Regulation of the mRNA-binding Protein AUF1 by Activation of the -Adrenergic Receptor Signal Transduction Pathway

Abstract In both cell culture based model systems and in the failing human heart, β-adrenergic receptors (β-AR) undergo agonist-mediated down-regulation. This decrease correlates closely with down-regulation of its mRNA, an effect regulated in part by changes in mRNA stability. Regulation of mRNA stability has been associated with mRNA-binding proteins that recognize A + U-rich elements within the 3′-untranslated regions of many mRNAs encoding proto-oncogene and cytokine mRNAs. We demonstrate here that the mRNA-binding protein, AUF1, is present in both human heart and in hamster DDT-MF2 smooth muscle cells and that its abundance is regulated by β-AR agonist stimulation. In human heart, AUF1 mRNA and protein was significantly increased in individuals with myocardial failure, a condition associated with increases in the β-adrenergic receptor agonist norepinephrine. In the same hearts, there was a significant decrease (50%) in the abundance of β-AR mRNA and protein. In DDT-MF2 cells, where agonist-mediated destabilization of β-AR mRNA was first described, exposure to β-AR agonist resulted in a significant increase in AUF1 mRNA and protein (100%). Conversely, agonist exposure significantly decreased (40%) β-adrenergic receptor mRNA abundance. Last, we demonstrate that AUF1 can be immunoprecipitated from polysome-derived proteins following UV cross-linking to the 3′-untranslated region of the human β-AR mRNA and that purified, recombinant p37 protein also binds to β-AR 3′-untranslated region mRNA.

In both cell culture based model systems and in the failing human heart, ␤-adrenergic receptors (␤-AR) undergo agonist-mediated down-regulation. This decrease correlates closely with down-regulation of its mRNA, an effect regulated in part by changes in mRNA stability. Regulation of mRNA stability has been associated with mRNA-binding proteins that recognize A ؉ U-rich elements within the 3-untranslated regions of many mRNAs encoding proto-oncogene and cytokine mRNAs. We demonstrate here that the mRNA-binding protein, AUF1, is present in both human heart and in hamster DDT 1 -MF2 smooth muscle cells and that its abundance is regulated by ␤-AR agonist stimulation. In human heart, AUF1 mRNA and protein was significantly increased in individuals with myocardial failure, a condition associated with increases in the ␤-adrenergic receptor agonist norepinephrine. In the same hearts, there was a significant decrease (ϳ50%) in the abundance of ␤ 1 -AR mRNA and protein. In DDT 1 -MF2 cells, where agonist-mediated destabilization of ␤ 2 -AR mRNA was first described, exposure to ␤-AR agonist resulted in a significant increase in AUF1 mRNA and protein (ϳ100%). Conversely, agonist exposure significantly decreased (ϳ40%) ␤ 2 -adrenergic receptor mRNA abundance. Last, we demonstrate that AUF1 can be immunoprecipitated from polysome-derived proteins following UV cross-linking to the 3-untranslated region of the human ␤ 1 -AR mRNA and that purified, recombinant p37 AUF1 protein also binds to ␤ 1 -AR 3-untranslated region mRNA.
The condition of heart failure is associated with heightened activity of the adrenergic nervous system (1), the severity of failure correlating with increases in circulating and cardiac concentrations of the catecholamine, norepinephrine (2). As a consequence of this increased "adrenergic drive" the cardiac ␤-AR 1 /G-protein/adenylyl cyclase pathway can become markedly desensitized. One major component of the desensitization is selective down-regulation of the dominant adrenergic receptor subtype within the human myocardium, the ␤ 1 -AR (1,(3)(4)(5). Recently, we (6) and others (7) have demonstrated that the observed decrease in ␤ 1 -adrenergic receptors in failing human heart is closely associated with a corresponding down-regulation of ␤ 1 AR mRNA. Therefore, it is of interest to better define potential mechanisms responsible for down-regulation of ␤-AR mRNA.
Experiments performed using hamster DDT 1 -MF2 smooth muscle cells (8,9) suggest that down-regulation of the endogenously expressed ␤ 2 -AR mRNA does not appear to be caused by a decrease in the rate of transcription; rather, it appears that agonist exposure decreases the half-life of ␤ 2 -AR mRNA from approximately 12 to 5 h (8). This regulatory mechanism has been demonstrated previously to be important for numerous mRNAs encoding proto-oncogenes, lymphokines, and cytokines (10 -27). For these gene products regulation of mRNA stability has also been associated with the interaction of the mRNA with a family of cytosolic proteins (M r 30,000 -40,000) that often bind to A ϩ U-rich elements (AREs) commonly located within the 3Ј-untranslated region (3Ј-UTR) of the mRNA (28 -34). This interaction induces mRNA degradation by mechanisms only partially understood. However, for some mRNAs including those containing AREs (35,36), the degradation of mRNA may be associated with the process of translation. The cytosolic A ϩ U-rich mRNA-binding proteins are in general considered to be distinct from other mRNA-binding proteins such as the heterogenous nuclear ribonucleoprotein particles (37,38), however, the role of heterogenous nuclear ribonucleoprotein particles A1 and C proteins as cytoplasmic factors regulating mRNA stability is currently undergoing reassessment (39, 40).
