A 20-Nucleotide (A + U)-rich Element of β2-Adrenergic Receptor (β2AR) mRNA Mediates Binding to β2AR-binding Protein and Is Obligate for Agonist-induced Destabilization of Receptor mRNA

The Mr 35,000 β-adrenergic receptor mRNA-binding protein, termed βARB protein, is induced by β-adrenergic agonists and binds to β2-receptor mRNAs that display agonist-induced destabilization. A cognate sequence in the mRNA was identified previously that provides for βARB protein binding in vitro. In the present work, the sequence established in vitro for binding of βARB protein to hamster β2-adrenergic receptor mRNA was probed in vivo by site-directed mutagenesis of the 3′-untranslated region and expression in Chinese hamster ovary cells. A 20-nucleotide, (A + U)-rich region in the 3′-untranslated region consisting of an AUUUUA hexamer flanked by defined U-rich regions constitutes the binding domain for βARB protein. U to G substitution in the hexamer region attenuates the binding of βARB protein, whereas U to G substitution of hexamer and flanking U-rich domains abolishes binding of βARB protein and stabilizes β2-adrenergic receptor mRNA levels in transfectant clones challenged with either isoproterenol or cyclic AMP. These results demonstrate that binding of βARB protein to the 20-nucleotide, (A + U)-rich domain mediates the agonist and cyclic AMP-induced mRNA decay of G protein-linked receptors, such as the β2-adrenergic receptor.

Agonist-induced down-regulation of G protein-linked receptors, such as the ␤ 2 -adrenergic receptor, provides an explanation for long term adaptation to chronic stimuli characteristically observed for members of this receptor family (1)(2)(3)(4)(5). For ␤ 2 -adrenergic receptors, steady-state levels of the receptor and its mRNA decline following a challenge with agonist (2,3). The basis for the decline in receptor mRNA induced by agonist is not transcriptional suppression, but rather post-transcriptional destabilization of receptor mRNAs (6). Recently, we identified a M r 35,000 protein with properties consistent with those expected for an RNA-binding protein selective for mRNAs of receptors that display agonist-induced down-regulation of their messages and protein expression. This M r 35,000 ␤-adrenergic receptor mRNA-binding protein, termed ␤ARB 1 protein (7), demonstrates several RNA binding properties (8) similar to (A ϩ U)-rich element (ARE)-binding proteins reported by others (9 -12).
Several classes of RNA-binding proteins have been implicated in regulating mRNA stability and turnover. The heterogeneous, nuclear ribonucleoprotein particles participate in several steps of mRNA maturation including packaging, translocation, and splicing of heterogenous nuclear RNA (26 -28). Splicing and further processing of pre-mRNAs possessing introns, a 5Ј-cap and 3Ј-poly(A) ϩ tract involves the small nuclear RNA-binding proteins (29,30). Cytosolic mRNA-binding proteins include the M r 72,000 poly(A)-binding protein, which binds to long stretches (ϳ25 nucleotides/protein) of poly(A) ϩ and stabilizes the RNA to 3Ј-to 5Ј-nuclease activity (31). A subset of smaller (ranging from M r 30,000 to 40,000), cytosolic mRNA-binding proteins have been identified that display recognition of AU-rich domains in the 3Ј-UTR (7, 9 -12, 32, 33). Although several of these proteins have been purified (32,33), the precise role that these smaller RNA-binding proteins play in regulating mRNA stability and turnover has not been elucidated.
The M r 35,000 cytosolic ␤ARB RNA-binding protein displays the following properties: binds selectively to ␤ 1 -and ␤ 2 -adrenergic and thrombin receptor mRNAs, examples of G proteinlinked receptors with mRNAs displaying AU-rich domains; fails to bind both rat and human ␤ 3 mRNA (34); and is induced by agonist treatment, its levels varying inversely with the level of receptor mRNA (7). The cognate sequences of ␤ 2 -adrenergic receptor mRNA important for binding to ␤ARB protein has been established in vitro via competition studies with 3Ј-UTRs of highly regulated mRNAs and RNA variants with specific mutations in the cognate domains, as well as via radiolabeling of these 3Ј-UTRs followed by UV-catalyzed cross-linking to cytosolic preparations containing ␤ARB protein (8). The predictive value of the presence of the cognate sequence to identify receptors displaying post-transcriptional regulation was tested using thrombin receptor as a model system. The thrombin receptor, whose mRNA harbors several AU-rich sequences in its 3Ј-UTR region, was found to display agonist-induced desta-bilization and its mRNA to bind ␤ARB protein (34). In the present work the sequences established for binding of ␤ARB protein to hamster ␤ 2 -adrenergic receptor mRNA in vitro were tested in vivo following site-directed mutagenesis in tandem with stable expression of receptor mRNA with mutated 3Ј-UTRs in Chinese hamster ovary cells. Although both pentamer and hexamer core AREs bind the ␤ARB protein (34), a 20nucleotide, (A ϩ U)-rich sequence consisting of an AUUUUA hexamer flanked by U-rich regions is shown to be obligate for binding of ␤ARB protein and regulation of ␤ 2 -adrenergic receptor mRNA stability in vivo.

