Role of β-arrestin1/ERK MAP kinase pathway in regulating adenosine A₁ receptor desensitization and recovery

Abstract

Exposure of cells to adenosine receptor (AR) agonists leads to receptor uncoupling from G proteins and downregulation of the A₁AR. The receptor levels on the cell surface generally recover on withdrawal of the agonist, because of either translocation of the sequestered A₁AR back to plasma membrane or de novo synthesis of A₁AR. To examine the mechanism(s) underlying A₁AR downregulation and recovery, we treated ductus deferens tumor (DDT₁ MF-2) cells with the agonist R-phenylisopropyladenosine (R-PIA) and showed a decrease in membrane A₁AR levels by 24 h, which was associated with an unexpected 11-fold increase in A₁AR mRNA. Acute exposure of these cells to R-PIA resulted in a rapid translocation of β-arrestin1 to the plasma membrane. Knockdown of β-arrestin1 by short interfering RNA (siRNA) blocked R-PIA-mediated downregulation of the A₁AR, suppressed R-PIA-dependent ERK1/2 and activator protein-1 (AP-1) activity, and reduced the induction of A₁AR mRNA. Withdrawal of the agonist after a 24-h exposure resulted in rapid recovery of plasma membrane A₁AR. This was dependent on the de novo protein synthesis and on the activity of ERK1/2 but independent of β-arrestin1 and nuclear factor-κB. Together, these data suggest that exposure to A₁AR agonist stimulates ERK1/2 activity via β-arrestin1, which subserves receptor uncoupling and downregulation, in addition to the induction of A₁AR expression. We propose that such a pathway ensures both the termination of the agonist signal and recovery by priming the cell for rapid de novo synthesis of A₁AR once the drug is terminated.

the physiological roles of adenosine are mediated by four different adenosine receptor (AR) subtypes, A₁AR, A_2AAR, A_2BAR, and A₃AR (7), all of which belong to the family of G protein-coupled receptors. These roles include diverse functions such as suppression of neurotransmitter release in the brain, sleep, bradycardia, vasodilation, inhibition of lipolysis, and inhibition of cancer cell proliferation. These responses are mediated via different signal transduction pathways activated via different guanine nucleotide regulatory (G) proteins. The A₁AR is coupled mainly to the G_i proteins, which leads to inhibition of adenylyl cyclase activity and a decrease in cyclic AMP production (32). In atrial cells, coupling of the A₁AR to G_i activates an outwardly rectifying K⁺ channel, leading to bradycardia (17). Activation of the A₁AR can also inhibit cardiac and neuronal Ca²⁺ channels (1). More recent findings have indicated that activation of the A₁AR increases the activity of extracellular signal-regulated kinase (ERK)1/2 in a variety of cell types (36), which subserves a number of functions including cytoprotection (8).

Signaling mediated via the A₁AR could be terminated or reduced by an agonist-dependent process termed desensitization, a homeostatic response invoked by the cell to limit receptor activity. Desensitization of A₁AR has been well studied both in vivo and in vitro. In rats, desensitization of the A₁AR evoked by the agonist R-phenylisopropyladenosine (R-PIA) is characterized by a loss in A₁AR-dependent inhibition of adenylyl cyclase activity in rat adipocytes, a reduction in A₁AR protein, and a decrease in the G_i coupling proteins (29, 27). These findings have been replicated in cultured rat adipocytes (10), in cardiac myocytes (39, 19), in DDT₁ MF-2 ductus deferens tumor cells (30, 24), and in cells transfected with A₁AR (26). Desensitization of the A₁AR in vitro was associated with receptor phosphorylation (31), probably via the β-adrenergic receptor kinase (24), protein kinase A, or protein kinase C (37).

Besides the posttranslational mechanism of desensitization described above, another mechanism that could decrease the level of A₁AR during the desensitization process is the decrease in the transcripts encoding this receptor. Such a mechanism has been described for the α₂ adrenergic receptor in DDT₁ MF-2 cells (11). However, little is known concerning the regulation of the A₁AR transcripts during desensitization. Fernandez et al. (6) reported that desensitization of the A₁AR in rat brain was not associated with any change in A₁AR mRNA. Similar findings were reported in cerebellar granule cells after long-term exposure to R-PIA (42). However, these investigators reported a consistent induction of A₁AR mRNA from 3 to 12 h after initiation of R-PIA administration (42). Thus regulation of A₁AR transcript is not a likely explanation for the decrease in A₁AR protein during desensitization. Nevertheless, decreases in expression and A₁AR protein in the rat brain were observed at the end of pregnancy (18) and in postmortem brains of multiple sclerosis patients (41).

Transcriptional regulation of the A₁AR appears to involve activator protein-1 (AP-1) and steroid responsive element (SRE)-2 consensus sequences in the A₁AR promoter (34). Deletion of AP-1 binding sites suppressed promoter activity, suggesting that these sites regulate the basal expression of the A₁AR. These sites could account for glucocorticoid regulation of A₁AR expression in DDT₁ MF-2 cells (9) and in the brain (40). The A₁AR promoter A harbors one of the two κB sites for the binding of the transcription factor nuclear factor-κB (NF-κB), which is implicated in the induction of A₁AR after oxidative stress (12, 25, 29).

