide at 42°C. Following hybridization overnight, the reactions were digested in 200 l of a 1:100 diluted mixture of RNase A and RNase T1 at 37°C for 45 min. The protected fragments were analyzed by electrophoresis on a 6% polyacrylamide gel containing 8 M urea (Promega). The gel was dried and exposed to X-Omat AR film (Kodak) with intensifying screens at Ϫ80°C for overnight or longer.
The density of each protected fragment on the RPA autoradiogram was measured using GPTools (BioPhotonics Corp., Ann Arbor, MI). The relative amount of mRNA represented by each fragment was calculated by normalizing the density with the number of [ 32 P]CTP incorporated into each fragment (197, 87, and 18 CTP for 1C3, 1C2, and 52 bp of E2, respectively). The amount of total GHR mRNA was obtained by summing the amounts of 1C2-, 1C3-, and E2-containing GHR mRNA.
Identification of GHR Genomic Clones-We previously isolated genomic clones containing the bovine GHR 5Ј-flanking DNA by screening a bovine genomic library in vector EMBL3 SP6/T7 (CLONTECH) with probes generated from the cDNA for the bovine GHR 1A or 1B 5Ј-UTR (21,22). The GHR 1A and 1B 5Ј-UTRs are transcribed by the P1 and P2 promoter, respectively. We therefore used PCR to screen the GHR genomic clones for the presence of newly isolated 5Ј-UTRs. The PCR amplification was carried out on 2 l of phage stocks (1 ϫ 10 9 plaque-forming units in 50% glycerol) with two primers (1C3F1 and 1C3R1) that were specific for the 1C3 5Ј-UTR (Table I). The PCR was for 30 cycles of 96°C for 10 s, 55°C for 30 s, and 72°C for 2 min. The 1C3 was amplified from several GHR 1B genomic clones that were isolated using the 1B 5Ј-UTR as a probe. A 2.3-kb XhoI-XhoI fragment was released from one genomic clone and subcloned into the pBluescript SK plasmid (Stratagene) to form pSK 41-1-6. This plasmid was sequenced as described before. Restriction sites for NotI, SmaI, SacI, XhoI, and HindIII were identified within the insert by the GCG program and confirmed by restriction endonuclease digestion.
Promoter-Reporter Plasmid Constructs-The respective GHR genomic region was tested for its ability to drive the expression of the firefly luciferase reporter gene in the promoter-less plasmid pGL2-Basic (pGL2B) (Promega). A 2.3-kb EcoRI-SacI fragment digested from pSK 41-1-6 was inserted into the polycloning sites SmaI and SacI upstream from the firefly luciferase reporter gene in pGL2B. This construct was designated 2300C-GL2. The second construct, designated 1000C-GL2, was generated by fusing a 1-kb NotI-SacI fragment, released from the pSK 41-1-6 plasmid and blunted at the NotI end, into pGL2B at the polycloning sites SmaI and SacI. The orientation and sequence of the inserts were verified by sequencing through the insert-vector junctions. Large scale plasmid DNA purification was done using the Qiagen Maxiprep kit (Qiagen).
Cell Culture and Transient Transfection Assay-Cell lines were purchased from the American Type Culture Collection (ATCC) (Rockville, MD). Media and reagents for cell culture were purchased from Sigma unless otherwise indicated. Human liver cell line Hep G2 and hamster kidney cell line BHK-21 were maintained in minimal essential medium containing L-glutamine, 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin at 37°C under 5% CO 2 . A second human liver cell line PLC/PRF/5 was maintained in the same medium but without L-glutamine. Cells were plated on 12-well plates, each well containing 3 ϫ 10 5 Hep G2 cells or 1 ϫ 10 5 PLC/PRF/5 cells or 1 ϫ 10 5 BHK-21 cells. Following a 24-h incubation, cells in each well were transfected with 2 g of promoter-reporter construct and 0.002 g of pRLSV-40 plasmid (Promega) using the calcium phosphate method. The pRLSV-40 plasmid was co-transfected to normalize the variation in transfection efficiency. pRLSV-40 encodes the Renilla luciferase, and its activity can be distinguished from that of the firefly luciferase encoded in pGL2B in the Dual-Luciferase Assay System (Promega). After 12 h of transfection, the cells were washed with phosphate-buffered saline and then maintained in fresh medium for 36 h prior to luciferase assay. In each experiment, the pGL2B plasmid was also transfected in separate wells to compare the specific activity of promoter-reporter constructs with the basic activity of the promoter-less plasmid.
