Abstract
α1-Adrenoceptor subtypes (α1A-, α1B-, α1D-) are known to couple to similar signaling pathways, although differences among the subtypes do exist. As a more sensitive assay, we used oligonucleotide microarrays to identify gene expression changes in Rat-1 fibroblasts stably expressing each individual subtype. We report the gene expressions that change by at least a factor of 2 or more. Gene expression profiles significantly changed equally among all three subtypes, despite the unequal efficacy of the inositol phosphate response. Gene expressions were clustered into cytokines/growth factors, transcription factors, enzymes, and extracellular matrix proteins. There were also a number of individual subtype-specific changes in gene expression, suggesting a link to independent pathways. In addition, all three α1-AR subtypes robustly stimulated the transcription of the prohypertrophic cytokine interleukin (IL)-6, but differentially altered members of the IL-6 signaling pathway (gp-130 and STAT3). This was confirmed by measurement of secreted IL-6, activated STAT3, and gp-130 levels. Activation of STAT3 Tyr705 phosphorylation by the α1-ARs was not through IL-6 activation but was synergistic with IL-6, suggesting direct effects. Interestingly, α1B-AR stimulation caused the dimerization-dependent phosphorylation of Tyr705 on STAT3 but did not activate the transcriptional-dependent phosphorylation of Ser727. The α1B-AR also constitutively down-regulated the protein levels of gp-130. These results suggest that the α1B-AR has differential effects on the phosphorylation status of the STAT3 pathway and may not be as prohypertrophic as the other two subtypes.
α1-Adrenoceptors (ARs) belong to the superfamily of G-protein-coupled receptors (GPCR) that mediates the functions of catecholamines. Once activated by binding, α1-ARs initiate the cellular pathways leading to the regulation of physiological effects, including blood pressure maintenance, glucose metabolism, renal sodium reabsorption, and cardiac inotropy (Michelotti et al., 2000; Piascik and Perez, 2001). Three α1-AR subtypes (α1A-, α1B-, and α1D-AR) have been cloned and characterized pharmacologically (Cotecchia et al., 1988; Lomasney et al., 1991; Perez et al., 1991). All three α1-AR subtypes display comparable binding affinities to catecholamines and show prevalence for coupling to the Gq family of G proteins, thereby activating IP3-mediated increases in calcium. Despite having similar binding affinities for the endogenous ligands and predominance toward Gq signaling, all three α1-AR subtypes are expressed in most tissues, thus raising the question of whether or not α1-AR subtypes carry out redundant functions.
Discerning the physiological roles of α1-AR subtypes in vivo has proven difficult, mainly because of the lack of probes with sufficient subtype selectivity (i.e., ligands and low avidity antibodies). This issue has been partly addressed by the use of cell lines transfected with individual α1-AR subtypes, and transgenic animal models in which α1-AR subtypes have been either knocked out (Cavalli et al., 1997; Rokosh and Simpson, 2002; Tanoue et al., 2002) or overexpressed (Milano et al., 1994; Zuscik et al., 2000; Lin et al., 2001).
Besides vascular effects, α1-ARs are known to stimulate hypertrophy of cardiac myocytes by activating established signaling pathways (reviewed in Varma and Deng, 2000). This effect of α1-ARs may be relevant during the pathogenesis of congestive heart failure, because plasma concentrations of the sympathetic neurotransmitter norepinephrine are elevated during the onset of heart failure. In addition to increased cardiac sympathetic drive, circulating levels of the proinflammatory cytokine IL-6 (and IL-6 family members LIF and cardiotrophin-1) are also elevated in patients with heart failure (Kanda et al., 2000; Eiken et al., 2001; Ng et al., 2002). A role for IL-6 in the development of heart failure has been proposed based upon evidence that mice overexpressing IL-6 or its receptor develop ventricular hypertrophy (Hirota et al., 1995), whereas IL-6 receptor or gp-130 knockout mice develop thin ventricular walls (Taga et al., 1996) or a massive apoptosis before the onset of heart failure (Hirota et al., 1999). Recent evidence indicates that other well-known cardiac stimuli, including angiotensin II (Sano et al., 2000) and activation of β-ARs (Murray et al., 2000), increase expression of IL-6 in the heart, indicating a correlation between neuroendocrine signaling and cytokine release.