In this study, we show that agonist-mediated desensitization of the A₁AR was associated with a loss in A₁AR but an increase in receptor mRNA. The loss of A₁AR appears dependent on its activation of ERK1/2 through its interaction with β-arrestin1. Withdrawal of the agonist resulted in a rapid recovery of the A₁AR levels that was dependent on de novo receptor synthesis mediated via the β-arrestin1/ERK1/2 MAP kinase pathway but independent of NF-κB. Thus the β-arrestin1/ERK1/2 MAP kinase pathway could play a crucial role in regulating both the desensitization and the recovery of the A₁AR.

MATERIALS AND METHODS

Materials.

R-PIA, 8-cyclopentyl-1,3-di[2′,3′]propylxanthine (DPCPX), 2-(2-amino-3-methoxyphenyl)-4H-1-benzopyran-4-one (PD98059), pyrrolidine dithiocarbamate (PDTC), cycloheximide, N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid (HEPES), Tris · HCl, soybean trypsin inhibitor, pepstatin, benzamidine, and adenosine deaminase were obtained from Sigma-Aldrich (St Louis, MO). The ATP-competitive ERK1/2 inhibitor 5-(2-phenyl-pyrazolo[1,5-a]pyridin-3-yl)-1H-pyrazolo[3,4-c]pyridazin-3-ylamine (FR180204) was from EMD Biosciences (Gibbstown, NJ). The antagonist radioligand [³H]DPCPX (160 Ci/mmol) was purchased from Perkin Elmer (Boston, MA). Monoclonal antibody against A₁AR was obtained from Dr. Yuko Sekino (Gunma University School of Medicine, Maebashi, Japan), while polyclonal antibody against A₁AR was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). p-ERK1/2, ERK1/2, p65, p50, and IκB-α antibodies were also from Santa Cruz Biotechnology. Anti-β-actin and anti-β-arrestin1 monoclonal antibodies were from Sigma-Aldrich and EMD Biosciences, respectively. All cell culture supplies, including Dulbecco's modified Eagle's medium (DMEM, high glucose), fetal bovine serum, and penicillin-streptomycin, were obtained from Invitrogen (Carlsbad, CA). All other reagents were of the highest available grade and were purchased from standard sources.

Cell culture.

Syrian Hamster ductus deferens smooth muscle cells (DDT₁ MF-2 cells) were cultured in medium consisting of DMEM (high glucose), supplemented with 10% fetal bovine serum, 50 U/ml penicillin, and 25 μg/ml streptomycin. Cells were grown at 37°C in the presence of 5% CO₂ and 95% ambient air. Cells were passaged twice a week, and all experiments were performed with confluent monolayers.

Preparation of cell membrane and nuclear preparations.

Membrane, cytosolic, and nuclear preparations from whole cell lysates were made by using procedures described previously (15).

Radioligand binding.

Quantitation of A₁AR was performed using the A₁AR antagonist radioligand [³H]DPCPX and ¹²⁵I-labeled N⁶-(4-aminobenzyl)-9-[5-(methylcarbonyl)-β-d-ribofuranosyl]adenine (AB-MECA). Iodination of AB-MECA was performed as described previously (15). Membrane preparations (∼50 μg protein/assay tube) were incubated with the radioligands for 1 h at 37°C in the absence or presence of 10 mM theophylline (to define nonspecific binding) in a 250-μl total volume of (in mM) 50 Tris · HCl (pH 7.4), 10 MgCl₂, and 1 EDTA. Saturation binding experiments were performed by incubating at least five different concentrations of [³H]DPCPX (0.5–5 nM) with membranes for 1 h at 37°C. The reaction mixture was then filtered over polyethyleneimine-treated (0.3%) Whatman GF/B glass fiber filters (Brandel, Gaithersburg, MD) and washed with 10 ml of ice-cold Tris buffer containing 0.01% 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS). The radioactive content of each filter was determined with a Beckmann liquid scintillation counter (LS5801).

Short interfering RNA transfections.

DDT₁ MF-2 cells were transfected with scrambled short interfering RNA (siRNA) or β-arrestin1 siRNA with the RNAi Human/Mouse Starter Kit (Qiagen, Valencia, CA) according to the manufacturer's protocol. Briefly, 2 × 10⁵ or 3 × 10⁵ cells were plated in six-well plates or T-25 flasks, respectively. On the next day, 10 nM β-arrestin1 or scrambled siRNA was diluted in 100 μl of serum-free culture medium and incubated at room temperature with 3 μl of HiPerfect transfection reagent for 5–10 min to allow formation of transfection complex, which was then added to the cell culture in a dropwise fashion. After 48-h transfection, medium was changed and respective treatments were performed. The target sequence used to design the siRNA against β-arrestin1 was 5′-AAAGCCUUCUGCGCGGAGAAU-3′.

Western blotting.

Total cell lysates were used to quantitate R-PIA-mediated activation of ERK1/2, while membrane and nuclear preparations were used to look for the expression of A₁AR and β-arrestin1 and p65/p50 subunits of NF-κB, respectively. Assays were performed essentially as described previously (15). Blots were visualized by exposure to Kodak XAR film (Fisher Scientific, Pittsburgh, PA) or with the LAS-3000 imaging system (Fujifilm USA, Valhalla, NY). Quantitation of the bands was performed with Un-Scan-It software (Silk Scientific, Orem, UT).

Immunocytochemistry.