Dual-Luciferase Activity Assay-Activities of the firefly luciferase and Renilla luciferase in a single sample were measured sequentially using the Dual-Luciferase Reporter Assay (DLR TM ) system (Promega) according to the manufacturer's instructions. Briefly, cells were rinsed twice with phosphate-buffered saline and then lysed in 200 l of Passive Lysis Buffer at room temperature for 15 min. 20 l of cell lysate was quickly mixed with 100 l of Luciferase Assay Reagent II in a luminometer tube. The light emission for the firefly luciferase was recorded immediately for 15 s after a 3-s premeasurement delay using a TD-20e Luminometer (Tuner Designs, Sunnyvale, CA). Subsequently, 100 l of Stop&Glo TM reagent was added to the same tube to inactivate the firefly luciferase while activating the Renilla luciferase. The light  Fig. 1.
output from the Renilla luciferase was integrated under the same conditions. Variation in transfection efficiency was normalized by dividing the measurement for the firefly luciferase activity with that for the Renilla luciferase activity. The luciferase activity of a construct plasmid was expressed as relative to that of pGL2B.

RESULTS
Novel 5Ј-UTRs for the Bovine GHR mRNA-Three distinct PCR products (designated 1C1, 1C2, and 1C3) were obtained after amplifying the 5Ј-end sequence of the bovine GHR mRNA in the uterus by using the RACE procedure (Fig. 1A). Cloning of these three PCR products and subsequent DNA sequencing showed that 1C1, 1C2, and 1C3 were each flanked by the reverse primer E2R (used in the last PCR amplification) at the 3Ј-end (Fig. 1B) and the tailed poly(dCTP) sequence at the 5Ј-end. This indicated that they were specific RACE products. Comparison of sequences of 1C1, 1C2, and 1C3 with the previously isolated bovine GHR 1A and 1B 5Ј-UTRs (21) revealed that they all shared an identical 52-bp region of exon 2 in the bovine GHR cDNA from their 3Ј-termini ( Fig. 2A). The 5Ј-UTRs, however, diverged in the remaining upstream sequence, 11 bp upstream from the translation start codon ATG ( Fig. 2A). Thus, the newly isolated RACE products were new 5Ј-UTRs for the bovine GHR cDNA. Sequence comparison for 1C1, 1C2, and 1C3 also revealed that they were identical sequences with different extensions at the 5Ј-end (Fig. 1B). Whether 1C3, 1C2, and 1C1 are contiguous in the genomic sequence will be examined below.
Alignment of 1C3 with the eight 5Ј-UTRs for the human GHR mRNA (23) did not reveal significant similarity (between 30 and 50% identity) for DNA sequence except for the homologous 52-bp fragment of exon 2 (Ͼ95% identity). Sequence alignment of 1C3 with the five 5Ј-UTRs for the rat GHR mRNA (24) showed that rat V4 may be homologous to the bovine 1C (Fig. 2B). Approximately 60% of the nucleotides within the 144-bp rat V4 sequence (24, GenBank accession S77487) were identical to the bovine 1C; higher identity (75%) was found within an 80-bp exon 1 region proximal to exon 2 (Fig. 2B).
Besides the translation start codon (ATG) in exon 2, a second ATG codon was found 68 bp upstream of the former one within 1C3, 1C2, and 1C1 (Fig. 1B). However, the upstream ATG codon was not in-frame with the major open reading frame for the bovine GHR (3). Assumed translation from the upstream ATG codon would lead to an early termination (codon TGA) within 30 amino acids. Therefore, expression of an NH 2 -terminal extended extracellular domain of the bovine GHR from 1C3, 1C2, or 1C1 is unlikely.