EXPERIMENTAL PROCEDURES
Cell Culture-DDT 1 MF2 vas deferens smooth muscle cells were cultured in Dulbecco's modified Eagle's medium supplemented with 5%, heat-inactivated fetal bovine serum (HyClone), penicillin (60 g/ml), and streptomycin (100 g/ml) as described by Scarpace et al. (35). CHO cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal bovine serum (HyClone). Cells were treated with either drugs prepared in a vehicle or with the vehicle alone, as described in each individual protocol.
Preparation of Cytosolic (S100) Extracts-Following drug treatment, cells were washed twice with phosphate-buffered saline and removed from the plate with 1.0 mM EDTA in phosphate-buffered saline. Approximately 5 ϫ 10 7 cells were collected gently by low speed (1,000 ϫ g) centrifugation, resuspended in phosphate-buffered saline, transferred to a sterile polypropylene ultracentrifuge tube, and collected again gently by centrifugation. The phosphate-buffered saline was aspirated from the cell pellet, and 5-l aliquots each of the protease inhibitors (10 mg/ml) aprotinin and leupeptin were added to the cell pellet. The cells were then subjected to ultracentrifugation (100,000 ϫ g) for 90 min at 4°C. The resulting supernatant fraction was transferred to Eppendorf tubes and maintained in an ice bath for immediate use. This cytosolic fraction is referred to as the S100 fraction throughout this work. Protein concentration was determined by method of Lowry et al. (36).
Mutagenesis and Plasmid Construction-Mutagenesis of ␤ 2 -adrenergic receptor cDNA in pSP70 was performed by overlap extension polymerase chain reaction. Briefly, mutagenic primers were constructed containing complementary sequences to ␤ 2 AR cDNA immediately 5Ј or 3Ј to the flanking 3Ј-UTR AUUUA pentamer (nucleotide 1520 -1524) designated mutant 1, or to AUUUUA hexamer nucleotide (1598 -1603) designated mutant 2 (Fig. 1). The polymerase chain reaction was performed on ␤ 2 AR cDNA template using SP6 or T7 promoter primer and one of the mutagenic primers. Products were separated by agarose gel electrophoresis, made visible by staining with ethydium bromide and UV irradiation, excised, and purified with a GeneClean-11 kit, as described by the manufacturer (Bio 101 Inc., La Jolla, CA). Amplified fragments (5-10 ng) from the forward and reverse polymerase chain reaction were mixed and subjected to a second round of amplification by polymerase chain reaction, and the products were separated and identified as above. The fragment corresponding to full-length, mutant ␤ 2 AR cDNA was excised and gel-purified. The fragment was subjected to digestion with EcoRI, and the EcoRI-digested fragment was then cloned into EcoRI sites of both pSP70 and pCMV5. After identification of the appropriate recombinants, orientation was determined by restriction digestion mapping. The mutated cDNAs were sequenced by dideoxy method to verify the sequence for the appropriate base substitution. Plasmid vector pSP70, into which wild-type and various mutants of ␤ 2 AR cDNA were inserted, was used for in vitro transcription after linearization with a restriction enzyme that cleaves the plasmid immediately 3Ј to the receptor cDNA insert.
Expression vector pCMV5 was used for stable transfection in CHO cells. Mutant 2 was used as the template for engineering both mutants 3 and 4 ( Fig. 1). Plasmids containing the 20-nucleotide AUUUUA hexamer flanked by poly(U) regions as well as those containing 20-nucleotide pentamer flanked by poly(U) regions were constructed by use of complimentary synthetic oligonucleotides flanked by restriction sequences for HindIII at the 5Ј-end and ClaI at the 3Ј-end. Complimentary oligonucleotide were annealed and cloned into pSP70. The resultant plasmids were linearized immediately 3Ј to the AU-rich region and employed as templates for in vitro transcription.
In Vitro Transcription-The wild-type (37) and mutant cDNAs for the hamster ␤ 2 -adrenergic receptor were inserted in pSP70 plasmid vector, which then was linearlized. Transcription was performed in vitro using SP6 DNA-directed RNA polymerase to produce full-length, 5Ј-capped, uniformly labeled, poly(A) ϩ mRNAs, based on the technique of Melton et al. (38). Briefly, mRNAs were transcribed in the presence of RNasin (Promega), radiolabeled ␣-[ 32 P]UTP (800 Ci/mmol, DuPont NEN), nucleotide, and buffer conditions as detailed by Promega. Cotranscriptional capping was performed by using the cap analogue m 7 (5Ј)Gppp(5Ј)G (New England Biolabs) at a concentration that was 10-fold in excess to the concentration of GTP. After the mRNA was transcribed, RNase-free DNase was added to the mixture to remove template DNA. The labeled transcript was extracted with phenol, then with chloroform, and precipitated finally with 2.5 volumes of ice-cold ethanol and 0.1 volume of 3 M Na-acetate. The labeled transcript was reconstituted in RNase-free water, maintained at Ϫ80°C, and used within 24 h of synthesis. Size and integrity of the transcripts were verified immediately prior to use by agarose/formaldehyde gel electrophoresis.