Cells were cultured on glass coverslips and fixed with 4% paraformaldehyde after 24-h R-PIA treatment. Fixed cells were processed and imaged for A₁AR as described previously (15).

Real-time reverse transcriptase-polymerase chain reaction.

Total RNA was isolated from DDT₁ MF-2 cells with Tri reagent (Sigma-Aldrich) as directed by the manufacturer. poly(A)⁺ RNA was made from this, 15 μl of poly(A)⁺ RNA was converted to cDNA with the iScript cDNA Synthesis Kit (Bio-Rad, Hercules, CA), and quantitative real-time reverse transcriptase-polymerase chain reaction (qRT-PCR) was performed as described previously (15). The primers used for A₁AR were A₁AR primer 5′-CAT CCC ACT GGC CAT CCT TAT-3′ (sense) and A₁AR primer 5′-AGG TAT CGA TCC ACA GCA ATG-3′ (antisense). A₁AR was normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH). The primer sequences included 5′-ATG GTG AAG GTC GGT GTG AAC-3′ (sense) and 5′-TGT AGT TGA GGT CAA TGA AGG-3′ (antisense).

AP-1 transcription factor assay.

Nuclear preparations made from DDT₁ MF-2 cells were used to determine AP-1 DNA binding activity with an ELISA-based TransAM AP-1 (c-Fos) transcription assay kit (Active Motif, Carlsbad, CA) according to the manufacturer's protocol. Twenty micrograms of nuclear preparation was diluted in complete lysis buffer to make the final volume of 20 μl per well. The samples were incubated for 1 h at room temperature with mild agitation on a rotating rocker platform. Wells were then washed three times each with the 1× wash buffer supplied with the kit. After the washes c-Fos antibody (diluted 1:1,000 in the 1× antibody binding buffer) was added to each of the wells and incubated at room temperature for 1 h without agitation. Subsequently, the wells were washed again three times with the 1× wash buffer, followed by 1-h incubation with the horseradish peroxidase secondary antibody diluted 1:1,000 in antibody binding buffer. After four washes with 1× wash buffer, 100 μl of the developing solution was added to each well and incubated at room temperature in the dark for 5 min. After the 5-min incubation, the reaction was stopped with a stop solution and absorbance was read on a microplate reader (Biotek Instruments, Winooski, VT) at 450 nM with a reference wavelength of 655 nM.

Protein determination.

The amount of protein in samples was determined with the Bio-Rad protein assay dye reagent concentrate (Bio-Rad), using bovine serum albumin to prepare standard curves.

Statistical analysis.

Values are presented as means ± SE, and statistical analyses were performed by using either one-way ANOVA or Student's t-test followed by Tukey's post hoc test (SPSS software, SPSS, Chicago, IL).

RESULTS

Agonist mediates A₁AR desensitization in DDT₁ MF-2 cells.

Exposure of DDT₁ MF-2 cells to the A₁AR agonist R-PIA led to desensitization of the A₁AR followed by receptor internalization, as described previously (24). Similar to that study, we demonstrate a time-dependent desensitization of A₁AR by R-PIA in plasma membrane preparations as evidenced by a decrease in antagonist radioligand binding (Fig. 1A) and uncoupling of the A₁AR from G proteins (as determined by agonist radioligand binding, indicative of receptor desensitization). Treatment with R-PIA for 8 h reduced [³H]DPCPX binding to 61 ± 5% of vehicle-treated cells (controls) and to 52 ± 8% of control by 12 h (Fig. 1A; P < 0.05, n = 3). The decrease in antagonist radioligand binding was maximal at 46 ± 8% of control by 24 h of R-PIA treatment and was maintained at similar levels up to 48 h. The binding of the agonist radioligand ¹²⁵I-AB-MECA was reduced to 61 ± 6% of control after 12-h exposure to R-PIA, indicative of receptor uncoupling. Immunolabeling of the A₁AR with a monoclonal antibody showed bright red fluorescence localized mainly to the plasma membrane in control cells, but this fluorescence was reduced in the plasma membranes of cells exposed to R-PIA for 24 h (Fig. 1B). Western blotting studies also showed reduced plasma membrane A₁AR in cells treated with R-PIA, confirming the loss of the plasma membrane receptor protein in the treated cells (Fig. 1B, inset). Withdrawal of R-PIA led to recovery of A₁AR to the normal control level by 24 h. This is supported by the radioligand binding study (Fig. 1C) and Western blotting (Fig. 1C, inset).

Fig. 1.