Expression of 1C3, 1C2, and 1C1 5Ј-UTRs in Bovine Tissues-An RPA was done to determine whether 1C3, 1C2, and 1C1 were expressed in non-uterine tissues. The RPA yielded two major protected fragments (Fig. 3). The long fragment was generated from protecting the 585-bp 1C3; the short fragment was likely generated from protecting the 247-bp 1C2. Besides uterus, expression of 1C3 and 1C2 was detected in most adult bovine tissues, with greater levels in liver, muscle, uterus, mammary gland, kidney, and adrenal gland than in lung, heart, pituitary gland brain, placenta, and spleen (Fig. 3). Expression of 1C3 and 1C2 was also detected in fetal liver and kidney (Fig. 3). Across all tissues examined, the fetal lung was the only location that did not appear to express 1C3 or 1C2. Expression of 1C3 and 1C2 was also expressed poorly in lung of adult cattle (Fig. 3). Although the RPA was not designed as a strict quantitative assay, the observed pattern of tissue distribution of 1C3 and 1C2 was comparable with that of the total GHR mRNA (2)(3)(4)(5).
The 1C1 5Ј-UTR, as isolated by RACE, was not detected by RPA, perhaps because the RPA was less sensitive than the RACE procedure. Alternatively, 1C1 may have been amplified from a shortened first-strand cDNA caused by incomplete reverse transcription and therefore was a RACE artifact. The 52-bp exon 2 fragment generated from GHR mRNA with other 5Ј-UTRs (i.e. GHR 1A and GHR 1B) was not shown (Fig. 3) because small fragments were electrophoresed out of the gel.
Proportions of 1C2-and 1C3-derived GHR Transcripts within the Total GHR mRNA Pool-Subsequent RPAs were done to determine the ratios of 1C2-and 1C3-derived GHR transcripts to the total GHR mRNA pool. Tissues that had a relatively high level of GHR mRNA were examined. The gel was electrophoresed for a shorter time so that the 52-bp fragment of exon 2 could be visualized (Fig. 4). The ratios of 1C2and 1C3-derived GHR transcripts to the total GHR mRNA pool are shown in Table II. In the adult bovine liver, where most of the GHR mRNA is transcribed from exon 1A by the liverspecific promoter P1 (25)(26)(27), 1C-derived GHR transcripts (1C3 ϩ 1C2) represented only 11% of the total GHR mRNA (Table  II). In non-hepatic tissues such as kidney, mammary gland, uterus, and muscle, 28 -35% of the total GHR mRNA was represented by 1C-derived transcripts. The percentages of 1Cderived GHR mRNA in non-hepatic tissues were significantly higher than that in liver (p Ͻ 0.05). Within fetal muscle and kidney, percentages of 1C-derived transcripts to the total GHR mRNA pool were similar to those observed in the adult tissues (Table II). Much less GHR mRNA was detected in the fetal liver than the adult liver because the primary GHR promoter in liver is P1, and the P1 promoter is not activated until birth (22). For the same reason, 1C-derived GHR transcripts represented a much higher proportion (30%) of the total GHR mRNA in the fetal liver than in the adult liver (Table II).
Within all tissues examined, expression of 1C2 was greater than 1C3 (p Ͻ 0.05) ( Table II). Ratios of 1C2 to 1C3 varied between 1.6 and 3.3 among various tissues (Table II). Higher ratios of 1C2 to 1C3 were found in liver and kidney than in other tissues (p Ͻ 0.05). Although the absolute amount of 1C2and 1C3-derived GHR mRNA as well as the total GHR mRNA appeared to be decreased in the uterus and mammary gland of pregnant cattle (Fig. 4), the ratios of 1C2 and 1C3 to the total GHR mRNA pool within various tissues were not affected by pregnancy. The present study was designed to measure the relative levels of 1C2-and 1C3-derived GHR mRNA to the total GHR mRNA pool. Absolute levels of 1C2-and 1C3-derived GHR mRNA across tissues as well as changes in these transcripts during pregnancy should be examined in controlled studies with additional animals.