α1-AR subtype-specific signaling has been suggested previously, particularly in studies of α1-AR stimulation of MAPK pathways (Zhong and Minneman, 1999), which integrate extracellular signals that regulate cell growth, proliferation, and fate. In the present study, we employed oligonucleotide microarrays to provide new insights, and a more sensitive assay of α1-AR subtype signaling. We compared the profiles of gene expression after short-term (60-min) epinephrine-stimulation of Rat-1 fibroblasts stably transfected with individual α1-AR subtypes and compare these with stimulated nontransfected cells. Our studies indicate that within the same cellular context, α1-AR subtypes can stimulate or inhibit identical profiles of expression, despite differences in IP efficacy, as well as profiles that are subtype- and/or pathway-specific. We report that α1-ARs activate the secretion of IL-6 and that there are synergistic effects on STAT3 activation, suggesting that α1-AR mediation of STAT3 is through non–IL-6 effects, which was confirmed through neutralizing antibodies. Interestingly, the α1B-AR was not synergistic or as synergistic with IL-6 activation. This could be because the α1B-AR displays constitutive down-regulation of gp-130 protein levels. Our data also suggests that α1B-AR activation can stimulate the dimerization-dependent phosphorylation of Tyr705 on STAT3 but not the transcriptional-dependent phosphorylation of Ser727, the first report of this differential activation of STAT3. These results could be important in regulating pro-hypertrophic pathways and may be differentially regulated by the α1-AR subtypes.
Materials and Methods
Materials.
[125-I]BE-2254, [myo-3H]inositol, and [γ-32P]dCTP were obtained from PerkinElmer Life Sciences (Boston, MA). AG 1-X8 resin (formate form), 4 to 15% Ready-Gels, and Bio-Rad protein assay were purchased from Bio-Rad Laboratories (Hercules, CA). Full-length rat p21-cK-Ras and c-fos plasmids were obtained from American Type Culture Collection (Manassas, VA). M-PER lysis reagent and West-Pico enhanced chemiluminescent reagent were from Pierce (Rockford, IL). The gp-130 (M-20) antibody was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Both STAT3 and Phospho-STAT3 (Ser727 and Tyr705) antibodies were obtained from Cell Signaling Technology (Beverly, MA). IL-6 and IL-6 neutralizing antibodies were from R & D systems (Minneapolis, MN). Horseradish peroxidase-conjugated anti-rabbit IgG was from Jackson Laboratories (Bar Harbor, ME). (−)-Epinephrine bitartrate, (S)-(−)-propranolol hydrochloride, rauwolscine hydrochloride, and prazosin hydrochloride were from Sigma-Aldrich (St. Louis, MO). Phentolamine mesylate was from RBI/Sigma (Natick, MA). Dulbecco's modified Eagle's medium (DMEM), penicillin, streptomycin, trypsin-EDTA, and Ca2+-Mg2+–free phosphate-buffered saline (PBS) were obtained from the Cleveland Clinic Foundation (CCF) house facility (Cleveland, OH). Fetal bovine serum was obtained from Cambrex Bio Science Walkersville, Inc. (Walkersville, MD). G418 was purchased from Calbiochem (La Jolla, CA).
Cell culture.
Rat-1 fibroblasts stably transfected with human α1-AR cDNAs corresponding to either the α1A-, α1B-, or α1D-AR subtype were a gift of GlaxoSmithKline (Uxbridge, UK). These cells were derived from clonal isolates. Cells were propagated in 75-cm2 flasks in a humidified atmosphere (37°C) in DMEM containing 5% fetal bovine serum, 10 U/ml penicillin, 100 μg/ml streptomycin, and 500 μg/ml of the selection antibiotic G418. Cells were detached by trypsinization (0.05% trypsin, 0.53 mM EDTA) and subcultured at a ratio of 1:5 upon reaching confluence. Nontransfected Rat-1 fibroblasts were maintained under the same conditions but in the absence of G418.