Desensitization of A₁ adenosine receptor (A₁AR) after agonist treatment. A: DDT₁ MF-2 cells were treated with 1 μM R-phenylisopropyladenosine (R-PIA) for different time periods, and A₁AR antagonist radioligand [³H-labeled 8-cyclopentyl-1,3-di[2′,3′]propylxanthine (DPCPX)] binding was performed on membrane preparations. *Statistically significant difference from control (P < 0.05; n = 3). B: DDT₁ MF-2 cells were grown on glass coverslips, treated with vehicle (control) or R-PIA (1 μM) for 24 h, and fixed with 4% paraformaldehyde. Immunocytochemistry for A₁AR was performed on these fixed cells with an A₁AR monoclonal antibody and tetramethylrhodamine isothiocyanate-labeled secondary antibody. Surface expression of A₁AR is seen as red fluorescence (arrows). Inset: Western blot for A₁AR showing decreased surface expression after 24-h R-PIA treatment. C: cells were treated either with vehicle (control) or R-PIA for 24 h. For R-PIA withdrawal treatment, fresh medium was added 24 h after R-PIA treatment, cells were cultured for an additional 24 h, and [³H]DPCPX radioligand binding was performed on cell membrane isolates. Values are means ± SE from 3 independent experiments performed in duplicates. Statistically significant difference: from *control (0 h) and **R-PIA treatment groups (P < 0.05; n = 3). Inset: Western blotting for A₁AR shows recovery of surface receptors after 24-h R-PIA withdrawal (WD). Lanes shown are a composite of 2 portions of the same gel, which are separated by dark vertical lines. D: saturation radioligand binding studies for cells treated with vehicle (control) or R-PIA (24 h) or after R-PIA withdrawal for 24 h (WD).

To validate that the changes in receptors measured by single concentrations of the radioligand reflected real changes in receptor number, we performed saturation curves for [³H]DPCPX. R-PIA treatment for 24 h reduced the total number (maximum binding site density, B_max) of A₁AR, but this number recovered to the control level on withdrawal of the drug (Fig. 1D). In the presence of R-PIA, B_max was 355 ± 56 fmol/mg protein, while after 24 h R-PIA withdrawal it recovered to 707 ± 96 fmol/mg protein (n = 3; P < 0.05 with R-PIA-treated cells). B_max for vehicle-treated control cells was 660 ± 86 fmol/mg protein.

Involvement of β-arrestin1 in agonist-mediated downregulation of A₁AR.

β-Arrestin1 has been shown to mediate the association of the phosphorylated receptor with clathrin and facilitate its downregulation (21). We next examined the possibility that β-arrestin1 also mediates A₁AR-dependent decrease in plasma membrane receptor. Cells were pretreated with β-arrestin1 siRNA for 48 h to decrease its expression (Fig. 2A, inset). When these cells were subsequently treated with R-PIA, the agonist-dependent reduction in A₁AR was almost completely abrogated. Cells treated with β-arrestin1 siRNA showed [³H]DPCPX binding comparable to cells treated with scrambled siRNA (Fig. 2A). [³H]DPCPX binding was 42 ± 5% in R-PIA-treated cells, while it increased to 82 ± 9% in cells transfected with β-arrestin1 siRNA followed by R-PIA treatment (P < 0.01; n = 4). The ability of R-PIA to effect downregulation of the A₁AR was not different between cells administered vehicle or scrambled siRNA (see Supplemental Fig. S1).¹ We next examined the effect of R-PIA on β-arrestin1 localization. R-PIA produced a rapid translocation of cytosolic β-arrestin1 to the membrane that was evident as early as 2 min after exposure and increased up to the 30-min time period examined (Fig. 2B).

Fig. 2.

β-Arrestin1 is involved in R-PIA-mediated A₁AR downregulation. A: DDT₁ MF-2 cells were treated with R-PIA for 24 h after 48-h transfection with scrambled (Scramble) or β-arrestin1 short interfering RNA (β-Arr1 siRNA). [³H]DPCPX binding was performed, and the results are presented as means ± SE. Statistically significant difference: from *Scramble and **R-PIA treatment groups (P < 0.01; n = 4). Inset: Western blot showing the effective knockdown of β-arrestin1 by β-Arr1 siRNA. B: Western blotting was performed for β-arrestin1 in membrane preparations isolated from cells treated with 1 μM R-PIA for different time periods. Blot is representative of 3 different experiments with similar results. Lanes presented for β-arrestin1 are a composite of 3 consecutive segments of the same gel, as indicated by dark vertical lines, which allowed for better horizontal alignment. No adjustments were needed for the β-actin lanes showing the same loading sequence.

R-PIA increases A₁AR transcripts through activation of β-arrestin1/ERK1/2 MAP kinase/AP-1 pathway.

To determine whether downregulation of the A₁AR was associated with changes in the steady-state levels of A₁AR transcripts, we performed real-time PCR on mRNA isolated from control cells and cells treated with R-PIA for 24–72 h. Surprisingly, we observed increases in A₁AR transcripts over this time period, with greatest increases observed over the 24-h time period (Fig. 3A). The increases in A₁AR transcripts were 11-, 3-, and 2-fold for R-PIA treatment periods of 24, 48, and 72 h, respectively. The increases observed at 24 h were significantly reduced in cells treated with PD98059 (10 μM) to inhibit MAP kinase kinase (MEK), the upstream activator of ERK1/2 (Fig. 3B), and by β-arrestin1 siRNA (Fig. 3C), implicating the β-arrestin1/ERK1/2 pathway in the induction of the A₁AR. Inhibition of the agonist-induced increase in A₁AR transcripts was also observed in cells pretreated with an ERK1/2 inhibitor FR180204 (Supplemental Fig. S2). Furthermore, cells incubated with R-PIA (1 μM) showed a transient activation of ERK1/2, which peaked at 5 min and returned to baseline by 20 min (Fig. 3D). The ERK1/2 activity was reduced in cells pretreated with DPCPX (1 μM) (Fig. 3E), but not by the A_2AAR antagonist ZM241385 (1 μM) (data not shown), implicating A₁AR and not A_2AAR in this activation. R-PIA-mediated ERK1/2 activation was also abrogated in cells pretreated with PD98059 (Fig. 3F) and FR180204 (Supplemental Fig. S3). Cells in which the levels of β-arrestin1 were reduced by siRNA failed to show R-PIA-dependent activation of ERK1/2, implicating β-arrestin1 in A₁AR-mediated ERK1/2 activation (Fig. 3G). Transfection of the cells with a scrambled siRNA did not inhibit the activation of ERK1/2 by R-PIA, ruling out a nonspecific effect of the transfection reagent (Fig. 3G).