From previous studies (41)(42)(43) using cytosolic extracts produced from DDT 1 -MF2 hamster smooth muscle cells, the properties of a ␤-AR mRNA-binding protein (␤-ARB), which binds to hamster ␤ 2 -AR and human ␤ 1 -AR mRNAs, has undergone pre-liminary characterization. Binding of ␤-ARB to mRNA was determined to involve regions of the 3Ј-UTR of the hamster ␤ 2 -AR mRNA containing an ARE (41,42). In addition, agonist stimulation of the ␤-AR pathway or protein kinase A activation by a cAMP analogue resulted in significant up-regulation (3-4-fold) of ␤-ARB protein as detected by UV cross-linking. Conversely, treatment of DDT 1 -MF2 cells with dexamethasone, which up-regulates ␤ 2 -AR mRNA, down-regulated ␤-ARB by ϳ50%. Therefore, agents that regulate hamster ␤ 2 -AR mRNA stability and abundance appear to affect reciprocally the abundance of ␤-ARB protein. Among the family of G-protein-coupled receptors, the mRNAs of the hamster ␤ 2 -AR, the human ␤ 1and ␤ 2 -AR, and the thrombin receptor have all been demonstrated to interact with ␤-ARB (41)(42)(43). To date, the identity of ␤-ARB has remained unresolved. However, ␤-ARB does share characteristics in common with several described A ϩ U-rich mRNA-binding proteins (41), including AUF1 (31).
The cytoplasmic RNA-binding protein, AUF1 (A ϩ U-rich element RNA-binding/degradation factor), has recently been cloned and characterized (31). AUF1 binds to the 3Ј-UTRs of several highly regulated mRNAs including c-myc, GM-CSF, and c-fos. Furthermore, there is evidence of "cause and effect" between AUF1 and regulation of mRNA stability in that partially purified AUF1 can selectively accelerate the degradation of c-myc mRNA in an in vitro mRNA decay system (34). Based on these findings, we endeavored to determine if AUF1 was expressed in human heart and in DDT 1 -MF2 cells, and if so, if AUF1 abundance was regulated by stimulation of the ␤-AR pathway. Here we report that the mRNA encoding AUF1 protein is expressed in both human heart and DDT 1 -MF2 cells. Furthermore, exposure of DDT 1 -MF2 cells to ␤-AR agonist, or high adrenergic drive, as manifest in the failing human heart, results in up-regulation of the AUF1 gene product. In addition, we show that purified, recombinant p37 AUF1 protein binds to an ARE within the 3Ј-UTR of the human ␤ 1 -AR mRNA, and that cellular AUF1 can be immunoprecipitated from polysomederived proteins following UV cross-linking to the 3Ј-UTR of the human ␤ 1 -AR mRNA. These data link for the first time a specific mRNA-binding protein known to be associated with the regulation of mRNA stability, with the mRNA of a G-proteincoupled receptor. In addition, they demonstrate that the abundance of this mRNA-binding protein is up-regulated by adrenergic stimulation, an effect known to destabilize ␤-AR mRNA.

MATERIALS AND METHODS
Tissue Procurement-Human ventricular myocardium was obtained from two categories of adult subjects. Failing hearts were obtained from patients undergoing heart transplantation for end stage heart failure (n ϭ 20) due exclusively to idiopathic dilated cardiomyopathy. These individuals had not received intravenous ␤-AR agonists, phosphodiesterase inhibitors, or ␤-blocking drugs prior to transplantation. Nonfailing hearts were obtained from adult organ donors whose hearts were unsuitable for cardiac transplantation due to blood type or size incompatibility (n ϭ 14). Organ donors' hearts had normal left ventricular function, as determined by echocardiography. Left ventricular aliquots were removed from the heart immediately upon explantation, and either immersed in liquid nitrogen for mRNA and protein quantification or placed in ice-cold, oxygenated Tyrode's solution for preparation of material for radioligand binding assays, as described previously (3).
AUF1 mRNA Measurement-A 233-base pair fragment of p37 AUF1 cDNA (31) was cloned from human heart DNA by the use of reverse transcription-PCR. Primers utilized for this reaction spanned a segment of the human p37 AUF1 coding region cDNA sequence from nucleotides 471 to 702 (31) and incorporated restriction enzyme recognition sites at the 5Ј ends (SmaI for the forward primer and XbaI for the reverse primer). Primer sequences were: 5Ј-CCCGGGAAGCTTGG-GAAAATGTTTATAGGAGGCC-3Ј for the forward primer, and 5Ј-GATCTCTAGAGCTTTGGCCCTTTTAGGATC-3Ј for the reverse primer. The PCR product was subcloned into pBluescript II KS (Stratagene, Inc., La Jolla, CA) and sequenced using the dideoxy method (Sequenase Version 2, U. S. Biochemical Corp.). Radiolabeled antisense riboprobes were transcribed from the HindIII digested p37 AUF1 cDNA fragment using T7 DNA-dependent RNA polymerase, [␣-32 P]UTP (800 Ci/mmol, DuPont NEN), and the Maxiscript kit (Ambion, Inc., Austin, TX). Total cellular RNA from human ventricular myocardium or from DDT 1 -MF2 cells was extracted by the method of Chomczynski and Sacchi (44) using RNA Stat-60 (Tel Test, Inc., Friendswood, TX), and quantified by absorbance at A 260 . In each ribonuclease protection assay (RPA), 10 g of RNA were hybridized overnight with 10 6 cpm of radiolabeled AUF1 riboprobe and a low specific activity 18 S rRNA riboprobe (Ambion, Inc.) using the RPA II kit (Ambion Inc.). Since 18 S rRNA abundance is in excess of mRNAs, 18 S probe was produced at a low specific activity to assure molar excess of probe to target without producing a signal beyond the linear range when measured simultaneously with AUF1. The hybridization reaction was digested with RNase A and RNase T1. RNA-RNA hybrids were resolved by electrophoresis in an 8% polyacrylamide, 8 M urea gel. Protected fragments corresponding to AUF1 and 18 S rRNA signals were quantified using a PhosphorImager (Bio-Rad).