Genomic Sequence of 1C3, 1C2, and 1C1 5Ј-UTRs-The DNA fragment corresponding to 1C3 was amplified by PCR from the bovine genomic clones that contained GHR 1B 5Ј-UTR. Sequencing of the subclone pSK 41-1-6 that contained a 2.3-kb XhoI-XhoI fragment of the GHR 1B genomic clone demonstrated that 1C3, 1C2, and 1C1 were contiguous in the genomic sequence ( Fig. 5; the DNA sequence has been deposited in GenBank under accession number AF046861) as they were in the cDNA sequence (Fig. 1). This indicated that 1C3, 1C2, and 1C1 are generated by initiating transcription from different sites in the same exon. The exon was named 1C in agreement with the nomenclature used for the bovine exon 1A and 1B, from which the 1A and 1B 5Ј-UTR are generated. Restriction enzyme mapping and DNA sequencing also revealed that the 3Ј-border of exon 1C is 770 bp downstream from exon 1B (Fig.  5). The exact distance of exon 1C to exon 2 remains to be determined. However, based on the preliminary results of PCR mapping and partial sequencing of several bovine GHR genomic clones, at least 20 kb separate exon 1C and exon 2 (data not shown). Preliminary results also indicated that exon 1A is at least 10 kb downstream from exon 1B and at least 6 kb upstream from exon 2 (Fig. 5).
Transcription Start Sites in Exon 1C-The 5Ј-terminal nucleotide of the 5Ј-end sequence is typically regarded as the transcription start site when the 5Ј-end of a mRNA transcript is isolated by the 5Ј-RACE procedure (30 -32). The 5Ј-terminal nucleotide A of 1C3 and T of 1C2 (Fig. 1B) were therefore assigned to be the transcription start sites in exon 1C. The assigned transcription start site for 1C2 was close to the transcription start site (8 bp apart) mapped by using RPA in which a DNA sequencing ladder was run alongside (Fig. 6). The 1C2 and 1C3 start sites are two major start sites in exon 1C because levels of 1C2-and 1C3-derived GHR mRNA are relatively high. The start site for 1C1 may be a minor start site or amplification

FIG. 2. Sequence alignment of the bovine 1C 5-UTR with the bovine 1A and 1B 5-UTRs (panel A) and with the rat 5-UTR V4 (panel B). In panel A,
only the 100-bp sequence upstream of the 3Ј-end of 1C is aligned with the corresponding regions within the previously isolated bovine 1A and 1B 5Ј-UTR variants (21,29); the three bovine 5Ј-UTRs share an identical 52-bp exon 2 region and diverge in the upstream region. In panel B, the 144-bp rat V4 sequence (24) is aligned with the corresponding region within the bovine 1C 5Ј-UTR (ϩ71 to ϩ215, Fig. 1B). In each 5Ј-UTR, exon 2 is italicized, and the translation initiation codon (ATG) is underlined.

FIG. 3. RPA of expression of 5-UTRs in bovine tissues.
The autoradiograph is representative of three repeated RPAs, in which tissues from two adult cows and one 8-month-old fetus were used. SK muscle, skeletal muscle. 20 g of total RNA was hybridized with 5 ϫ 10 4 cpm of 32 P-labeled riboprobe that contained a 585-bp antisense sequence of 1C3 5Ј-UTR. Two major protected fragments, each corresponding to the expression of 1C3 and 1C2 5Ј-UTR of GHR mRNA, are denoted with arrows. M indicates a 32 P end-labeled 100-bp DNA ladder (Promega). tRNA denotes the negative control containing 20 g of yeast tRNA.
FIG. 4. RPA of proportions of 1C 5-UTRs to the total GHR mRNA pool in bovine tissues. The autoradiograph is representative of two replicated RPAs. The RPA included liver, kidney, muscle, mammary gland, and uterine samples from nonpregnant or pregnant (p) cows, as well as a liver, muscle, and kidney sample from an 8-month-old fetus. Mam, mammary gland; Muscle, skeletal muscle. The RPA was carried out as described in Fig. 3. Three major protected fragments, each corresponding to 1C3, 1C2, and 52 bp of exon 2 (E2), are denoted with arrows. of 1C1 was due to RACE artifact.