UV Cross-linking and Label Transfer-An aliquot of radiolabeled mRNA (1-4 ϫ 10 6 cpm), 5 g of yeast tRNA, and competing unlabeled RNA transcripts (at the molar excess over probe indicated) were each added to a mixture containing the S100 cytosolic fraction (30 -100 g total protein), 4 mM dithiothreitol, 5 g of heparin, and 65 units of RNasin in a total volume of 50 l. Aliquots of the mixture of S100 cytosolic fraction and radiolabeled mRNA were distributed in wells of a 24-well microliter plate and allowed to incubate for 10 min at 22°C. Samples were aspirated, placed in an ice slurry, and then exposed to short wave (254 nm) UV irradiation at a distance of 7 cm for 30 min. The mRNA not cross-linked to protein was digested with RNase A (0.5 mg/ml) and RNase T1 (10 units/ml) at 37°C for 30 min.
SDS-Polyacrylamide Gel Electrophoresis of RNA Protein Adducts-Samples were solubilized in 50 l (1:1) of Laemmli loading buffer (39) for 10 min at 67°C. The samples were then loaded onto a SDS-polyacrylamide gel (10% acrylamide separating gel with 5% acrylamide stack) and subjected to gel electrophoresis for 110 mA/h. Resolved proteins were stained with Coomassie Blue R, and the gels destained, dried, and subjected to autoradiography for 3-7 days. The relative intensity of radiolabeled species on the gel was quantified by direct analysis of radioactivity using a ␤-phosphorimager.
Transfection of CHO Cells-CHO wild-type cells were co-transfected with vectors harboring mutant or wild-type receptor cDNAs, or empty vector plasmids, each in combination with plasmid pCW1 containing the neomycin resistance gene, using Lipofectin (Life Technologies, Inc). Stable transfectant clones were selected by neomycin resistance in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum and G418 (10 g/ml). Expression of ␤ 2 -adrenergic receptor was determined for CHO clones stably expressing receptors with wild-type and mutant 3Ј-UTRs, by ICYP binding (6).
Determination of ␤ 2 AR mRNA Half-life-CHO cells were pretreated with isoproterenol (10 M), CTP-cyclic AMP (10 M), or vehicle. After 24 h of treatment with isoproterenol or after 12 h of treatment with CPT-cyclic AMP, actinomycin D (5 g/ml) was added to arrest transcription at specific times. Total RNA was extracted from individual culture dishes at the indicated time, and the amount of receptor mRNA was established by use of an RNase protection assay performed as described previously (34). Two different radiolabeled, antisense riboprobes corresponding to 600 (740 -1338) and 285 (1201-1486) nucleotides from the coding region of ␤ 2 AR mRNA were employed for the RNase protection assay.

␤ARB Protein Binding to Full-length Wild-type and Mutant
Hamster ␤ 2 AR mRNA-Previous studies using ␤ 2 -adrenergic receptor mRNA and 3Ј-UTRs of highly regulated mRNAs and RNA variants with specific mutations in the AU-rich region demonstrated the importance of AUUUA pentamers flanked by poly(U) regions in the binding of ␤ARB protein (8). 3Ј-UTR of ␤ 2 AR mRNA contains one AUUUA pentamer, which is not flanked by poly(U) regions, and another AUUUUA hexamer, which is flanked immediately on either side by poly(U) regions. These regions were mutated with U to G substitutions ( Fig. 1) and then tested for the ability of the mRNA transcribed from these mutants to bind ␤ARB protein in vitro. The mutations were focused on the AUUUA pentamer (1524 -1528) and AUUUUA hexamer and the flanking poly(U) regions (1592-1611). U to G substitutions, in the most 5Ј-localized AUUUA pentamer of the 3Ј-UTR, yield mutant 1. U to G substitutions in the AUUUUA hexamer yield mutant 2. Transcribed in vitro, capped, uniformly labeled mRNAs and cytosolic S100 fractions were incubated and subjected to UV-catalyzed cross-linking to identify the RNA-binding proteins (28), as described earlier (7,8).