␤ 2 -AR mRNA Measurement-The abundance of hamster ␤ 2 -AR mRNA from DDT 1 -MF2 cells was measured by RPA using a specific riboprobe. All measurements were made as described above for AUF1 including normalization of the ␤ 2 -AR mRNA signal to the signal for 18 S rRNA. The riboprobe (311 nucleotides) was generated from plasmid DNA encoding the hamster ␤ 2 -AR using PCR primers corresponding to nucleotides 1201 to 1511 (45) and including restriction sites for XhoI and EcoRI. The forward primer was 5Ј-GATCCTCGAGGATTTCAG-GATTGCCTTCCA-3Ј, and the reverse primer was 5Ј-GATCGAAT-TCTAGTGTCCTGTCAGGGAGGG-3Ј. The PCR product, corresponding to the 3Ј end of the coding region and the 5Ј end of the 3Ј-UTR, was subcloned into pBluescript II KS, and the nucleotide sequence was verified by DNA sequence analysis. Antisense RNA probe was generated using T7 RNA polymerase from riboprobe construct linearized with XhoI as described above.
␤ 1 -AR mRNA Measurement-Human ␤ 1 -AR mRNA abundance from human ventricular myocardium was measured by quantitative reverse transcription-PCR as described in detail previously (6). Briefly, poly(A) ϩ -enriched RNA was extracted from samples of human ventricular myocardium using oligo(dT)-cellulose (Micro-Fast Track TM mRNA Isolation Kit Version 1.2, Invitrogen Corp., San Diego, CA). mRNA was subjected to a reverse transcriptase reaction in the presence of a fixed amount of synthetic (84mer) RNA "internal standard" such that target mRNA (␤ 1 AR) and internal standard were amplified colinearly. PCR primers were end-labeled with [␥-32 P]ATP and the absolute amounts of ␤ 1 -AR and internal standard PCR products were determined for each heart by linear modeling of at least 3 points on the linear portion of the amplification curves.
␤-AR Quantification-␤-AR density from human ventricular myocardium was determined in a crude membrane fraction as described previously (3). Briefly, the total population of ␤ 1 -plus ␤ 2 -AR was measured by the nonselective radioligand [ 125 I]iodocyanopindolol with and without the use of 1 M L-propranolol to determine total and nonspecific binding, respectively. Maximum binding (B max ) and [ 125 I]iodocyanopindolol dissociation constant (K d ) were determined by nonlinear leastsquares computer modeling of the specific binding curve. ␤ 1 -and ␤ 2 -AR subtype proportions were determined using the ␤ 1 -AR selective ligand CGP-20712A (3). Protein concentrations were determined by the Peterson modification of the method of Lowry (46).
Immunoblot Analysis of AUF1-Abundance of AUF1 polypeptides was determined in extracts of DDT 1 -MF2 cells and in human heart tissues using a polyclonal antibody described previously (31). In human K562 cells, this antibody recognizes p37 AUF1 ,p40 AUF1 an apparent splice form of p37 AUF1 , and 45-kDa protein, an immunologically related but uncharacterized protein (31). DDT 1 -MF2 cells were either untreated or treated with 10 M (Ϫ)-isoproterenol for 24 or 48 h. Cells were harvested with ice-cold phosphate-buffered saline containing 1 mM EDTA, centrifuged for 5 min at 1000 ϫ g, and resuspended in lysis buffer (20 mM Tris-HCl, pH 7.5, 0.1% Triton X-100, 5 mM ␤-mercaptoethanol, 5 g/ml aprotinin, and 5 g/ml leupeptine). Cell lysates were subjected to eight rounds of freeze/thaw (dry ice/ethanol for 1 min, 37°C water bath for 1 min). The samples were centrifuged at 16,250 ϫ g for 10 min, and the supernatant was collected. Total protein concentrations were measured using the BCA reagent kit (Pierce, Rockford, IL). Equal amounts ␤-Adrenergic Receptor-mediated Regulation of AUF1 of protein were resuspended in Laemmli loading buffer (47), boiled for 5 min, and separated by SDS-PAGE (10% resolution phase and 5% stacking phase). Western blotting was performed as described previously (48). Proteins were transferred to 0.1-m nitrocellulose membranes (Schleicher and Schuell) for 2 h at 40 V. Blots were blocked overnight in phosphate-buffered saline with 5% nonfat dry milk, washed 5 ϫ 5 min with 0.5% phosphate-buffered saline, incubated for 1 h with 1:1000 anti-AUF1 antibody (31), washed again in phosphatebuffered saline, and incubated with 1:1000 GAR (Jackson Immunoresearch Laboratories, Inc., Westgrave, PA). Signal was visualized using ECL detection using the manufacturer's protocol (Amersham) and Kodak X-Omat AR film. Signal intensity was determined using an Alpha Innotech IS-1000 Digital Imaging System (San Leandro, CA). Linear range of the protein signal for AUF1 was determined by comparison of increasing amounts of protein (1-25 g) on the immunoblot. All subsequent quantification was performed using 2.5 g of total cellular protein, a concentration at the lower end of the linear range.
