Agonist Regulation of Human b 2 -Adrenergic Receptor mRNA Stability Occurs via a Specific AU-rich Element*

Prolonged agonist stimulation of b 2 -adrenergic recep- tors results in receptor down-regulation, which is closely associated with a reduction of the corresponding mRNA, an effect mediated in part by changes in mRNA stability. Transfection experiments with human b 2 -ad- renergic receptor cDNAs bearing or lacking the untranslated regions suggested that the essential agonist sensitivity of the mRNA resides within the 3 * -untrans-lated region. The importance of this region was further confirmed in gel shift experiments; cytosolic prepara-tions from agonist-stimulated DDT 1 -MF2 smooth muscle cells caused a shift of b 2 -adrenergic receptor mRNAs containing the 3 * -untranslated region. Progressive 3 * terminal truncations of the receptor cDNA led to the identification of an AU-rich element at positions 329– 337 of the 3 * -untranslated region as the responsible cis acting element. Substitution of this motif by cytosine residues almost completely abolished mRNA down-reg-ulation and inhibited the formation of the RNA-protein complex. Even though the b 2 -adrenergic receptor AU- rich element showed two U 3 A transitions compared with the recently proposed AU-rich element consensus sequence, it revealed an almost identical destabilizing potency. Fusion of the b 2 -adrenergic receptor 3 * -un-translated region to the b -globin coding sequence dramatically reduced the half-life of the chimeric transcript in an agonist- and cAMP-dependent manner. This suggests that the agonist-induced b 2 -adrenergic receptor mRNA destabilization is regulated by cAMP-dependent RNA-binding protein(s) via a specific AU-rich element. Chronic stimulation of the b 2 -adrenergic

Chronic stimulation of the ␤ 2 -adrenergic receptor (␤ 2 AR) 1 results in a decrease of receptor responsiveness, a process called agonist-induced receptor desensitization (1,2). Long term desensitization often involves a significant reduction of receptor numbers, which is termed receptor down-regulation. Several distinct molecular mechanisms affecting both mRNA and protein levels contribute to receptor down-regulation (2)(3)(4), which appear to be operative to varying extents in different cell lines. To date there is evidence that the expression of the ␤ 2 AR gene can be regulated at the level of transcription (5,6), posttranscriptionally at the level of mRNA stability (7) or at the level of translation via a short peptide encoded within the 5Ј-untranslated region (UTR) of the ␤ 2 AR gene (8).
Posttranscriptional mechanisms are of particular interest, since they participate in the stability and turnover of various highly labile mRNAs, such as granulocyte-macrophage colonystimulating factor, interleukin-3, and the oncogenes c-fos and c-myc (9,10). AU-rich elements (AREs) are often found in the 3Ј-UTRs of these mRNAs and appear to be key determinants of their short half-lives, even if mRNA turnover does not strictly depend on these motifs. The optimal destabilization motif was recently suggested to be UUAUUUA(U/A)(U/A) (11,12), but there is also evidence that an AUUUA pentamer need not be an integral part of a functional ARE (13). On the contrary, it appears that each ARE represents a combination of structurally distinct domains, such as AUUUA motifs, AU nonamers, and U-rich elements, and that it is the combination of these sequence elements that determines its ultimate destabilizing function (14). AREs appear to represent recognition sites for several cytoplasmic and nucleus-associated RNA-binding proteins, which mediate RNA degradation (15)(16)(17)(18)(19). Some of these proteins have been purified, but their precise roles in the regulation of mRNA stability remain unclear.
For the ␤ 2 AR mRNA, three binding proteins have been described so far: (i) the ␤-adrenergic receptor mRNA-binding protein (␤ARB), a M r 35,000 cytosolic protein identified in hamster DDT 1 -MF2 smooth muscle cells (20); (ii) a M r 85,000 factor mediating ␤ 2 AR transcript destabilization in adult rat hepatocytes (21); and (iii) the M r 37,000 AU-rich element RNAbinding/degradation factor (AUF1), which has been shown to bind also ␤ 1 AR mRNA (22). Although AU-rich sequence motifs within the ␤ 2 AR 3Ј-UTR have been demonstrated to function as recognition sequences for these proteins in vitro (21)(22)(23)(24), the exact nature of the particular cis-acting elements mediating ␤ 2 AR mRNA destabilization in vivo has not been established. During the preparation of this manuscript, Tholanikunnel and Malbon (25) reported the first characterization of such an element, a 20-nucleotide AU-rich domain with an unusual AUUUUA hexamer core, which is obligate for the destabilization of the hamster ␤ 2 AR mRNA. However, a sequence alignment revealed no equivalent within the human ␤ 2 AR transcript (26,27), which in turn suggests that ␤ 2 AR mRNA stability is regulated via species-specific cis-acting elements.