Promoter Activity-The GHR promoter activity for 5Ј-UTR 1C3 and 1C2 in vivo was suggested by the relatively high expression of 1C3 and 1C2 5Ј-UTRs in hepatic and many nonhepatic tissues. The promoter activity of the 5Ј-flanking region of exon 1C was assessed in its ability to drive the expression of the luciferase reporter gene in the promoter-less pGL2B plasmid in liver-derived cell lines Hep G2, PLC/PRF/5, and kidneyderived cell line BHK-21. After transient transfection in these cells, construct 2300C-GL2, which contained a 2.3-kb 5Ј-flanking region of 1C2 (Fig. 5), showed 20-, 18-, and 15-fold greater activity than the promoter-less pGL2B plasmid alone (p Ͻ 0.05) in Hep G2, PLC/PRF/5, and BHK-21 cells, respectively (Fig. 7). It was possible that the promoter activity of this 2.3-kb region was associated with the P2 promoter because the P2 promoter and exon 1B were included in the upstream region of the 2.3-kb fragment. To examine this possibility, a second construct, 1000C-GL2, was made by including only the 1,000-bp 5Ј-flanking region of 1C2 fused to the luciferase reporter gene. The 1,000-bp 5Ј-flanking fragment excluded the 5Ј-flanking region of exon 1B (Fig. 5). As shown in Fig. 7, the second construct maintained significant promoter activity, being 11-, 15-, and 5-fold over the pGL2B plasmid alone (p Ͻ 0.05) in Hep G2, PLC/PRF/5, and BHK-21 cells, respectively. However, removal of the P2 promoter region also significantly decreased (p Ͻ 0.05) the promoter activity of 1000C-GL2 when transfected into BHK-21 and Hep G2 cells, compared with that of 2300C-GL2. This result indicated that transcription of 5Ј-UTR 1C3/1C2 is initiated by a separate functional promoter from the P2 promoter. The new promoter was named P3. The activity of promoter P3 can be enhanced by the upstream P2 promoter. DISCUSSION In this study, we initially isolated three new 5Ј-UTRs for the bovine GHR mRNA from bovine uterine total RNA by using RACE (Fig. 1, A and B). Expression of two of the 5Ј-UTRs, 1C2 and 1C3, was detected by RPA in various tissues including uterus (Figs. 3 and 4). Cloning of the corresponding GHR genomic region (Fig. 5) and subsequent transfection analysis (Fig. 7) defined a new promoter (P3) that initiated the transcription of these 5Ј-UTRs from different sites in exon 1C (Fig.  5). Isolation of a third promoter for the GHR gene adds to the complexity of the structural organization of the GHR regulatory region and the molecular mechanisms for the regulation of GHR gene expression.
Use of alternative promoters to initiate transcription has been demonstrated as a strategy to achieve developmental and tissue-specific expression of many genes including IGF-II (33), the aldolase A gene (34), ␥-glutamyl transpeptidase (35), mineralocorticoid receptor gene (36), and Na ϩ -Ca ϩ exchanger gene (ncx1) (37). In the case of the GHR gene, two promoters had been isolated and characterized before this study. The first TABLE II Ratios of 1C3-and 1C2-derived GHR mRNA to the total GHR mRNA pool and ratio of 1C2-derived GHR mNRA to 1C3-derived GHR mRNA The relative amounts of 1C3-, 1C2-, and E2-containing GHR mRNA were derived from RNase protection analysis of adult and fetal bovine RNA samples (a representative RPA is presented in Fig. 4). Data are expressed as mean Ϯ S.E. when multiple samples were analyzed. The numbers in the parentheses are the numbers of samples in the analysis. Data were analyzed for statistical significance by using analysis of variance (ANOVA) and Student's t test.  promoter, termed P1, was originally isolated in sheep (25) and later in mouse (26), cattle (29), and most recently in humans (27). The P1 promoter is liver-specific. The second promoter, termed P2, was only isolated in sheep (13), but its transcripts were also present in cattle (21,22) and humans (23). The P2 promoter appears to be a ubiquitous promoter that governs the expression of the GHR gene in various tissues including liver. The new GHR promoter, named P3, isolated in the present study, appears to be a second ubiquitous promoter for the bovine GHR gene. The evidence for this conclusion included that P3-derived 5Ј-UTRs were detected in all bovine tissues examined (Fig. 3) and that P3 was able to drive the expression of a reporter gene in both hepatic and non-hepatic cell lines (Fig. 7).