For human heart tissues, approximately 100 mg of tissue frozen in liquid N 2 was placed in 200 l of lysis buffer (20 mM Tris-HCl, 0.1% Triton X-100, 5 mM ␤-mercaptoethanol, aprotinin 10 g/ml, and leupeptin 10 g/ml). The tissue was homogenized using 25 strokes of a Teflon pestle and a T-liner at 4°C. The homogenate was subjected to 8 rounds of freeze/thaw and centrifuged at 16,250 ϫ g for 15 min. The supernatant was transferred to a fresh tube and the protein concentration determined as above. To perform Western analysis, 100 g of total cellular protein was subjected to SDS-PAGE as described above and transferred to a polyvinylidine difluoride membrane (Millipore, Marlborough MA) for 2.5 h at 40 V. Subsequent immunodetection methods were as outlined above.
Sequencing of the cDNA Encoding the Human ␤ 1 -Adrenergic Receptor 3Ј-UTR-The ϳ2.4-kilobase cDNA encoding the human ␤ 1 -AR (49) was subcloned into pBluescript II KS at the EcoRI site and its orientation confirmed by DNA sequencing. Nucleotide sequence was determined from purified, double stranded plasmid DNA by the dideoxy method (Sequenase Version 2). Sequencing primers corresponding to the published sequence of the ␤ 1 -AR coding region and to the T3 primer were used initially. Internal primers were used once additional sequence had been established. TAQ-uense DNA sequencing kit (U. S. Biochemical Corp.) was used to sequence the T-rich portion of the cDNA. Each DNA strand was sequenced at least twice to insure accuracy. The cDNA sequence of the 3Ј-UTR of the human ␤ 1 -AR has been submitted to GenBank (U29690).
In Vitro Transcription of RNA for UV Cross-linking-A 919-base pair cDNA fragment corresponding to the ␤ 1 -AR 3Ј-UTR was synthesized by PCR and subcloned into pcDNA3 (Invitrogen) utilizing the XhoI and XbaI restriction endonuclease sites. The resulting vector was linearized with XbaI, and in vitro transcription was performed as described previously (41). Briefly, radiolabeled RNA was synthesized using T7 DNAdirected RNA polymerase and [␣-32 P]UTP (800 Ci/mmol, DuPont New England Nuclear) to produce uniformly labeled, 5Ј-capped RNA. After transcription, RNase-free DNase I was added to the mixture to remove template DNA. The labeled transcript was extracted with phenol/chloroform, then chloroform only, precipitated with ethanol and washed with 70% ethanol, resuspended in RNase-free water, and maintained at Ϫ80°C until use.
Preparation of Ribosomal Salt Wash-A 0.3 M KCl ribosomal salt wash (RSW) was produced from DDT 1 -MF2 cells using the method of Brewer and Ross (50).
Purification of Recombinant p37 AUF1 Protein-The coding region of p37 AUF1 resides on a 910-base pair BsmAI fragment spanning nucleotides 236 to 1146 of the cDNA (31). This fragment was blunted and inserted into the SmaI site of the pGEM7Z(ϩ) vector (Promega) to yield the pGEM7Z/P37CR plasmid. To generate the corresponding His 6 -AUF1 fusion peptide expression vector, an Asp718-HindIII fragment from pGEM/P37CR was inserted into Asp718-HindIII digested pTrcHisB (Invitrogen) resulting in pTrcHisB/P37CR. The reading frame of the His 6 -AUF1 fusion polypeptide was confirmed by both dideoxy sequencing and reactivity of the fusion polypeptide with polyclonal AUF1 antiserum.
An Escherichia coli TOP10 (Invitrogen) clone containing pTrcHisB/ P37CR was induced to express plasmid-encoded protein by culturing with 1 mM isopropyl-␤-D-thiogalactopyranoside (U. S. Biochemical). His 6 -AUF1 fusion polypeptide was purified using the Xpress System (Invitrogen) under native conditions as described by the manufacturer. Selected fractions were electrophoresed, and the protein profile assessed by Coomassie staining. Fractions 4 -11 were pooled, and human ␣-lactalbumin (Sigma) was added to a final concentration of 100 g/ml to aid in preserving the activity of the recombinant AUF1 protein during storage at Ϫ80°C. The concentration of purified recombinant AUF1 was determined by comparison with known amounts of bovine serum albumin using Coomassie-stained SDS-polyacrylamide gels and immunoblot analysis using anti-AUF1 polyclonal antiserum.
UV Cross-linking-UV cross-linking was performed as described previously (41). Briefly, an aliquot of radiolabeled RNA (1-4 ϫ 10 6 cpm) was added to a mixture containing 20 l of RSW (ϳ5 ϫ 10 6 cell equivalents/l) from DDT 1 -MF2 cells, 5 g of yeast tRNA, 4 mM dithiothreitol, 5 g heparin, and 65 units of RNasin in a total volume of 50 l. After incubation for 10 min at 22°C, samples were placed in an ice slurry and exposed to short-wave (254 nm) UV radiation for 3 min in a Stratagene (model 1800) UV Stratalinker. The cross-linked RNA was digested with RNase A (0.5 mg/ml) and RNase T1 (10 units/ml) at 37°C for 30 min. Samples were solubilized in 50 l of Laemmli loading buffer for 10 min at 70°C, and proteins were resolved by SDS-PAGE. Gels were stained with Coomassie Blue R-250 (Sigma) followed by destaining and drying, and subjected to autoradiography for 1-5 days.