In this study, we report the identification of a nonconsensus AU-rich nonamer within the ␤ 2 AR 3Ј-UTR as a critical determinant for the agonist-induced destabilization of the human receptor transcript and provide evidence that the participation of a RNA-binding protein and of cAMP are required for ␤ 2 AR mRNA down-regulation in vivo.

MATERIALS AND METHODS
Plasmid Construction-The ␤ 2 AR vectors used for transient transfections were constructed based on the plasmid pBC12BI-␤ 2 (26), from which a 1.95-kb fragment corresponding to the complete human ␤ 2 AR transcript was excised and inserted into the expression vector pcDNA3 * This work was supported by Deutsche Forschungsgemeinschaft Grant SFB355 and a grant from the Fonds der Chemischen Industrie. 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 vectors used as templates for in vitro transcription reactions were constructed by amplification of two ϳ1-kilobase pair fragments, each comprising one-half of the receptor transcript. These two fragments, ␤ 2 AR5Ј and ␤ 2 AR3Ј, were cloned into pGEM9Zf(Ϫ) (Promega) under the control of the T7 promoter. A 130-bp poly(A) stretch was inserted downstream of the ␤ 2 AR cDNAs to obtain polyadenylated mRNAs.
For the generation of the ␤-globin/␤ 2 AR 3Ј-UTR fusion plasmid a 2.2-kilobase pair fragment corresponding to the ␤-globin primary transcript was excised from pGEM1␤-globin and inserted into pcDNA3. After that, a 540-bp DraI-XbaI fragment comprising the ␤-globin 3Ј-UTR was replaced by the 546-bp ␤ 2 AR 3Ј-UTR excised from pcDNA3-␤ 2 ⌬5ЈUTR. The correctness of all constructs was confirmed by double-stranded DNA sequencing.
Cell Culture and Transfection-Hamster DDT 1 -MF2 smooth muscle cells and human embryonic kidney cells (HEK 293) were grown in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal calf serum, 2 mM glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin (all purchased from Life Technologies, Inc.). Monolayer cultures were harvested at 60 -70% confluence, and DDT 1 -MF2 suspension cultures were maintained at a cell density of about 5 ϫ 10 5 cells/ml. The calcium phosphate precipitation method (28) was used for transfection of HEK293 cells. Transfection efficiencies were determined by cotransfection of a ␤-galactosidase encoding plasmid, pSV␤gal.
48 h after transfection the cells were stimulated with 10 M (Ϫ)isoproterenol (Sigma) or 10 M forskolin (Sigma) for the times indicated. To block transcription and/or translation, 5 g/ml actinomycin D (Roth) and/or 0.5 g/ml Pseudomonas exotoxin A (Sigma), respectively, were added 30 min after the begin of agonist stimulation. At the times indicated, the cells were harvested, washed twice with phosphate-buffered saline, and subjected to RNA analysis.
RNA Isolation and Northern Analysis-Total RNA was isolated by the AGPC extraction method (29), separated on formaldehyde gels, and subsequently transferred to nylon membranes (Qiagen) by downward alkaline capillary transfer (30). Single-stranded DNA probes were prepared in two steps. First, the respective region was amplified in a "standard" PCR. The resulting double-stranded DNA fragment served as a template in a second, asymmetric PCR that included only the 3Ј-primer. 20 M DIG-11-dUTP (Boehringer Mannheim) was added for labeling along with 30 M dNTPs. The primers used for the preparation of the ␤ 2 AR probe were ␤ 2 AR.seq.2 and ␤ 2 AR.rev.4, spanning a 672-bp fragment immediately downstream of the start codon. Probes specific for ␣ B -crystallin (468 bp, used as an internal standard) and ␤-globin (319 bp) were amplified using the primers cry.seq/cry.rev and h␤g.seq/ h␤g.rev, respectively. Hybridization was done at 37°C for 24 -48 h in 50% formamide, 5 ϫ SSC, 3 ϫ Denhardt's solution, 0.5% SDS, 0.2% sodium laurylsarcosinate, and 5% dextran sulfate. Chemiluminescent detection was performed using the DIG Luminescent Detection kit (Boehringer Mannheim). The signal intensity on the x-ray films was analyzed densitometrically.