The binding of transcripts to ␤ARB protein was assessed by direct label transfer using radiolabeled mRNA of hamster wildtype and various mutant ␤ 2 ARs as well as by evaluating the ability of unlabeled wild-type and mutant mRNAs to compete with 32 P-labeled ␤ 2 AR transcripts for ␤ARB protein binding. The cross-linked, radiolabeled RNA-binding protein adducts were subjected to SDS-polyacrylamide gel electrophoresis and made visible by autoradiography (Fig. 2). The direct label transfer studies identify several classes of RNA-binding proteins capable of binding labeled transcript of wild-type (8) and of the mutant ␤ 2 ARs. ␤ARB protein (M r ϭ 35,000) was labeled prominently by the radiolabeled transcripts of wild-type ␤ 2 AR as well as those of the mutant with the U to G substitution in the most 5Ј-AUUUA pentamer (mutant 1). Several other, slower migrating proteins, less prominently labeled, were also visible. Using radiolabeled ␤ 2 AR transcripts from the mutant in which U to G substitutions occur in the AUUUUA hexamer at 1592 (mutant 2) substantially reduced the amount of label transfer to ␤ARB protein compared with that obtained with the wild-type and mutant 1 transcripts. U to G substitutions to the 5Ј-flanking poly(U) region of the AUGGUA hexamer as well as U to G substitutions to both the 5Ј-and 3Ј-flanking poly(U) regions of the mutated hexamer abolished binding of the transcripts to ␤ARB protein. Quantification by ␤-phosphorimaging of the data from replicate label transfer experiments reveals that a more than 90% label transfer to ␤ARB protein is abolished by these U to G substitutions (Fig. 2, right panel). The results from the direct label transfer studies using labeled transcripts of each mutant were tested further via competition studies in which the unlabeled transcript of mutants 2 and 3 were employed to compete with the binding of the radiolabeled wild-type ␤ 2 AR mRNA to ␤ARB protein (Fig. 3). Unlike the wild-type transcripts, the unlabeled transcripts for mutants 2 and 3 failed to compete with 32 P-labeled ␤ 2 AR mRNA for label transfer to ␤ARB protein (Fig. 3).
␤ARB Protein Binding to 20-Nucleotide ARE-Since mutation in the 20-nucleotide region abolished binding of ␤ARB protein, it was of interest to test whether this 20-nucleotide AU-rich region is the major element involved in the binding of ␤ARB protein. The AUUUA pentamer is a highly conserved sequence that is often repeated in tandem in the 3Ј-untrans-lated region of RNAs encoding short-lived cytokines and protooncogenes, and their presence appears to alter stability of some mRNAs (13). In contrast, the element identified in hamster ␤ 2 AR mRNA is an AUUUUA hexamer flanked by U-rich regions uniquely different from the known cis-acting elements identified in several, highly regulated mRNAs. To compare the ability of hexamer and pentamer AREs to binding to ␤ARB protein, we engineered plasmids expressing the 20-nucleotide ARE found in the wild-type ␤ 2 AR mRNA (see Fig. 1) and a second plasmid expressing the same 20-nucleotide ARE in which the AUUUUA hexamer is replaced by a AUUUA pentamer. Uniformly radiolabeled, capped, RNA corresponding to each of these probes was prepared together under identical Autoradiogram of UV-catalyzed cross-linking of S100 cytosolic fractions prepared from DDT 1 -MF2 cells with full-length, capped, and uniformly labeled in vitro transcribed mRNA encoding the hamster ␤ 2 -adrenergic receptor. 32 P-labeled ␤ 2 -adrenergic receptor mRNA was subjected to competition by adding increasing amounts (1-, 5-, and 10-fold molar excesses) of full-length, unlabeled RNAs transcribed from ␤ 2 -adrenergic receptors templates with either wild-type 3Ј-UTR (lanes 2-4) or mutant 2 (lanes 6 -8) and mutant 3 (lanes 10 -12) 3Ј-UTR harboring disruptions of the hexamer-containing ARE. Unlabeled RNAs and 32 P-radiolabeled RNA were added simultaneously to the mixture containing the S100 cytosolic extracts. The mixture was incubated for 10 min, prior to UV-catalyzed cross-linking.
conditions, and binding to ␤ARB protein was tested by direct label transfer (Fig. 4). The radiolabeled, 20-nucleotide hexamer ARE (*HEXAMER) displays prominent binding to the M r 35,000 ␤ARB protein. In sharp contrast, the uniformly labeled 20-nucleotide ARE in which the hexamer is replaced by a pentamer (*PENTAMER) binds ␤ARB protein less well, displaying less than 50% of the binding observed with hexamer ARE. Binding of hexamer versus pentamer was investigated further through competition studies in which the ability of the unlabeled, 20-nucleotide hexamer ARE competed with radiolabeled hexamer and pentamer AREs for binding to ␤ARB protein (Fig. 4). The unlabeled, 20-nucleotide hexamer ARE competes effectively with the binding of the labeled hexamer ARE as well as the pentamer ARE, as displayed in the autoradiogram (Fig. 4, top). Quantification of data from replicate, independent label transfer studies in which both labeled AREs were prepared and used simultaneously reveal the hexamer-containing ARE as the preferred binding site for ␤ARB (Fig. 4, bottom).