Immunoprecipitation of AUF1-Polysomes and RSW from DDT 1 -MF2 cells and radiolabeled human ␤ 1 -AR 3Ј-UTR RNA were prepared as described above. Polysomes or RSW (ϳ2 ϫ 10 6 cell equivalents) were mixed with 5 ϫ 10 6 cpm of ␤ 1 -AR 3Ј-UTR RNA, UV cross-linked, and digested with RNase A and T1 as described above. Polysomes or RSW were precleared with preimmune serum and Pansorbin cells (Calbiochem, La Jolla, CA) and immunoprecipitated as described by Zhang et al. (31). The pellet was resuspended in Laemmli buffer, boiled 5 min, and proteins resolved by SDS-PAGE (10%). Gels were dried, and radiolabeled proteins were visualized by autoradiography for 2 to 5 days.

RESULTS
Human Heart-In failing ventricular myocardium, ␤ 1 -AR mRNA and receptor protein are significantly down-regulated to a similar extent (6,7). Furthermore, as discussed below, sequencing of the cDNA for the 3Ј-UTR of the human ␤ 1 -AR has revealed that there is at least one potential ARE. Based on the precedent of agonist-mediated destabilization of the hamster ␤ 2 -AR mRNA (8) and the binding of ␤-ARB to this mRNA (41), we wished to determine if the gene encoding the mRNA-binding protein AUF1 was expressed in the human heart. Secondarily, we wished to determine if AUF1 gene expression was affected by heart failure. Left ventricular myocardium was obtained from two groups: (i) individuals with idiopathic dilated cardiomyopathy (n ϭ 20) undergoing orthotopic cardiac transplantation, and (ii) organ donors whose hearts were unsuitable for cardiac transplantation (n ϭ 14). To measure AUF1 and human ␤ 1 -AR mRNA, total cellular RNA was isolated from left ventricular myocardium. As determined by RPA, the mRNA encoding AUF1 was significantly up-regulated in failing heart (190% of control, p Ͻ 0.05) compared to nonfailing donor hearts (Table I). In each case the signal for AUF1 mRNA was normalized to that of 18 S rRNA. Although 18 S rRNA is a significantly more abundant RNA that AUF1, the signal for 18 S was often less intense than that of AUF1 due to the intentionally low specific activity of the 18 S probe. Therefore, the signal strength ratio of AUF1 to that of 18 S RNA was often Ͼ1. Heart failure had no effect on 18 S rRNA expression (data not shown). In addition, immunoblot analysis of AUF1 proteins was performed on ventricular myocardium from nonfailing (n ϭ 8) and failing (n ϭ 8) human hearts. Fig. 1A is a representative immunoblot of eight of these hearts, four nonfailing and four failing. Compared to nonfailing hearts, the relative abundance of immunoreactive p45 and p40 AUF1 protein were both significantly increased in failing hearts (Fig.  1B). p37 AUF1 protein was distinctly present but variably detectable and at low abundance even when using as much as 100 g of total cellular protein per lane. Therefore, quantitative analysis was not performed on the signal for p37 AUF1 . A more detailed analysis of AUF1 protein expression in response to ␤-agonist stimulation was performed in a cell culture model system, described below.
Although the identity of p45 is currently unknown, it is obviously immunologically related to p37 AUF1 and to p40 AUF1 .

␤-Adrenergic Receptor-mediated Regulation of AUF1
Furthermore, as demonstrated below (Fig. 6), p45 recognized and bound to A ϩ U-rich mRNAs thus appearing to be similar in this regard to AUF1 proteins. Lastly, like p40 AUF1 , p45 abundance is up-regulated in individuals with heart failure. The relationship of p45 to AUF1 or to other mRNA-binding proteins beyond these shared characteristics remains to be determined.
As determined by quantitative reverse transcription-PCR, ␤ 1 -AR mRNA abundance was significantly decreased (ϳ40%) in failing as compared to nonfailing, control hearts (Table I). These results are consistent with previous findings (6). ␤-AR density and subtype proportions also were determined in the same failing and nonfailing hearts. ␤ 1 -AR density was also significantly reduced (ϳ61%) in failing compared to nonfailing hearts (Table I). By contrast, ␤ 2 -AR density was not different in failing compared to nonfailing hearts (23.2 Ϯ 2.0 versus 27.4 Ϯ 3.4 fmol/mg protein, respectively).
In summary, these data indicate that (i) AUF1 mRNA and protein are expressed in human ventricular myocardium; and (ii) in individuals with heart failure, AUF1 mRNA and protein are significantly up-regulated and, both ␤ 1 -AR mRNA and protein are down-regulated. From these data we conclude that up-regulation of AUF1 in human heart may be involved in the regulation of ␤ 1 -AR mRNA stability and thus may be at least partially responsible for the decline in ␤ 1 -AR mRNA and subsequently protein abundance in the failing heart.
DDT 1 -MF2 Cells-The use of a cell culture model system and the cloning of p37 AUF1 (31) has made it possible to explore in greater detail the role of ␤-agonist stimulation in regulating the expression of the mRNA-binding protein, AUF1. DDT 1 -MF2 cells were chosen because: (i) agonist-mediated destabilization of the endogenous hamster ␤ 2 -AR mRNA was originally described in these cells (8), and (ii) ␤-ARB, which binds to the human ␤ 1 -AR and hamster ␤ 2 -AR mRNAs was also originally described in these cells (41). When DDT 1 -MF2 cells were treated with 10 m (Ϫ)-isoproterenol for 24 h (n ϭ 2) or for 48 h (n ϭ 3), steady-state AUF1 mRNA abundance, as determined by RPA, was modestly increased to 129 Ϯ 6% (pϽ0.05, n ϭ 5, pooled data) of untreated controls. AUF1 mRNA was increased to the same extent at both 24 and 48 h. In each case, mRNA abundance was normalized to signal for 18 S rRNA. Treatment with isoproterenol had no effect on 18 S rRNA expression (data not shown). Fig. 2A is a representative immunoblot of DDT 1 -MF2 whole cell lysates using polyclonal anti-AUF1 antibody. Three distinct bands are evident: p37 auf1 , p40 auf1 , and p45, an immunologically related polypeptide (31). Fig. 2B demonstrates the presence of AUF1 polypeptides in the polysome fraction of DDT 1 -MF2 cells under both basal and isoproterenol stimulated conditions. As is evident, AUF1 proteins are preferentially localized to the polysome fraction. This finding is consistent with localization determined previously for AUF1 in K562 cells (31) and for ␤-ARB in DDT 1 -MF2 cells (41).