In Vitro Transcription-Transcripts were generated from 1 g of linearized template DNA in a total reaction volume of 20 l in the presence of 1 unit/l RNase inhibitor, 1 mM ATP/CTP/GTP, 0.65 mM UTP, 0.35 mM DIG-UTP, and 2 units/l T7-RNA-polymerase (all reagents purchased from Boehringer Mannheim). Cotranscriptional capping was performed by using the cap analogue m 7 (5Ј)Gppp(5Ј)G (New England Biolabs) in a concentration 10 times that of GTP. The reaction mixtures were incubated at 37°C for 2 h. RNase-free DNase I (Boehringer Mannheim) was added to remove template DNA. The labeled transcripts were extracted twice with phenol and then once with chloroform and precipitated with ethanol.

RESULTS
Mapping of the Agonist-sensitive Region within the ␤ 2 AR mRNA-Since in many cell types the ␤ 2 AR mRNA down-regulation appears to be mediated by several different processes (2-4), we decided to use a transient transfection system to identify the agonist-sensitive region(s) within the ␤ 2 AR transcript. For this purpose, HEK293 cells were chosen due to their very low level of endogenous ␤AR as well as the very high transfection efficiency compared with other cell lines. 48 h after the transfection of vectors bearing human ␤ 2 AR cDNAs corresponding to the complete receptor transcript and to mutants lacking either one or both UTRs, respectively, the cells were stimulated for 12 h with 10 M isoproterenol, and the ␤ 2 AR mRNA levels were quantified in Northern analyses (Fig. 1A). ␣ B -Crystallin, a widely expressed heat-shock protein, was used as an internal standard in all experiments. Cotransfection of a ␤-galactosidase encoding plasmid and subsequent staining of the cells revealed transfection efficiencies of about 90% in all samples (not shown), so that influences resulting from different transfection efficiencies of various constructs should be minimal. As shown in Fig. 1B, the ␤ 2 AR wild-type transcript and the 5Ј-UTR deletion mutant were down-regulated upon agonist stimulation by 60 Ϯ 6% and 68 Ϯ 8%, respectively, compared with unstimulated controls. The predominant role of elements encoded within the 3Ј-UTR for transcript destabilization was further confirmed by the respective deletion mutant which showed only a small reduction of the ␤ 2 AR mRNA level by 12 Ϯ 8%. The mRNA concentration of the transcript covering only the coding sequence remained almost unchanged. These results suggested that the agonist sensitivity of the human ␤ 2 AR   (31), we have demonstrated that the agonistinduced ␤ 2 AR mRNA down-regulation occurs in a cell typespecific manner. In DDT 1 -MF2 smooth muscle cells, it is caused predominantly at the posttranscriptional level via decreased mRNA stability. It has been shown for several highly labile mRNAs that destabilization motifs may function as recognition sequences for RNA-binding proteins (15)(16)(17)(18)(19). Therefore, we attempted to prove the existence of such a factor in DDT 1 -MF2 cells and to analyze a possible interaction with the human ␤ 2 AR transcript. DDT 1 -MF2 cells were grown in suspension cultures, and transcription was blocked by adding actinomycin D to the medium. After various incubation periods, the ␤ 2 AR mRNA concentrations were quantified in Northern analyses (Fig. 2). The ␤ 2 AR mRNA half-life of untreated control cells was determined to be about 120 min, whereas in agoniststimulated cells a ϳ50% reduction was observed, resulting in a half-life of ϳ50 min. These values are similar to those measured under similar conditions (suspension cultures) by Collins et al. (5). Translational blockade by exotoxin A increased the ␤ 2 AR mRNA half-life to about 80 min in stimulated cells, which indicates that ␤ 2 AR stimulation indeed induces the synthesis of a protein component, which accounts, at least in part, for receptor mRNA destabilization.