Although liver is the primary tissue for the expression of GHR, the activity of the P3 promoter seems to be more important in non-hepatic tissues. In the liver of adult animals, the P3-derived GHR transcripts represented only a small portion (11%) of the total GHR mRNA (Table II), whereas the majority of the GHR mRNA was transcribed from the liver-specific P1 promoter. In non-hepatic tissues, GHR transcripts generated from the P3 promoter represented about 30 -40% of the total GHR mRNA pool (Table II). Most, if not all, of the remaining GHR transcripts are probably transcribed from a second ubiquitous promoter, P2. Exon 1C-derived GHR 5Ј-UTRs (Fig. 3) and exon 1B-derived 5Ј-UTR (22) 2 are expressed in a similar pattern of tissue distribution: ubiquitous but with higher amounts in uterus, liver, muscle, kidney, and mammary gland than in other tissues. The consistency of the relative expression of 1C-and 1B-derived GHR mRNA across tissues was also demonstrated by the similar ratios of 1C-derived GHR mRNA in the total GHR mRNA pool across tissues in pregnant and non-pregnant animals, and in adult and fetal animals (Table II and Fig. 4). This suggests that the activities of P3 and P2 promoters might be co-regulated. The co-regulation mechanism of P3 and P2 promoters, which together generate most of the GHR transcripts in non-hepatic tissues, is consistent with the wide ranging actions of the GH-IGF-I axis in stimulating cellular growth, proliferation, and differentiation in multiple tissues (8 -11).
The P3 promoter was located 700 bp downstream from exon 1B in the bovine GHR gene (Fig. 5). The physical proximity of promoter P3 to P2 allows for an interaction between the two promoters as well as the co-regulation of their activities. The transient transfection analyses demonstrated that inclusion of the upstream P2 promoter region in the P3 promoter-reporter construct increased the activity of the P3 promoter (Fig. 7). This result indicated that the P2 promoter may serve as an enhancer for the P3 promoter, or that enhancer(s) for the P2 promoter may be shared by the P3 promoter. A binding site (CCAAT) for ubiquitous factor CTF/NF-1 and several SP1 sites (GGGCGG) are located in the proximal region of the P2 promoter. Perhaps the CTF/NF-1 and SP1 binding sites constitute a ubiquitous enhancer for both promoters. Similar mechanisms for co-regulation of two ubiquitous promoters were reported for the human aldolase gene (38). Whether this mechanism applies to the bovine GHR gene remains to be verified.
Although P3-derived GHR transcripts were ubiquitously expressed (Fig. 3), differential regulation of P3 activity between The RPA was done as described in Fig. 3. An antisense probe to 1C3 was hybridized with 20 g of bovine liver and kidney RNA. The sequencing ladder (lanes G, A, C, and T), corresponding to the complementary strand to 1C3 5Ј-UTR in Fig. 1, was generated with 32 P-labeled primer E2R (Table I)  Constructs 2300C-GL2 and 1000C-GL2 each contain 2,300-bp and 1,000-bp portions of the 5Ј-flanking region of the bovine GHR gene inserted immediately upstream from the firefly luciferase reporter gene in the promoter/enhancerless plasmid pGL2B. 2 g of promoter-reporter construct and 0.002 g of pRL-SV40 (encoding the Renilla luciferase) as transfection efficiency control were co-transfected into cells using the calcium phosphate method. The transfection of each construct was repeated five, five, and three times in BHK-21, Hep G2, and PLC/PRF/5 cells, respectively. Activity of the firefly luciferase and Renilla luciferase was assayed using the Dual-Luciferase Assay System (Promega). The activity of promoter-reporter plasmids, relative to that of the pGL2B plasmid, is expressed as mean Ϯ S.E. of repeated transfections. The difference between the relative luciferase activities of the two constructs within each cell line was compared by using Student's t test. a and b denote p Ͻ 0.05. For details on plasmid construction, transient transfection, and luciferase assay, see "Experimental Procedures." tissues cannot be excluded. Greater expression of 1C-derived 5Ј-UTRs in tissues such as liver, muscle, uterus, and kidney than in other tissues such as lung, heart, ovary, and brain ( Fig.  3) indicates that the P3 promoter may be controlled by constitutive as well as tissue-specific transcription factors. Previous studies have reported that the GHR mRNA in kidney (39) and in brain (40) (predominately P2-and P3-transcribed 5Ј-UTRs) was regulated differentially by GH and steroids from that in liver (predominately P1-transcribed 5Ј-UTR). In the present study, the levels of 1C-derived GHR mRNA and the total GHR mRNA in uterus and mammary gland appeared to be decreased in pregnant animals compared with non-pregnant animals (Fig. 4), further suggesting that P3 and P2 promoters may be subject to regulation by hormones. However, unlike the P1 promoter, whose activity increases dramatically after birth (22), developmental up-regulation of the activity of the P3 promoter in tissues such as liver and kidney was not suggested because similar levels of 1C3 and 1C2 5Ј-UTRs were detected in both adult and fetal tissues (Figs. 3 and 4).