Fig. 3.

Increased transcription of A₁AR by agonist treatment is dependent on extracellular signal-regulated kinase (ERK)1/2. Real-time PCR was performed on RNA isolated from DDT₁ MF-2 cells treated with R-PIA for 24–72 h (A), 24-h R-PIA treatment alone or after 30-min PD98059 (10 μM) pretreatment (B), or β-Arr1 siRNA for 48 h (C). Fold change shown is the mean ± SE of 3 independent experiments. *Statistically significant difference from control or Scramble; **statistically significant difference from R-PIA-treated cells (P < 0.05; n = 3). D: DDT₁ MF-2 cells were serum starved for 12 h and treated with 1 μM R-PIA for various time periods. Western blotting was performed for p-ERK1/2 on whole cell lysates and ERK1/2 used for normalization. Similar experiments were performed in cells pretreated with either 1 μM DPCPX (E) or 10 μM PD98059 (F) for 30 min, followed by 1 μM R-PIA treatment for specified time points. G: cells were transfected with β-Arr1 siRNA or Scramble for 48 h. After overnight serum starvation, cells were exposed to 1 μM R-PIA for 2 and 5 min, and ERK1/2 activity was determined. Downregulation of β-arrestin1 by siRNA suppressed R-PIA activation of ERK1/2. Blot shown is a composite of 3 different segments of the same blot, denoted by dark vertical lines, which were rearranged for optimal presentation. Lanes presented reflect those showing similar protein loading, as determined by β-actin immunoreactivity.

Since increased ERK1/2 activity could contribute to increased AP-1 activity and A₁AR transcription (34), we determined the level of AP-1 activity in cells treated with R-PIA. Treatment of cells for 6 h with R-PIA (1 μM) increased AP-1 activity to 166 ± 19% of control (P < 0.05, n = 3), which was significantly attenuated by PD98059 (113 ± 9% of control), implicating ERK1/2 as a contributor to the activation process (Fig. 4A). PD98059 added alone inhibited AP-1 activity below that obtained in the control cells, likely reflecting some degree of ERK1/2 activity in the control cells over the treatment period. In cells transfected with β-arrestin1 siRNA for 48 h and then treated with R-PIA, A₁AR-mediated activation of AP-1 was significantly attenuated. AP-1 activity in the presence of scrambled siRNA + R-PIA, β-arrestin1 siRNA + R-PIA and β-arrestin1 siRNA alone were 140 ± 9%, 70 ± 5%, and 62 ± 11% of scrambled siRNA, respectively (Fig. 4B; P < 0.05; n = 5). The administration of scrambled siRNA did not alter basal or R-PIA-stimulated AP-1 activity (see Supplemental Fig. S4). These data suggest a role of the β-arrestin1/ERK pathway in mediating the A₁AR-dependent AP-1 activity. Additionally, A₁AR antagonist DPCPX also reduced the R-PIA-induced AP-1 activity, further confirming the role of A₁AR in this process (Fig. 4C).

Fig. 4.

A₁AR agonist increases β-arrestin1/ERK1/2-dependent activity of activator protein-1 (AP-1) transcription factor. Nuclear preparations were prepared from cells treated with vehicle (control), R-PIA (for 6 h), or PD98059 without or with R-PIA (A). Similarly, cells were treated with scramble or β-Arr1 siRNA for 48h and then administered vehicle or R-PIA for 6 h (B). C: AP-1 activity in cells pretreated with vehicle or DPCPX and then administered vehicle or R-PIA. Equal amounts of protein were used to determine the AP-1 DNA binding activity with an ELISA-based AP-1 transcription factor assay kit. Histograms represent means ± SE of at least 4 different experiments. Statistically significant differences: from *control and **R-PIA (P < 0.05). Control is the mean of the raw values of the vehicle-treated wells and was normalized to 100%. Calculated SEs represent the variation among these raw values.

Recovery of A₁AR after withdrawal of R-PIA involves de novo protein synthesis and activity of ERK1/2 pathway.