To more fully explore the binding character of ␤ARB protein with the hexamer-as compared with pentamer-containing AREs, we examined the ability of the unlabeled pentamercontaining ARE to compete with labeled pentamer-and hexamer-containing AREs (Fig. 5). The autoradiogram reveals the more prominent binding of the hexamer-containing versus pentamer-containing AREs (Fig. 5, top). Unlabeled pentamer ARE competed for binding to ␤ARB protein with labeled hexamer and pentamer AREs, confirming earlier studies in which the binding of the pentamer-containing ARE was used to elucidate the properties of RNA binding to ␤ARB protein (8, 34). Analysis of the combined results from the competition studies suggests that the hexamer-containing ARE displays greater binding to ␤ARB protein, competing more robustly than the pentamer in these studies (Figs. 4 and 5).
Wild-type and Mutant ␤ 2 AR mRNA, Steady-state Levels in Vivo-Agonist-mediated changes in receptor mRNA levels can be mimicked in culture using DDT 1 -MF2 smooth muscle cells (6) as well as cell lines stably transfected and expressing G protein-linked receptors (40,41). Wild-type CHO cells express very few endogenous ␤ 2 AR and provide an ideal cell-type for the study of agonist-induced regulation of ␤ 2 AR with essentially no background signal, once stably transfected with expression vectors (42,43). To address not only the binding character of ␤ARB protein, but more importantly the influence that mutations of the hexamer-containing ARE would have on the steady-state level of ␤ 2 AR mRNA harboring these mutated 3Ј-untranslated regions in vivo, we adopted the CHO cells and transfected the cells with vectors expressing wild-type ␤ 2 AR mRNA as well as mutants 1, 2, 3, and 4.
CHO clones stably transfected and expressing ␤ 2 AR mRNA with wild-type and mutated 3Ј-untranslated regions of interest were created. S100 fractions prepared from CHO cells display ␤ARB protein binding of uniformly labeled, capped, and polyadenylated ␤ 2 AR mRNA, as evidenced in autoradiograms of the M r 35,000 protein following direct label transfer studies (Fig.  6). The results of the direct label transfer studies performed with S100 fractions from the CHO cells agree well with those performed with S100 fractions from DDT 1 -MF2 smooth muscle cells (Fig. 2). Due to the significant expression of endogenous

FIG. 4. Competition of molar excesses of unlabeled, 20-nucleotide hexamer-containing ARE with radiolabeled, hexamer-versus pentamer-containing 3-UTR ARE for binding to ␤ARB protein: demonstration that the hexamer-containing ARE is preferred to the pentamer-containing ARE.
Upper panel, autoradiogram of UV-catalyzed cross-linking of S100 cytosolic fractions from DDT 1 -MF2 cells with radiolabeled, in vitro transcribed RNA corresponding to the 20-nucleotide ARE region containing either an AUUUUA hexamer (*HEXAMER, lanes [1][2][3][4][5] or AUUUA pentamer (*PENTAMER, lanes 6 -10) in the absence and presence of increasing amounts of (1-, 5-, 10-, and 20-fold molar excess) unlabeled RNA corresponding to ARE containing the hexamer. Bottom panel, quantification of the 32 P (counts/min) transferred. The values are displayed as "percent of label transfer," setting at 100% the transfer obtained in the absence of competing unlabeled RNA. The values are means Ϯ S.D. from three replicate experiments.

FIG. 5. Competition of molar excesses of unlabeled, 20-nucleotide pentamer-containing ARE with radiolabeled, hexamer-versus pentamer-containing 3-UTR ARE for binding to ␤ARB protein: demonstration that the hexamer-containing ARE is preferred to the pentamer-containing ARE.
Upper panel, autoradiogram of UV-catalyzed cross-linking of S100 cytosolic fractions from DDT 1 -MF2 cells with radiolabeled, in vitro transcribed RNA corresponding to the 20-nucleotide ARE region containing either an AUUUUA hexamer (*HEXAMER, lanes [1][2][3][4][5] or AUUUA pentamer (*PENTAMER, lanes 6 -10) in the absence and presence of increasing amounts of (1-, 5-, 10-, and 20-fold molar excess) unlabeled RNA corresponding to ARE containing the pentamer. Bottom panel, quantification of the 32 P (counts/min) transferred. The values are displayed as "percent of label transfer," setting at 100% the transfer obtained in the absence of competing unlabeled RNA. The values are means Ϯ S.D. from three replicate experiments. ␤ 2 AR in the DDT 1 -MF2 cells, the CHO cells with virtually no endogenous expression of ␤ 2 AR presented the only real candidate for these studies. In agreement with the results obtained with S100 from the smooth muscle cells (Fig. 2), mutation of the hexamer ARE attenuated its ability to bind to ␤ARB protein (Fig. 6, lane 3, M3). Further mutation of the hexamer ARE with 5Ј (mutant 3)-and both 5Ј-and 3Ј (mutant 4)-flanking sequences abolishes the ability of these mRNAs to bind ␤ARB protein (Fig. 6, lanes 4 and 5).