As determined by immunoblot analysis, p37 auf1 protein was increased to 230 Ϯ 50% of control (n ϭ 5, pooled data) in cells treated with 10 M (Ϫ)-isoproterenol for 24 (n ϭ 2) or 48 h (n ϭ 3) compared to untreated controls (Table II). The relative abundance of p37 auf1 protein was increased to a similar extent by isoproterenol treatment at both the 24-and 48-h time points. In the same preparations, the relative amounts of p40 auf1 and p45 were also determined. In cells treated with isoproterenol, p40 auf1 increased to 160 Ϯ 30% of control (n ϭ 5, pooled data) and p45 increased to 180 Ϯ 30% of control (n ϭ 5, pooled data) (Table II). At each time point, the relative abundance of p40 auf1 and p45 were roughly similar, both being of considerably greater abundance than p37 auf1 . Also of note is the finding that the relative abundance values of p37 auf1 and p40 auf1 are greater (in arbitrary units) at 48 h compared to 24 h. This may indicate an increased relative amount of AUF1 proteins as the cells continue to grow.
In the same cells, the relative abundance of the endogenous ␤ 2 -AR mRNA was measured. Treatment of DDT 1 -MF2 cells with 10 M (Ϫ)-isoproterenol for 24 (n ϭ 2) or 48 h (n ϭ 3) produced a decrease in ␤ 2 -AR mRNA to 63 Ϯ 2% (pϽ0.05, n ϭ 5) of control. The degree of down-regulation was highly similar at 24 and 48 h (63% versus 62% of control). As with AUF1, the TABLE I Expression of AUF1 mRNA and ␤ 1 -AR mRNA and protein in nonfailing and failing human left ventricular myocardium Tissue samples were obtained from patients undergoing heart transplantation for idiopathic dilated cardiomyopathy (Failing) or organ donors with normal contractile function (Nonfailing). Relative AUF1 mRNA abundance was measured by RPA and referenced to the signal for 18 S rRNA. The relative densitometric values for both AUF1 and 18 S RNAs are arbitrary and dependent on the specific activity of each probe in each assay (a ratio of "2" does not imply twice as much AUF1 as 18 S RNA). Absolute amounts of ␤ 1 -adrenergic receptor mRNA were measured by quantitative reverse transcriptase (RT)-PCR from poly(A)selected mRNA. ␤-AR density was measured in membrane preparations of human ventricular myocardium using multiple concentrations of the radioligand. [ 125 I]iodocyanopindolol, to determine total adrenergic receptor binding. The density of the ␤ 1 -and ␤ 2 -AR subtypes was determined by competitive binding using the ␤ 1 -selective antagonist CGP20712A.  1. AUF1 proteins in human left ventricular myocardium. A, a representative immunoblot of AUF1 proteins from four nonfailing and four failing human hearts. Whole tissue lysates of human ventricular myocardium were prepared and assayed for AUF1 polypeptides using a polyclonal anti-AUF1 antibody (31). In each case, equivalent amounts of total cellular protein (100 g) were analyzed for each subject. The data from these hearts and an additional eight hearts are summarized in B. B, the relative abundance of AUF1 immunoreactive proteins was investigated in nonfailing (n ϭ 8) and failing (n ϭ 8) human heart. Data are presented as X Ϯ S.E. The relative abundance of p45 and p40 AUF1 proteins are expressed in arbitrary units (A.U.) Statistical analysis of nonfailing compared to failing hearts was performed by use of a two-tailed, unpaired t test.

␤-Adrenergic Receptor-mediated Regulation of AUF1
protected signal for the hamster ␤ 2 -AR mRNA was normalized to the invariant signal for 18 S rRNA (data not shown). The degree of down-regulation of ␤ 2 -AR mRNA was in close agreement with a previous investigation (8) and correlates well with the degree of receptor down-regulation.
In summary, these results demonstrate that in DDT 1 -MF2 hamster smooth muscle cells: (i) stimulation of the ␤-AR pathway produces an increase in AUF1 mRNA and p37 auf1 protein.
In addition, the abundance of p40 auf1 and p45 proteins are also increased; (ii) there is a reciprocal decrease in ␤ 2 -AR mRNA abundance; and (iii) p37 auf1 protein is expressed and localized to a polysome fraction. We conclude that agonist-mediated up-regulation of AUF1 protein in DDT 1 -MF2 cells may contribute to agonist-mediated destabilization and down-regulation of the hamster ␤ 2 -AR observed in these cells.
Human ␤ 1 -AR 3Ј-UTR-Although previously cloned (49), the nucleotide sequence of the 3Ј-UTR of cDNA for the human ␤ 1 -AR had not been determined. In order to determine if the ␤ 1 -AR 3Ј-UTR contained potential mRNA stability regulatory domains such as an ARE and to facilitate mapping of mRNA-binding proteins, we sequenced this portion of the cDNA (Fig. 3). The ␤ 1 -AR 3Ј-UTR contains a uniquely long poly(U) tract in its proximal region. This domain is similar to other AREs (20,51). Several other A ϩ U-rich regions are denoted including a putative mRNA destabilizing sequence "UUAUUUAU" (52,53). In addition, four potential poly(A) addition sites (AAUAAA or AUUAAA) are present. It is currently unknown which site or sites are used for poly(A) addition.