Cells that were treated with R-PIA for 24 h and then exposed to fresh medium without the agonist showed a rapid rate of recovery of A₁AR to the plasma membrane. Time course studies indicate that the levels of [³H]DPCPX binding were 75 ± 5% of control by 4 h and 102 ± 6% of control by 8 h after drug withdrawal (Fig. 5A). No further changes were evident on radioligand binding studies at 12 and 24 h. Interestingly, the binding of the agonist radioligand, which represents high-affinity receptors (coupled to G proteins), showed a similar recovery profile (data not shown). The increase in A₁AR during the withdrawal period could result from recycling of sequestered or internalized receptors or could involve de novo receptor synthesis. To test these possibilities, DDT₁ MF-2 cells were treated with 1 μM R-PIA for 24 h, followed by a 24-h withdrawal period in the presence of cycloheximide (5 μg/ml) to inhibit protein synthesis. In these studies, R-PIA reduced [³H]DPCPX binding to 45 ± 3% of control (n = 3; P < 0.01 from vehicle-treated control cells), while drug withdrawal for 24 h resulted in a recovery to 109 ± 6% (n = 3; P < 0.05 from R-PIA-treated group). Cycloheximide treatment for 24 h reduced this rebound increase to 63 ± 5% (n = 3; P < 0.05 compared with control recovery) (Fig. 5B). These findings suggest that de novo protein synthesis contributes, in part, to the recovery in surface expression of the A₁AR. When cells were exposed to R-PIA for 72 h and the drug was withdrawn, the recovery was almost completely blocked by cycloheximide (data not shown). This suggests that de novo protein synthesis is essential for mediating recovery following long-term agonist exposure. Therefore, these studies highlight two mechanisms underlying receptor recovery, those that are dependent or independent of protein synthesis.

Fig. 5.

A₁AR recovery after agonist withdrawal involves de novo receptor synthesis via ERK1/2 pathway. A: [³H]DPCPX radioligand binding assays were performed on membranes isolated from DDT₁ MF-2 cells cultured in the presence of R-PIA for 24 h (time 0) or in which the drug was withdrawn for up to 24 h. B: radioligand binding was performed in vehicle-treated control cells, cells treated with R-PIA for 24 h, cells treated with R-PIA for 24 h in which R-PIA was withdrawn for 24 h (WD), cells in which R-PIA was withdrawn in the presence of cycloheximide (CH + WD) or actinomycin D (Act D + WD), and cells in which these compounds were added alone (CH) or (Act D). Values are means ± SE from 3 independent experiments. Statistically significant differences: from *control, **R-PIA, and **WD (P < 0.05; n = 3). C: drug withdrawal in absence or presence of PD98059 or presence of PD98059 alone. Statistically significant differences: from *control, **R-PIA-treated, and ^#WD (P < 0.05; n = 3). D: cells were transfected with Scramble or β-Arr1 siRNA for 48 h and treated for 24 h with R-PIA followed by 24 h of no drug. Crude membrane extracts were used to perform the [³H]DPCPX radioligand binding experiments. Results are presented as means ± SE of 3 separate experiments. Statistically significant differences: from *control and **R-PIA treatment groups, respectively (P < 0.01; n = 3). All control groups were treated with vehicle for 24 h.

Pretreatment of cells with actinomycin D also reduced the recovery of A₁AR during drug withdrawal. This effect was apparent when drug withdrawal was initiated after both 24-h treatment with R-PIA (Fig. 5B) and 72-h treatment (data not shown). This suggests that loss of A₁AR occurring during these time periods required A₁AR transcription for receptor recovery. In cells pretreated with PD98059 (10 μM) the recovery of A₁AR following drug withdrawal was blunted (Fig. 5C), suggesting a role of the ERK/MAP kinase pathway in the process. [³H]DPCPX binding was reduced to 56 ± 3% of control by R-PIA treatment for 24 h and recovered to 90 ± 8% of control on drug withdrawal for 24 h, but this recovery was significantly inhibited to 57 ± 3% in the presence of PD98059 (Fig. 5C). Previous data (Fig. 3G) showed the role of β-arrestin1 in ERK1/2 activation by R-PIA. Hence, we determined whether β-arrestin1 siRNA can also reduce the recovery of the A₁AR (as seen with PD98059). However, interestingly, we found that β-arrestin1 siRNA did not affect the A₁AR recovery following R-PIA withdrawal (Fig. 5D). On the basis of these data, we conclude that ERK1/2 activity is essential for mediating the recovery of A₁AR after withdrawal of R-PIA, but β-arrestin1 is not required for this recovery.

R-PIA-mediated regulation of A₁AR does not involve NF-κB pathway.

In addition to AP-1, A₁AR promoter also has two binding sites for transcription factor NF-κB (25). In a previous study in DDT1 MF-2 cells, we showed (15) that the B-oligomer subunit of pertussis toxin can activate NF-κB and increase the expression of A₁AR. Treatment of cells with B-oligomer for 20 min decreased the steady-state levels of IκB-α, indicative of NF-κB activation. Hence, we first determined the levels of IκB-α in cells treated with R-PIA at different time intervals up to 60 min. As shown in Fig. 6A, R-PIA treatment did not affect the level of IκB-α over that observed for untreated controls. NF-κB dimer is present in an inactive form in the cytosol, and it translocates to the nucleus after activation. Next we performed Western blotting studies to assess for the translocation of p65 and p50 subunits of NF-κB. Similar to IκB-α, R-PIA treatment did not affect the activation of both p65 and p50 subunits as demonstrated by lack of increased expression of these subunits in nuclear preparations (Fig. 6, B and C). Western blots for p65 and p50 were normalized to the nucleus-specific protein lamin A/C. These data suggest a lack of involvement of NF-κB in R-PIA-mediated increase in A₁AR transcription. To further validate our findings, we determined the effect of the NF-κB inhibitor PDTC (10 μM) on R-PIA withdrawal. A₁AR antagonist radioligand binding studies showed a failure of PDTC to inhibit the recovery of A₁AR following drug withdrawal. After R-PIA withdrawal, [³H]DPCPX binding was 98 ± 18% of control (100%), while in the presence of PDTC it was 101 ± 17% (P < 0.05; n = 3). These findings suggest that R-PIA does not affect NF-κB activation, discounting a role of this transcription factor in the recovery of A₁AR following exposure to R-PIA.