As a prelude to a determination of the effects of the mutated 3Ј-untranslated regions on the steady-state levels of ␤ 2 AR mRNA, the expression of ␤ 2 ARs and their capacity for downregulation was examined. Expression of ␤ 2 AR was determined by radioligand binding studies with the high affinity, ␤-adrenergic antagonist ligand ICYP. Stable transfectant CHO clones were found to express similar levels of ␤ 2 AR (2.0 -3.0 fmol/10 5 cells). The transfected clones expressing the wild-type receptors displayed agonist-induced down-regulation of receptor expression. Treatment with agonist (10 M isoproterenol) for 24 h resulted in a 40 -50% decline in receptor number, as measured by ICYP binding (not shown).
Agonist-induced destabilization of the ␤ 2 AR mRNA was explored in the CHO transfectant clones expressing ␤ 2 AR mRNA with the wild-type 3Ј-untranslated region. The stability of the ␤ 2 AR transcript was assessed in cells challenged with ␤-adrenergic agonist as well as with the second messenger cyclic AMP. Clones stably transfected with vector expressing wildtype ␤ 2 AR were challenged with either isoproterenol (10 M, 24 h) or CPT-cyclic AMP (50 M, 12 h), and actinomycin D was added to arrest transcription for the periods indicated. ␤-Adrenergic receptor mRNA levels were quantified by use of an RNase protection assay (Fig. 7A). In cells not stimulated with either agonist or cyclic AMP (i.e. control) the half-life for the ␤ 2 AR mRNA is greater than 10 h. The half-life for ␤ 2 AR mRNA in DDT 1 -MF2 smooth muscle cells was established to be 12-15 h (6). Treating cells with the ␤-adrenergic agonist isoproterenol or the poorly hydrolyzed, water-soluble analogue of cyclic AMP, CPT-cyclic AMP altered dramatically the stability of the ␤ 2 AR transcript (Fig. 7A). In the clone expressing the wild-type ␤ 2 AR mRNA, the half-life is reduced to ϳ 7 h by challenge with agonist or with CPT-cyclic AMP (Fig. 7A, lower panel). The half-life for the ␤ 2 AR mRNA following agonist-induced downregulation in CHO cells is similar to that obtained in the DDT 1 -MF2 smooth muscle cells in culture (5).
Identification of cis-Acting Elements Responsible for ␤ 2 AR mRNA Destabilization-The mutations in the pentamer and FIG. 6. Identification of ␤ARB protein in S100 cytosolic fractions prepared from CHO cells: UV-catalyzed cross-linking of radiolabeled hamster ␤ 2 AR mRNA. Left panel, CHO cells were treated with CPT-cyclic AMP (50 M) for 12 h to induce ␤ARB protein (7,34). Autoradiogram of UV-catalyzed cross-linking of S100 cytosolic fractions from CHO cells and full-length, capped, and uniformly labeled, in vitro transcribed mRNAs corresponding to hamster ␤ 2 AR with  7. Cells expressing ␤ 2 -adrenergic receptor mRNAs with either wildtype 3-UTR or 3-UTR with U to G substitution disrupting the AUUUA pentamer display normal agonist-induced destabilization of receptor mRNA. Autoradiograms obtained from RNase protection analysis of ␤ 2 AR mRNA isolated from CHO cells stably transfected with ␤ 2 AR harboring either wildtype 3Ј-UTR (panel A) or a 3Ј-UTR with mutation 1 interrupting the AUUUA pentamer (panel B). Cells were challenged with either no agent (control), ␤-adrenergic agonist (isoproterenol, iso), or CPT-cAMP (cAMP) and the half-life of ␤ 2 AR mRNA (lower panels of A and B) determined, as described under "Experimental Procedures." The RNase-resistant bands were quantified by phosphorimaging analysis of each band. The autoradiograms are representative of at least three replicate experiments. Quatitative data are the mean values Ϯ S.D. of three replicate experiments for each, ␤ 2 AR mRNA with either wild-type 3Ј-UTR or 3Ј-UTR with mutation 1. hexamer ARE that altered the ability of the transcripts to bind to ␤ARB protein were investigated to ascertain what influence, if any, these mutations would have upon mRNA stability and agonist-induced destabilization of receptor mRNA. As shown in Fig. 6, disruption of the 3Ј-untranslated region pentamer with U to G substitutions (Fig. 1) had no significant influence in the stability of the ␤ 2 AR mRNA. In CHO cells expressing ␤ 2 AR with either wild-type 3Ј-UTR or a 3Ј-UTR with U to G substitutions in the AUUUA pentamer, ␤ 2 ARs displayed agonistinduced down-regulation, mutant 1 clones displaying a decline from 2.1 to 1.09 fmol/10 5 cells in response to agonist. The half-life of the receptor mRNA was found to be very similar for ␤ 2 AR mRNA with either wild-type 3Ј-untranslated region or those with the most 5Ј-AUUUA pentamer disrupted to AUGUA (Fig. 7B). These data clearly indicate that the integrity of the pentamer itself is not a requisite for the destabilization of the ␤ 2 AR mRNA. Furthermore, the results agree well with the demonstration by others that the presence of an isolated AUUUA sequence itself does not ensure destabilization of the mRNA (44).