UV Cross-linking of Polyribosome-associated Proteins and Recombinant p37 AUF1 Polypeptide to the ␤ 1 -AR mRNA-To determine which mRNA-binding proteins bind to the human ␤ 1 -AR mRNA, we performed UV cross-linking of radiolabeled RNA substrates to ribosome-associated proteins from isoproterenol (10 M for 48 h) stimulated DDT 1 -MF2 cells. Proteins were solubilized from polyribosomes by extraction with 0.3 M KCl, termed a RSW. The rationale for using RSW rather than S100 cytosol or polysomes as a starting point was that this preparation has been shown to contain AUF1 in a partially purified form as well as being sufficient to reproduce decay of proto-oncogene mRNA in an in vitro mRNA decay system (18). RNAs encoding the human ␤ 1 -AR coding region only, the ␤ 1 -AR 3Ј-UTR only, or the c-myc 3Ј-UTR, were in vitro transcribed and the radiolabeled RNAs incubated with RSW. Mixtures were UV irradiated, treated with RNase A ϩ T1, and separated by SDS-PAGE. Prominent bands are apparent at ϳM r of 65,000, 55,000, and 38,000. The M r 65,000 band binds only to the 3Ј-UTR of c-myc, whereas the M r ϳ55,000 bands and the M r 38,000 band both bind to the 3Ј-UTRs of the ␤ 1 -AR and c-myc. By contrast, none of these bands bind to the coding region of the human ␤ 1 -AR mRNA (Fig. 4). In addition to the distinct band at M r 38,000, there is significant trailing of protein binding between M r 38,000 and 41,000 indicating that perhaps more than one protein in this size range binds to the 3Ј-UTR of the human ␤ 1 -AR mRNA. This is by contrast to the c-myc 3Ј-UTR where only the M r 38,000 band is in evidence. There are also bands of low intensity between M r 41,000 and 45,000 that bind to the ␤ 1 -AR and c-myc 3Ј-UTRs.
Binding of all RSW proteins to the ␤ 1 -AR 3Ј-UTR is effectively competed by a 10-fold molar excess of unlabeled ␤ 1 -AR 3Ј-UTR (Fig. 5). By contrast, only binding of M r 38,000 protein(s) is competed by a 10-fold molar excess of the A ϩ U-rich GM-CSF 3Ј-UTR RNA but not by a 50-fold molar excess of ⌬GM-CSF. The ⌬GM-CSF RNA contains only one of the five pentameric AUUUA motifs present in the wild-type GM-CSF RNA (30). By these criteria, and as demonstrated previously for the hamster ␤ 2 -AR (41), M r 38,000 (␤-ARB) has the properties of an A ϩ U-rich mRNA-binding protein.
To test the hypothesis that the M r 38,000 protein (␤-ARB) may be an AUF1-related protein, RSW proteins from isoproterenol (10 M for 48 h) stimulated DDT 1 -MF2 cells were UV cross-linked to radiolabeled ␤ 1 -AR 3Ј-UTR, as described under "Materials and Methods." The reaction was immunoprecipitated using a polyclonal anti-AUF1 or with non-immune se-FIG. 2. Immunoblot of AUF1 proteins in DDT 1 -MF2 cells. A, whole cell lysates of DDT 1 -MF2 cells smooth muscle cells were prepared and assayed for AUF1 polypeptides using a polyclonal anti-AUF1 antibody (31). Three bands are in evidence, p37 auf1 , p40 auf1 , and p45, a polypeptide immunologically related to AUF1. B, post-nuclear S130 cytosol and polysome fractions from DDT 1 -MF2 cells were analyzed for the presence of AUF1 polypeptides.

␤-Adrenergic Receptor-mediated Regulation of AUF1
rum. Anti-AUF1 serum selectively immunoprecipitated two proteins (Fig. 6A), a single major band of ϳM r 45,000 and a band of weaker intensity immediately below the major band. This finding is in exact concordance with that of Zhang et al. (31) when immunoprecipitating AUF1 proteins UV crosslinked to the c-myc ARE. No proteins were evident when immunoprecipitating with non-immune serum.
In polysomes extracts with KCl, as in the RSW preparations used above, proteins at ϳM r 40,000 to 45,000 are not readily apparent in UV cross-linking experiments. By contrast, polypeptides at M r 40,000 to 45,000 are readily detected by UV cross-linking to the 3Ј-UTR of the human ␤ 1 -AR when using polysomes prior to KCl extraction. We therefore performed similar UV cross-linking immunoprecipitation experiments using a polysome preparation. Here, three proteins of ϳM r 40,000 to 45,000 are immunoprecipitated corresponding to bands of similar migration in the input lane (i.e. not subjected to immunoprecipitation, labeled "X-link"). By contrast, no polypeptides co-migrating with the M r 38,000 protein (␤-ARB) were immunoprecipitated (Fig. 6B). Therefore, all three proteins recognized by the anti-AUF1 antibody associate with the 3Ј-UTR of the human ␤ 1 -AR mRNA and can be immunoprecipitated after UV cross-linking.