To identify the region(s) of the human ␤ 2 AR mRNA that might be interacting with such a protein component, we established a gel shift assay using the human ␤ 2 AR mRNA as a template. In vitro transcribed, DIG-labeled, capped, and polyadenylated RNAs corresponding to the 5Ј-half (189-nucleotide 5Ј-UTR plus 806-nucleotide coding sequence) and the 3Ј-half (441-nucleotide coding sequence plus 554-nucleotide 3Ј-UTR) of the ␤ 2 AR mRNA, respectively, were incubated with cytosolic extracts prepared either from DDT 1 -MF2 control cells or cells stimulated with isoproterenol for 12 h. The samples were separated on nondenaturing polyacrylamide gels and transferred onto nylon membranes. After chemiluminescent detection, a protein-mRNA complex was only found if cytosolic fractions of stimulated cells were mixed with the 3Ј-half of the ␤ 2 AR mRNA (Fig. 3). The addition of exotoxin A to the cells to block de novo protein synthesis inhibited the formation of this complex. These observations are a further indication that the 3Ј-UTR contains elements critical for the stability of the human ␤ 2 AR mRNA. Furthermore, they provide evidence that binding protein(s) induced in hamster DDT 1 -MF2 cells can bind to the human transcript and are apparently involved in this regulation.
Identification of an AU-rich Destabilization Motif within the ␤ 2 AR 3Ј-UTR-For a more detailed characterization of the human ␤ 2 AR 3Ј-UTR, two truncation mutants lacking the 3Јterminal 157 bp and 313 bp of the receptor cDNA, respectively, were generated, and their degree of agonist-induced ␤ 2 AR mRNA down-regulation after transient transfection into HEK293 cells was determined (Fig. 4A). The mRNA level changes of the mutant lacking 157 bp were comparable with those of the wild-type receptor, whereas deletion of 313 bp completely abolished agonist-mediated down-regulation. The respective levels of ␣ B -crystallin mRNA remained unchanged in all three cases. These results provide evidence that the region between positions 241 and 397 of the ␤ 2 AR 3Ј-UTR is critical for receptor mRNA destabilization (Fig. 4B).
AU-rich elements have been shown to be key determinants for the destabilization of several highly regulated mRNAs (9,10). Their consensus sequence has recently been proposed to be UUAUUUA(U/A)(U/A) (11,12). Therfore, we looked for elements consisting of at least nine consecutive adenosine or uridine residues within the ␤ 2 AR 3Ј-UTR. None of the four regions identified (Table II and Fig. 4B) exactly fits the proposed consensus sequence. Because of its location in the region shown to be critical for ␤ 2 AR mRNA destabilization, the motif UAAUAUAUU found at positions 329 -337 was of special interest. The sequence differed from the consensus only at positions 2 and 5. To test the importance of this element for ␤ 2 AR mRNA stability, the wild-type sequence (Fig. 5, WT) was re- placed by a stretch of nine cytosine residues (M), and the two constructs were transiently transfected into HEK293 cells as before. Upon agonist stimulation, the mutant ␤ 2 AR mRNA levels, normalized for the respective values for ␣ B -crystallin, were reduced to only 90 Ϯ 8% of control levels compared with 35 Ϯ 5% for the wild type (Fig. 5). Therefore, this element appears to be absolutely essential for the destabilization of the human ␤ 2 AR transcript. Three additional ARE point mutants (M1-3, Table III) were generated to provide further insights in the minimal sequence requirements of this motif. In mutant M1, positions 2 and 5 were changed (A 3 U) so that the resulting ARE corresponded to the suggested consensus sequence (11,12). The respective transcript showed an almost identical degree of down-regulation in HEK293 cells compared with the wild-type sequence, with a reduction to 32 Ϯ 8% of the unstimulated control (Fig. 5). The flanking adenosine residues of the consensus sequence have been shown to be critical for the destabilizing potency of an ARE (11,12). In mutant M2, the adenosine residues at positions 2, 3, and 7, respectively, were exchanged for cytosines. The ␤ 2 AR mRNA levels after agonist stimulation were only slightly reduced to 87 Ϯ 8%, which confirms the importance of these residues. Finally, in the mutant M3, the three central nucleotides were replaced by cytosines to investigate the function of the ARE core domain. A small but significant decrease in the respective ␤ 2 AR mRNA levels to 75 Ϯ 7% was observed (Fig. 5), demonstrating that at least for the ␤ 2 AR ARE the core is not as essential as proposed for the consensus sequence (11,12).