Fig. 6.

Lack of involvement of nuclear factor-κB (NF-κB) in R-PIA-mediated downregulation and recovery. Cytosolic and nuclear cell lysates were prepared from DDT₁ MF-2 cells after the indicated treatment periods and used for Western blotting studies for IκB-α (A) and p65 (B) and p50 (C) subunits of NF-κB. The same blots were stripped and probed for β-actin (A) or nucleus-specific protein lamin A/C (B, C) for normalization. D: [³H]DPCPX radioligand binding was performed on the cell membranes prepared from DDT₁ MF-2 cells treated with vehicle (control) or R-PIA, after drug withdrawal for 24 h (WD), or withdrawal in presence of PDTC. Values are means ± SE from 3 different studies. *Statistically significant difference from control; **statistically significant difference from R-PIA (P < 0.05; n = 3).

DISCUSSION

The present study provides novel insights into the response of cells to agonist-mediated desensitization of the A₁AR. We show that upon exposure to agonist the A₁AR uncouples from its G proteins and downregulates via a β-arrestin1-dependent process. Association with β-arrestin1 is also important for A₁AR-dependent ERK1/2 activation. Finally, we show that the β-arrestin1/ERK pathway is important for mediating A₁AR activation of the AP-1 transcription factor and positive feedback regulation of A₁AR expression. Such a positive feedback loop is essential for ensuring rapid receptor recovery upon removal of the drug. Thus β-arrestin1 subserves A₁AR signaling in multiple ways, either through aiding in terminating the signal, by facilitating uncoupling and downregulation, or by aiding in recovery through increasing receptor transcription and synthesis.

The process of desensitization of the A₁AR has been extensively studied in various cell lines (for review see Ref. 16). As in other receptor systems, desensitization of the A₁AR limits the duration of its activation (especially in presence of a nonhydrolyzable agonist such as R-PIA) in order to facilitate homeostasis. This is ensured through multiple mechanisms, including reduced A₁AR coupling to G proteins possibly through receptor phosphorylation (30) or by decreasing cell surface expression of the A₁AR. However, unlike other G protein-coupled receptors, such as the β₂-adrenergic receptor, a more detailed study of the A₁AR desensitization has not been undertaken. While we have shown similarities in the desensitization process of the A₁AR and the β₂-adrenergic receptor (31, 24), another study has discounted the observation that the G protein-coupled receptor kinase (GRK) pathway is involved in A₁AR desensitization (26).

The present study highlights several molecular events that occur during the period of agonist exposure which facilitate the process of desensitization. We show that the A₁AR activation promotes translocation of β-arrestin1 to the plasma membrane. GRK-2 binds to the βγ-subunit of the receptor associated G protein(s) and generally promotes receptor phosphorylation on serine and threonine residues, a prerequisite for the binding of β-arrestin to the receptor and receptor uncoupling. We previously showed (24) uncoupling of the A₁AR by agonist, as determined by a loss in agonist high affinity, within 45 min of agonist exposure. This uncoupling of the A₁AR from its G proteins was associated with phosphorylation, albeit at low levels, since it could be partly reversed by dephosphorylation with phosphatases (24). The kinase(s) mediating A₁AR phosphorylation is still unclear. We have proposed a role of GRK-2 in this process based on a limited degree of phosphorylation of the purified bovine brain A₁AR by purified GRK-2 (31) and from the observation that arrestin reduces coupling of the GRK-2 phosphorylated A₁AR with purified G protein G_i/G_o preparations (24). A study by Ferguson et al. (5) had similarly indicated that a 30-min activation of the human A₁AR recruits β-arrestin2 to the plasma membrane of HEK293 cells, even though this was not associated with receptor internalization. Given that the rate of desensitization of the A₁AR is slow compared with the other AR subtypes, it is unclear whether these early events are significant in the process of A₁AR downregulation. However, they bear better correlation with a more rapid pace of agonist-mediated uncoupling of the A₁AR from G proteins in this cell line (24). Our study also supports a role of β-arrestin1 in the long-term downregulation of the A₁AR since knockdown of this protein by siRNA abolished agonist-mediated desensitization of the A₁AR. As such one can speculate that β-arrestin1 subserves both the uncoupling and the downregulation of the A₁AR.

A number of studies have demonstrated that activation of the A₁AR promotes ERK1/2 activation. This signaling pathway mediates, at least in part, the cardioprotective effect of A₁AR activation (33, 28). However, the mechanism underlying A₁AR-dependent ERK1/2 activation is not clear. The present study suggests that in this cell culture model β-arrestin1 mediates A₁AR-dependent activation of ERK1/2. We show rapid translocation of β-arrestin1 within the time frame of ERK1/2 activation. Furthermore, we show that knockdown of β-arrestin1 led to loss of receptor activation of ERK1/2. These data suggest that β-arrestin1 plays a key role in regulating A₁AR function, either through terminating its activity by uncoupling or endocytosis or through activating the ERK1/2 signaling pathway. As with other G protein-coupled receptors (21), the activation of ERK1/2 might proceed from the formation of a complex involving the internalized A₁AR, β-arrestin, and ERK1/2. Receptor internalization is dependent on the interaction of β-arrestin with clathrin, since a truncated form of β-arrestin that is unable to bind clathrin reduced proteinase-activated receptor 2 internalization (and ERK1/2 activation) (21).