Examination of the proximal region upstream of 1C2 or 1C3 within the bovine GHR gene failed to identify a consensus TATA box. Absence of a consensus TATA box supports the presence of at least two different 1C-derived 5Ј-UTRs because TATA-less promoters usually initiate transcription from multiple start sites (41,42). The proximal region of the P3 promoter had a high content of GC (83% in the 700-bp proximal region) and lacked a CCAAT box. These features were thought to be unique to many constitutive gene promoters (43) but now are also found in regulated genes (44 -46). A search for potential transcription factor binding sites using the computer program MatInspector (47) identified several sites for constitutively expressed transcription factor SP1 (48) and also a number of sites for regulated factors such as STAT, AP1, AP2, and AP4 in the proximal region as well as in the more upstream region of the P3 promoter. Thus, the molecular mechanism for the hormonal regulation of the P3 and P2 promoters remains an important area of study.
Mapping of the P3 promoter relative to exon 2 was not completed partly because of the large size of the exon 1 region in the GHR gene. Preliminary results generated from sequencing genomic clones indicate that there are at least 20 kb between exons 1C and 2. Although a definite map or DNA sequence for this region has not been established, our estimate is comparable to an estimate of 34 kb between 1B and exon 2 in the ovine GHR gene (13).
In most species, one or two 5Ј-UTRs homologous to the bovine 1A or 1B or both have been isolated for the GHR mRNA (49). In human, eight GHR mRNA 5Ј-UTRs (namely V1 to V8) were reported (23). Among them, the DNA sequences of V1 and V2 are about 80% identical to the bovine 1A and 1B, respectively. Comparison of the bovine 1C with the human V3 to V8 only revealed a 30 -50% similarity, suggesting that the human homolog of the bovine 1C was not included among the remaining six human 5Ј-UTRs. In the bovine GHR gene, exon 1C was located 700 bp from exon 1B (Fig. 5). The DNA sequence of the bovine GHR exon 1B and its proximal 5Ј-flanking region was more than 90% identical to the corresponding region in the human GHR gene (GenBank accession AJ002175) and also to the corresponding region in the ovine GHR gene (13). Thus, the human and ovine homologs of the bovine exon 1C might exist in the human and ovine GHR mRNA. Five 5Ј-UTRs (V1 to V5; nomenclature for the rat and for the human 5Ј-UTRs are not equivalent) were also reported for the rat GHR mRNA (24). The rat 5Ј-UTR V2 and V1 are similar to the bovine 5Ј-UTR 1A (therefore ovine 1A, human V1) and the bovine 5Ј-UTR 1B (ovine 1B, human V2) in both DNA sequence (approximately 70% identity) and expression pattern (liver-specific and ubiquitous, respectively). Our alignment of cDNA sequences suggests that the rat V4 might be the homolog of bovine 1C because rat V4 and bovine 1C share a region with 75% identity (Fig. 2B). Furthermore, the rat V4 and the bovine 1C 5Ј-UTRs are expressed in similar locations, i.e. liver and many nonhepatic tissues (24), suggesting the ubiquitous activity of the P3 promoter in the rat and perhaps in other species. However, compared with 30 -40% of representation of 1C-derived GHR transcripts in the total GHR mRNA within non-hepatic tissues, V4 is the predominant form of GHR mRNA transcripts in rats (24). The relative activity of P3 and P2 promoters, therefore, may be different for different species. Although three GHR promoters (P1, P2, and P3) appear to exist in various species, isolation of other 5Ј-UTRs for the human GHR mRNA (23), the rat GHR mRNA (24), and the bovine GHR mRNA 2 suggests that the GHR promoters may not be limited to P1, P2, and P3. Other promoters remain to be identified and characterized for the GHR gene.