Additionally, we performed gel shift experiments to answer the question whether the protein(s) identified in cytosolic ex- To inhibit protein synthesis, 0.5 g/ml exotoxin A were added 30 min after the begin of stimulation (E). The samples were resolved on 4% nondenaturing polyacrylamide gels and blotted onto nylon membranes by capillary transfer. The signals were visualized on x-ray films using a chemiluminescent detection protocol. FIG. 4. A, mapping of the agonist-sensitive region within the ␤ 2 AR 3Ј-UTR. Two 3Ј-terminal truncation mutants of the ␤ 2 AR cDNA were generated lacking the last 157 bp (⌬157) and 313 bp (⌬313) of the 3Ј-UTR, respectively. After transfection into HEK293 cells and 12 h stimulation (S) with 10 M (Ϫ)-isoproterenol (or medium for controls (C)) mRNA levels for the ␤ 2 AR and ␣ B -crystallin were determined by Northern analyses. The wild-type receptor (WT) was used as a control. B, schematic representation of the ␤ 2 AR 3Ј-UTR. The ␤ 2 AR coding sequence (CDS) is shown in black, and the 3Ј-UTR is hatched. The positions of the two truncation mutants are indicated by triangles. The region between positions 241-397 of the ␤ 2 AR 3Ј-UTR critical for agonist-mediated mRNA destabilization (according to Fig. 4A) is shown as a rectangle, and the positions of AU-rich destabilization motifs are marked by arrows and are numbered according to Table II.  (11,12), only elements consisting of at least nine consecutive adenosine or uridine residues are considered. The positions are numbered starting at the adenosine residue following the stop codon. tracts of agonist-stimulated DDT 1 -MF2 cells reacts in a similar manner to mutations of the ARE in the ␤ 2 AR 3Ј-UTR. As shown in Fig. 6, a gel shift was observed with both the wild-type transcript and the mutant M1, indicating that the protein does not discriminate between these two sequences. No interaction was found with the mutants M and M2, supporting the idea that the flanking adenosine residues are critical for ARE function. For mutant M3, two signals were identified, one corresponding to the free mRNA template and a second with a lower intensity comigrating with the RNA-protein complex (Fig. 6). This agrees with the slight isoproterenol-induced down-regulation of the M3 mutant mRNA shown in Fig. 5. Summarizing these results, the agonist-induced ␤ 2 AR mRNA destabilization appears to be mediated, at least in part, by an RNA-binding protein recognizing a nonconsensus AU-rich motif at positions 329 -337 of the ␤ 2 AR 3Ј-UTR, probably via a specific regulatory mechanism.