Our data also extend our understanding of intracellular events that mediate cellular homeostasis after long-term agonist exposure. The data suggest that 24- and 72-h exposure to agonist led to losses in A₁AR that required transcription and de novo protein synthesis to ensure recovery upon withdrawal of the drug. One mechanism implicated in transcriptional activation of the A₁AR is receptor activation of the ERK1/2 and AP-1 signaling pathways during agonist exposure (35). We provide evidence implicating the ERK/AP-1 signaling in mediating the increased A₁AR transcription induced by R-PIA. First, we observed that the activation of ERK1/2 and the induction of A₁AR mRNA by R-PIA were inhibited by PD98059, an inhibitor of MEK, the upstream activator of ERK1/2, and also by FR180204, an ERK1/2 inhibitor. Second, inhibition of ERK1/2 activation in cells transfected with siRNA against β-arrestin1 reduced the agonist-dependent increase in A₁AR mRNA. Third, R-PIA stimulated nuclear AP-1 activity, which was suppressed by inhibition of ERK1/2 activity by PD98059 and β-arrestin1 siRNA. The A₁AR gene promoter B contains consensus binding sites for the AP-1 transcription factor, which regulates the basal expression of the A₁AR (35). Thus we propose that induction of A₁AR mRNA by agonist involves activation of the ERK1/2 pathway, leading to induction of c-fos (and associated c-jun genes), the formation of Fos/Jun complexes, and activation of the AP-1 transcription factor.

Interestingly, we observed that activation of the A₁AR did not alter NF-κB activity and that inhibition of this transcription factor did not affect recovery of the A₁AR after desensitization. In our view, these findings would suggest a role of AP-1 and not NF-κB in the homeostatic regulation of the A₁AR mRNA during desensitization. In contrast, we propose that NF-κB, but not AP-1 transcription factor, mediates stress-dependent induction of the A₁AR (25). Together, our present findings and previous observations suggest differential utilization of the AP-1 and NF-κB transcription factors in regulating A₁AR expression depending on the nature of drug exposure. Both of these signaling pathways serve to ensure adequate levels of expression in the cell.

We are not aware of similar findings among G protein-coupled receptors in which an agonist can mediate its desensitization and downregulation, and recovery through activation of a single signaling cascade, such as the β-arrestin1-ERK-AP/1 pathway. Previous studies have shown that chronic agonist treatment can decrease G protein-coupled receptor (GPCR) mRNA, and this can further contribute to the process of receptor downregulation (2, 13). Collins et al. (3) showed the stimulation of β₂-adrenergic receptor transcription by cyclic AMP. However, this was after short-term (generally in minutes) agonist exposure. On long-term agonist exposure (for hours) the mRNA for β₂-adrenergic receptor was downregulated. A recent study indicates that activation of the histamine H1 receptor (H1R) increased H1R gene promoter activity as well as H1R protein by activating protein kinase C (4). Such a mechanism might account for the upregulation of H1R observed in the nasal mucosa during allergic rhinitis (14). The difference between that study and ours is that the agonist induction of A₁AR mRNA was not associated with an increased in receptor protein until the agonist was withdrawn. Several studies have also demonstrated heterologous regulation of receptors. For example, Miyoshi et al. (22) reported upregulation of H1R mRNA by stimulation of M₃ muscarinic receptor (m₃AchR) but that activation of the β₂-adrenergic receptor led to downregulation of the H1R (23). The actions of the m₃AChR and the β₂-adrenergic receptors were mediated by protein kinase C and A, respectively (22, 23). In addition, activation of the type 1 angiotensin receptor reduced transcription of the type 2 angiotensin receptor via ERK1/2 MAP kinase pathway in transfected PC12 cells (38).

A yet unanswered question is why the levels of A₁AR are maintained at a reduced level (∼50% of control) over the period of agonist exposure. This probably indicates inhibition of some posttranscriptional mechanism(s) that suppresses translation or increased degradation of newly synthesized receptors. This suppression is dependent on the presence of the agonist, since a rapid recovery in the level of the A₁AR was observed after withdrawal of the drug. On the basis of these observations, we propose that the increase in A₁AR mRNA during the desensitization process serves to prime the cell for rapid de novo receptor synthesis upon termination of agonist exposure.

In summary, the present results highlight novel mechanisms by which activation of the A₁AR can regulate its function and expression. On the basis of the present and existing data, we propose that the β-arrestin ERK MAP kinase pathway mediates the physiological role of the A₁AR, contributes to the termination of A₁AR signaling through desensitization and downregulation, and primes the transcriptional machinery for rapid synthesis of the receptor upon termination of the drug.

GRANTS

This study was supported by the Southern Illinois University School of Medicine Excellence in Academic Medicine Grant and National Institutes of Health Grants RO1-DC-02396 and R15-CA-135494.

NOTE ADDED IN PROOF

The authors state that Figs. 1C (inset), 2B, and 3G in the Articles in PresS version of this article have been modified in this final version to indicate that they form composites of different lanes from the same gel and that the figure legends have been modified to reflect these changes. These changes do not alter the conclusions of this study.