U to G substitutions in the hexamer of the ARE sharply reduced the ability of the ␤ 2 AR mRNA with mutated 3Ј-untranslated region to bind ␤ARB protein. Further U to G substitutions in the poly(U) regions of the flanking 5Ј and 3Ј sequences of the mutated hexamer led to complete loss of the ability of the mutated ␤ 2 AR mRNAs to bind ␤ARB protein. Mutants 2-4 were transfected into CHO cells and the clones tested for agonist-induced down-regulation of ␤ 2 AR mRNA and of ␤ 2 AR, in response to agonist and CPT-cyclic AMP treatment. U to G substitutions disrupting the AUUUUA hexamer (mutant 2) not only attenuated the binding of the transcript to ␤ARB protein, but also attenuated sharply the destabilization of the ␤ 2 AR mRNA in response to either agonist or treatment with CPT-cyclic AMP (Fig. 8A). When mutation of the AUUUUA hexamer was accompanied by U to G substitutions in the 5Ј-flanking U-rich region (mutant 3), both binding to ␤ARB protein and agonist-induced destabilization were abolished (Fig. 8B). The disruption of the 5Ј-and 3Ј-flanking regions of the mutated hexamer AUGGUA displayed the same loss of destabilization of the ␤ 2 AR mRNA in response to either agonist challenge or treatment with CPT-cyclic AMP (Fig. 8C). Agonist-induced down-regulation of ␤ 2 AR was sharply reduced by mutation 3 in the 3Ј-UTR, declining only 15%. In clones expressing ␤ 2 AR with mutation of the hexamer alone, agonistinduced down-regulation of ␤ 2 AR was variable in extent ranging from 10 to 30%. Taken together, these results demonstrate that the integrity of the hexamer ARE is obligate for agonistinduced destabilization of the ␤ 2 AR mRNA (Fig. 8, A-C, lower  panels). Disruption of the hexamer, but not the pentamer, attenuated, whereas further mutation of the U-rich regions flanking 5Ј and 3Ј to the hexamer nearly abolished binding of the transcript to ␤ARB protein and agonist-induced destabilzation of ␤ 2 AR mRNA. To confirm that the 600-nucleotide probe hybridizes with the target RNA to near completion (more than 90%) we established a standard curve using different quantities of in vitro transcribed ␤ 2 AR mRNA and a 285-nucleotide riboprobe as well as the 600-nucleotide riboprobe used throughout the earlier work ( Fig. 9A and B). Receptor mRNA was quantified equally well with either the 285-or the 600nucleotide riboprobe. Agonist-induced destabilization data and the mRNA half-life, as determined using the 285-nucleotide riboprobe in RNA extracted from CHO cells expressing ␤ 2 AR with wild-type 3Ј-UTR, agree well with data obtained with the 600-nucleotide probe (Fig. 9C). DISCUSSION Agonist-mediated down-regulation of receptor expression is commonly observed for members of the G protein-linked receptor (GPLR) superfamily. Clues to mechanisms underlying this mode of receptor down-regulation were provided when it was first observed that ␤-adrenergic agonists induce both a decline in ␤ 2 ARs and ␤ 2 AR mRNA levels in response to chronic stimulation (3). The basis for the agonist-induced decline in ␤ 2 AR mRNA levels was shown subsequently to reflect a destabilization of preexisting receptor mRNA (6). Since this first report of agonist-mediated destabilization of an mRNA encoding a GPLR by Hadcock et al. (6), other members of this superfamily of receptors have been reported to be regulated at the level of mRNA stability, including rat m1-muscarinic (41), AT1-angiotensin II (45), rat 5-HT 2A-serotonin (46), and human thrombin (34) receptors. A search for well known destabilizing elements, i.e. AUUUA pentamer flanked by poly(U) regions, identified these and other GPLRs (34).
To identify candidate proteins involved in the agonist-induced destabilization, uniformly labeled, full-length, capped and polyadenylated ␤ 2 AR mRNA in tandem with UV-catalyzed direct label transfer to cross-link putative RNA-binding proteins (8). Prominently labeled is the M r 35,000 ␤ARB protein, displaying specificity for binding to mRNA of ␤ 2 AR and thrombin receptors which display agonist-induced destabilization of mRNA, while not recognizing mRNA of ␤-adrenergic subtypes (rat and human ␤ 3 ) (34) which shows transcriptional repression without alterations in the t1 ⁄2 of the receptor mRNA (47).