Stability of a Chimeric ␤-globin/␤ 2 AR 3Ј-UTR Transcript-To test the hypothesis of a ␤ 2 AR-specific regulation of mRNA stability, we asked whether the elements within the ␤ 2 AR 3Ј-UTR are sufficient to destabilize a normally stable gene. To consider a possible participation of sequence motifs beside the ARE, the complete human ␤ 2 AR 3Ј-UTR was fused to the coding sequence of the human ␤-globin gene, and the half-life of the resulting chimeric transcript was measured in comparison with wild-type ␤-globin. To analyze whether the activation of the ␤ 2 AR has an influence on the function of the cis-acting elements within the transcript, a vector haboring the ␤ 2 AR cDNA was cotransfected together with the two ␤-globin constructs in HEK293 cells. 48 h after transfection, the cells were stimulated either with 10 M isoproterenol or with 10 M forskolin to directly activate the adenylyl cyclase. The stability of the ␤-globin wild-type transcript remained unaffected by agonist stimulation. The ␤-globin mRNA half-lives were about 13 h under all conditions, i.e. in control cells as well as in cells stimulated with either isoproterenol of forskolin ( Fig. 7 and Table IV). The exchange of the endogenous ␤-globin 3Ј-UTR against the respective region from the ␤ 2 AR dramatically reduced the stability of the chimeric mRNA. In addition, its stability became ␤AR agonist-dependent. Upon stimulation of the ␤ 2 AR with isoproterenol, the half-life of the chimeric mRNA decreased from about 4 to 2.5 h. The same regulatory pattern was observed with forskolin, which reduced the half-life of the chimeric transcript from 4.8 to 3.3 h. Therefore, the elements encoded within the 3Ј-UTR of the human ␤ 2 AR mRNA are sufficient to regulate mRNA stability in an agonist-dependent manner in a heterologous system. The finding that the degree of transcript destabilization is almost the same using either isoproterenol or forskolin further suggests that ␤ 2 AR mRNA stability is essentially regulated by cAMP. DISCUSSION The ␤ 2 AR is a prototypical member of the large family of G-protein-coupled receptors and is subject to a complex regu-lation by hormones and other signaling molecules (2-4). Distinct molecular mechanisms on both mRNA and protein levels, which may be operative to varying extents in different cell lines of the human ␤ 2 AR 3Ј-UTR The mutations were introduced in the ␤ 2 AR 3Ј-UTR by PCR as described under "Materials and Methods" using the primers given in Table I. The wild-type sequence (WT) is included for comparison. The mutant M1 corresponds to the ARE consensus sequence (11,12).  7. Fusion of the ␤-globin coding sequence with the ␤ 2 AR 3-UTR. The endogenous 3Ј-UTR of the ␤-globin gene was replaced by the respective region of the ␤ 2 AR (see "Materials and Methods"). Expression vectors encoding the cDNAs for the wild-type ␤-globin and the chimeric transcript, respectively, were transfected into HEK293 cells together with plasmid pBC-␤ 2 wt harboring the complete ␤ 2 AR cDNA. The half-lives of the ␤-globin mRNA and the chimeric transcript, respectively, were determined analogous to that of the ␤ 2 AR mRNA (Fig.  2). The results are expressed as a percentage of control and are the mean Ϯ S.E. of three independent experiments. and tissues, contribute to this regulation. Response elements specific for cAMP (CRE), glucocorticoids (GRE), and thyroid hormones (TRE) regulating ␤ 2 AR gene transcription have been identified in the ␤ 2 AR promoter region and the coding sequence (5,6,(32)(33)(34). Here, we report the identification and functional characterization of a cis-acting element within the 3Ј-UTR of the human ␤ 2 AR mRNA that is sufficient to cause destabilization of the mRNA. Many highly labile mRNAs possess AREs within their 3Ј-UTRs, although rapid mRNA turnover does not strictly depend on these motifs (9,10). In cytokine mRNAs, for example, Brown et al. (35) recently identified another class of destabilizing elements, which require at least one stem-loop (hairpin) in the secondary structure. Nevertheless, AREs are considered to be the predominant destabilization determinants.
Analyses using synthetic AU-rich sequences revealed that a nonamer with an AUUUA pentanucleotide core, UUAUUUA-(U/A)(U/A), is a key destabilization motif (11,12), although it is not a prerequisite for ARE function (13). In recent years, evidence has accumulated that it is the combination of structurally and functionally distinct AU-rich domains that determines the ultimate destabilizing function of an ARE (14). Several AU-rich sequences have also been identified in the 3Ј-UTRs of G-protein-coupled receptors (22,24), and binding of the three ␤ 2 AR mRNA-specific proteins identified so far, ␤ARB, P85, and AUF1, is selectively competed by poly(U) RNA (20 -22). Furthermore, in vitro binding of ␤ARB to the hamster ␤ 2 AR mRNA requires both an AUUUA pentamer and U-rich flanking domains (23). However, neither the hamster nor the human ␤ 2 AR mRNAs (26,27) possess any AU-rich consensus motifs within their 3Ј-UTR. Therefore, it is tempting to speculate that agonist-induced ␤ 2 AR mRNA destabilization occurs via unique cis-acting elements.