An additional experiment was performed to test whether or not ␤-ARB and AUF1 proteins are the same. Polysome-derived proteins were UV cross-linked to the radiolabeled mRNA corresponding to the ␤ 1 -AR 3Ј-UTR. Following separation by SDS-PAGE, the gel was transferred to a polyvinylidine difluoride (Millipore) membrane and autoradiography and immunoblotting were performed to ensure that signals could be superimposed precisely. Here again, the proteins recognized by the anti-AUF1 antibody do not co-migrate with the M r 38,000 signal for ␤-ARB (Fig. 6C). By immunologic and migratory criteria, the results presented above in Fig. 6, A-C, all argue strongly against p38␤-ARB being an AUF1 protein.
Together, the UV cross-linking and immunoprecipitation experiments indicate that: (i) a number of polysome-derived proteins between ϳM r 38,000 and 45,000 bind to the 3Ј-UTR of the  6 and 7). ␤ 1 -AR and GM-CSF but not ⌬GM-CSF competed effectively for ␤-ARB binding.

␤-Adrenergic Receptor-mediated Regulation of AUF1
human ␤ 1 -AR including a prominent polypeptide at ϳM r 38,000 as well as several polypeptides between M r 40,000 and 45,000 UV cross-link to the 3Ј-UTR but not the coding region of the human ␤ 1 -AR mRNA (a protein of the same apparent molecular weight binds to c-myc and GM-CSF mRNA; data not shown); (ii) anti-AUF1 antibody immunoprecipitates proteins between ϳM r 40,000 to 45,000 when cross-linked to the human ␤ 1 -AR 3Ј-UTR, and (iii) purified recombinant p37 AUF1 binds to the 3Ј-UTR of the human ␤ 1 -AR mRNA. These results indicate that M r 38,000 (␤-ARB) is not an AUF1-related protein. The results also demonstrate by multiple methods that AUF1 proteins bind to the mRNA encoding the ␤ 1 -AR 3Ј-UTR. DISCUSSION Agonist-mediated down-regulation of G-protein-coupled receptors is a well established regulatory paradigm. One of the ways in which the amount of receptor protein may be downregulated is by an alteration in the steady-state abundance of its mRNA which, in turn, is controlled by transcription rate, or by mRNA stability, or both. Since the first report of agonistmediated destabilization of an mRNA encoding a G-proteincoupled receptor by Hadcock et al. (8), the mRNA abundance of a number of other G-protein-coupled receptors have been reported to be regulated by changes in mRNA half-life (Table III). However, it should not be assumed that for each receptor of a particular subtype that it is regulated at the level of mRNA stability in all species or cell types. For example, the ␣ 1b -AR mRNA derived from rabbit aorta smooth muscle cells is regulated by an agonist-mediated response dependent on protein kinase C activity (54). By contrast, in DDT 1 -MF2 smooth muscle cells, the stability of hamster ␣ 1b -AR mRNA appears not to be regulated by agonist exposure (55). Furthermore, the hamster ␣ 1b -AR mRNA from DDT 1 -MF2 cells is not A ϩ U-rich and does not interact with ␤-ARB (41). It is unknown whether or not the rabbit ␣ 1b -AR interacts with A ϩ U-rich mRNA-binding proteins. As another example, the half-life of the mRNA encoding the m1-muscarinic receptor is decreased by its cognate agonist, carbachol, an effect that necessitates an intact 3Ј-UTR (56). While the 3Ј-UTR of the m1-muscarinic receptor does not contain A ϩ U-rich regions, there are sequence motifs that may form hairpin loops that might act as binding sites for RNAbinding proteins. For example, 5Ј-AUU-3Ј/5Ј-UAA-3Ј motifs appear to be important for directing endonucleolytic cleavage in mRNAs containing stem-loop structures (62). These structures have been shown to interact with mRNA-binding proteins of ϳM r 34,000 (63).
Several other mRNAs important to G-protein-coupled receptor signal transduction are also regulated at the level of mRNA FIG. 6. A, immunoprecipitation and Western blot analysis of AUF1 polypeptides from UV cross-linking reactions. KCl-extracted polysomeassociated proteins (RSW) from DDT 1 -MF2 cells treated with isoproterenol (10 M for 48 h) were pre-cleared with preimmune serum. RSW (2 ϫ 10 6 cell equivalents) was UV cross-linked to 5 ϫ 10 6 cpm of capped, uniformly labeled in vitro transcribed RNA corresponding to the human ␤ 1 -AR 3Ј-UTR. Following cross-linking, the reactions were treated with RNase A ϩ T1, diluted with NET-gel buffer (31), and AUF1 proteins subjected to immunoprecipitation using polyclonal anti-AUF1 antiserum or nonimmune serum. Proteins were resolved by SDS-PAGE and detected by autoradiography. B, polysome-associated proteins (not subjected to KCl-extraction) were subjected to immunoprecipitation as described in A. In addition, an input lane, labeled "X-link," represents 10% of the cross-linking reaction not subjected to immunoprecipitation. C, polysome-associated proteins from isoproterenol-treated (10 M for 24 h) DDT 1 -MF2 cells were UV cross-linked to radiolabeled mRNA encoding the 3Ј-UTR of the human ␤ 1 -AR. The reactions were digested with RNase A ϩ T1, resuspended in Laemmli buffer, and subjected to SDS-PAGE. The gel was transferred to a polyvinylidine difluoride membrane and proteins detected as described under "Materials and Methods" (right panel). Proteins recognized by the anti-AUF1 antibody are delineated by a bracket. The same membrane was also analyzed by autoradiography (left panel). A band corresponding to p38 ␤ARB is specifically labeled. Molecular weights are as indicated.