In the present study the sequence of hamster ␤ 2 AR mRNA necessary for binding to ␤ARB protein was revealed to be a 20-nucleotide ARE consisting of an AUUUUA hexamer flanked on 5Ј and 3Ј reaches by U-rich regions. Several eukaryotic, mRNA-stability determinants rich in AREs have been identified in the 3Ј-untranslated regions of transiently expressed mRNAs. Extensive mutational analysis and preparation of chimeric RNAs using AU-rich sequences of various proto-oncogenes and cytokine mRNAs as well as stable mRNA such as that for ␤-globin have permitted identification of AUUUA pentamers that are present both in tandem repeats (13) and nonamers such as UUAUUUA(U/A)(U/A) (24) and UUAUU-UAUU (25), as potent destabilizing elements in these RNAs.
Regulatory, trans-acting factors identified to date, include several cytoplasmic mRNA-binding proteins, which specifically interact with the ARE (7,9,11,12,32,48,49). Based on their patterns of induction as well as their subcellular localization, these RNA-binding proteins have been shown to act as stabilizers (9,33), destabilizers (7,11,12,32), and nucleocytoplasmic transporters (50). Agonist stimulation of ␤ 2 ARs provokes a significant up-regulation of ␤ARB protein, as established by UV-catalyzed cross-linking (7). Treatment of DDT 1 -MF2 smooth muscle cells with the glucocorticoid dexamethasone up-regulates ␤ 2 AR expression and ␤ 2 AR mRNA levels, and simultaneously down-regulates the level of ␤ARB protein (7). Based on the pattern of induction and the sign of the change in ␤ 2 AR mRNA levels, the M r 35,000 ␤ARB protein appears to be a destabilizer of mRNA.
Bohjanen et al. (51) identified three different kinds of RNAbinding proteins, termed AU-A, AU-B, and AU-C in human lymphocytes. The M r 34,000 AU-A protein is constitutively expressed, interact with AUUUA multimers, and other U-rich FIG. 9. Ribonuclease protection assay of ␤ 2 AR mRNA using riboprobes of 285 versus 600 nucleotides. Panel A, autoradiogram obtained from RNase protection analysis of increasing quantities ␤ 2 AR mRNA using riboprobes of 285 and 600 nucleotides. The indicated quantities of in vitro transcribed "sense strand RNA" was normalized by addition of yeast RNA and then hybridized with a 285-and 600-nucleotide riboprobe. A linear relationship between amount of ␤ 2 AR mRNA and the intensity of the protected band was obtained. Panel B, quantification of RNase-resistant bands using phosphorimaging analysis of each band and plot of the standard curve. Panel C, CHO cells stably transfected with ␤ 2 AR harboring wild-type 3Ј-UTR were challenged with either no agent (control), ␤-adrenergic agonist (isoproterenol, iso), or CPT-cAMP (cAMP), RNA was isolated, and the half-life was determined using the 285-nucleotide riboprobe. sequences, including poly(U) sequences. AU-B and AU-C are 30,000 and M r 43,000 cytoplasmic proteins requiring at least three tandem repeats of AUUUA pentamers for efficient binding and do not bind to AUUUUA-containing AREs. ␤ARB protein, in sharp contrast, binds efficiently hexamer AREs. The binding by ␤ARB protein is reduced by 50% when the hexamer ARE of the ␤ 2 AR mRNA is replaced by a pentamer. ␤ARB protein also does not bind to poly(U) sequences (34). An additional AU-binding factor, AUF1, is a M r 37,000 protein purified and cloned from the cytoplasmic extract of human leukemia cells (32). Although AUF1 has been shown to bind to ␤ 2 AR mRNA via the 3Ј-UTR, its nonidentity with ␤ARB protein was established by immunoprecipitation and immunoblotting analyses of AUF1 polypeptides of UV cross-linking products (52). DeMaria and Brewer (53) demonstrated that AUF1-ARE binding affinity is directly related to the potency with which an ARE destabilizes a heterologous mRNA. The more prominent binding of the hexamer-containing ARE as compared with the pentamer-containing ARE to ␤ARB protein suggests the ␤ARB protein binds preferentially to the hamster ␤ 2 AR mRNA hexamer-containing ARE than to other mRNAs with ARE harboring AUUUA pentamers, such as those found in regulated mRNAs.
Although AU-rich sequences have been identified within the 3Ј-untranslated regions of many GPLR mRNAs (34,52), the current report is the first to identify a particular cis-acting element obligate for binding to a specific RNA-binding protein, in this case ␤ARB protein. Although having some similarities with other known AREs, the ARE identified in the present work is unique in several important ways. First, the core region is an AUUUUA hexamer and not the more commonly found pentamer. Second, ␤ARB protein binding is significantly greater for the hexamer rather than pentamer-containing ARE. Many ARE-binding proteins display just the opposite, i.e. much less binding affinity for hexamer-than pentamer-containing AREs as the core region (51). Finally, we demonstrate that the hexamer-ARE is obligate for agonist-induced destabilization of the ␤ 2 AR mRNA in vivo. Although preferring the hexamer-to pentamer-containing ARE of the ␤ 2 AR mRNA, ␤ARB protein fails to bind, and ␤ 2 AR mRNA fails to undergo destabilization if the integrity of the hexamer-containing ARE is lost.