Measurements of the human ␤ 2 AR mRNA steady-state levels using mutants lacking either one or both UTRs predicted a predominant regulatory role for the 3Ј-UTR. In contrast, deletion of the 5Ј-UTR revealed a reduction of ␤ 2 AR mRNA levels similar to that of the wild-type receptor. Although the human ␤ 2 AR 3Ј-UTR is sufficient to significantly destabilize the ␤-globin mRNA when fused to the ␤-globin coding region, we cannot exclude the existence of additional determinants within the 5Ј-UTR that might contribute to mRNA stability. The importance of the ␤ 2 AR 3Ј-UTR was further comfirmed by the observation that a protein, whose synthesis was induced in DDT 1 -MF2 smooth muscle cells by the ␤ 2 AR stimulation, selectively bound to the 3Ј-half of the receptor transcript. This supports data from studies (15)(16)(17)(18)(19) in which AREs also functioned as recognition motifs for RNA-binding proteins.
A more detailed characterization of the ␤ 2 AR 3Ј-UTR identified an AU-rich nonamer, UAAUAUAUU, at positions 329 -337 as the critical element for ␤ 2 AR mRNA regulation. Its substitution by a stretch of nine cytosine residues almost completely abolished mRNA down-regulation and inhibited the interaction with the ␤ 2 AR mRNA-binding protein induced in DDT 1 -MF2 cells. Therefore, one may conclude that this motif represents a potent destabilization determinant. This motif differs from the consensus sequences at positions 2 and 5, which have both been shown to be important for the destabilizing potency of an ARE (11,12). However, mutational analysis of this specific ARE revealed a potency identical to the consensus element. This shows that a functional ARE does not have to contain an AUUUA pentamer (13). In accordance with previous reports (11,12), the flanking adenosine residues appear to constitute the most critical nucleotides for ARE function, since their substitution by cytosine residues was sufficient to abolish ␤ 2 AR transcript destabilization. Surprisingly, the replacement of the three central nucleotides still allowed mRNA down-regulation by about 25%. Furthermore, a weak interaction with the RNA-binding protein induced in DDT 1 -MF2 cells could be detected. A possible explanation for this unexpected result is that this mutant comprises a minimal sequence capable of functioning as an ARE, at least in the case of the human ␤ 2 AR mRNA, in which an intact core domain is not required. Since the region, in which the ARE is embedded, also does not resemble the U-rich sequences found in other highly labile mRNAs (9, 10), the human ␤ 2 AR mRNA stability appears to be regulated via a potent but nonconsensus ARE. This parallels a recent study (25), in which a 20-nucleotide AU-rich domain with an unusual AUUUUA hexamer core was identified as an obligate element for destabilization of the hamster ␤ 2 AR mRNA. Therefore, ␤ 2 AR mRNA stability seems to be regulated via species-specific cis-acting elements. Another possibility is that the deviations from the consensus can be compensated by other ␤ 2 AR-specific elements, such as the additional AU-rich domains within the 3Ј-UTR or secondary structure elements.
On the other hand, the binding protein(s) observed in DDT 1 -MF2 cells upon ␤ 2 AR stimulation does not discriminate between these sequence motifs. Therefore, one may assume that a rather general factor is responsible for ␤ 2 AR transcript destabilization. Two ␤ 2 AR mRNA-binding proteins, ␤ARB and AUF1, have been detected in this cell line so far; the latter one was also identified in the human myocardium (20,22). The binding affinity of AUF1 has recently been shown to correlate directly with the destabilizing potency of the respective ARE in vitro (36). The biochemical and functional relatedness of the two proteins initially led to the assumption that they might be identical, but immunochemical experiments recently suggested that they were distinct (22).
The analysis of the ␤-globin/␤ 2 AR 3Ј chimeric transcript confirmed that the regulation of ␤ 2 AR mRNA stability occurs in an agonist-dependent manner and requires the presence of an RNA-binding protein. Although the stability of the chimeric mRNA was only about one-third of the ␤-globin wild-type transcript, stimulation with either isoproterenol or forskolin caused a further decrease of the mRNA half-life by almost a factor of 2. This suggests that the coordinated interplay between ␤ 2 AR activation and the induction of specific binding protein(s) is required for efficient destabilization of the receptor transcript. The almost identical results with isoproterenol and forskolin show a predominant role for cAMP as a regulator of ␤ 2 AR mRNA stability. The biochemical mechanisms mediating this cAMP-dependent regulation remain to be elucidated.