Saturation Binding.
Saturation binding experiments were performed using membranes from the Rat-1 cells stably transfected with the α1A-, α1B- or α1D-AR subtypes. Membrane isolations were performed as reported previously (Perez et al., 1991). Binding assays were performed in triplicate in a 0.25-ml assay volume with eight concentrations (0–400 pM) of 125I-BE-2254 and 10 μg of membrane protein per tube. After equilibration for 30 min at 25°C, free and bound radioligand were separated by rapid filtration using a Brandel cell harvester and Whatman GF-C glass-fiber filters. Nonspecific binding was defined as the amount of radioactivity that remained bound to the filters in the presence of 10 μM phentolamine.Bmax (maximum receptor density) andKd (affinity) values were obtained using the nonlinear regression function of Prism (GraphPad, San Diego, CA).
Quantitation of Epinephrine-Stimulated Total IP Accumulation.
Confluent cells expressing α1-AR subtypes were grown for 24 h in the presence of 5 μCi of [3H]myo-inositol. Propranolol (1 μM) and rauwolscine (0.1 μM) were added to the growth media for 30 min to block putative β- and α2-ARs, respectively, followed by a 60-min incubation of 10 μM epinephrine in the presence of 10 mM LiCl. After drug incubations, the monolayers were washed with ice-cold PBS and the cells were lysed with 0.4 M perchloric acid. Lysates were scraped, collected, and neutralized by the addition of a 0.5-ml volume of a 0.72 N KOH/0.6 M KHCO3solution. Soluble [3H]IPs were isolated from the lysates by column chromatography using AG 1X-8 resin-packed columns. The columns were washed with 0.1 M formic acid, and the resin-bound [3H]IPs were displaced by elution with a 0.1 M formic acid solution containing 1.0 M ammonium formate. The eluant was collected in scintillation vials and the radioactivity was detected using a β-counter (Beckman, Irvine, CA). EC50 (potency) values and maximal epinephrine responses were compared using Prism (GraphPad).
Preparation of Biotin-Labeled cRNA and Hybridization to Oligonucleotide Arrays.
Five pooled flasks (150 mm2 each) of either nontransfected or α1-AR subtype-transfected Rat-1 fibroblasts were used for individual hybridization to the oligonucleotide microarray. Two separate hybridizations from two separate RNA preparations were performed per cell group. Briefly, confluent monolayers were first incubated for 30 min with propranolol (1 μM) and rauwolscine (0.1 μM), followed by an additional 60-min incubation with 10 μM epinephrine. After incubations, the media was removed and the monolayers rinsed twice with ice-cold PBS. Poly(A)+ RNA was immediately isolated using the FastTrack 2.0 Kit from Invitrogen (Carlsbad, CA) and stored overnight at −70°C. Double-stranded cDNA was then synthesized from 1.0-μg aliquots of poly(A)+ RNA using the SuperScript Choice double-stranded cDNA synthesis kit from Invitrogen. From each sample, cDNA was purified by phenol/chloroform extraction and ethanol precipitation, and biotin-labeled cRNA was synthesized via an in vitro transcription reaction using the BioArray high-yield RNA transcript labeling kit (Enzo Diagnostics, Farmingdale, NY). cRNA transcripts were then purified from the in vitro transcription reaction using the RNeasy Mini kit from QIAGEN (Valencia, CA). The fragmentation of biotin-labeled cRNAs and hybridization of these fragments to the oligonucleotide arrays were both carried out by the Gene Expression Core Service at the CCF. To determine the quality of the mRNA preparations and the subsequent manipulations between test samples, an aliquot of the biotinylated cRNA fragments from each sample were hybridized to an Affymetrix (Santa Clara, CA) “test chip” before the rat genome U34A array was used. This test chip analyzes the percentage of genes that are present in the sample that hybridizes to the genes present in the test chip. The analysis also includes a measure of the amount of full-length cRNA transcribed as determined by the ratios of 3′/5′ regions of both glyceraldehyde-3-phosphate dehydrogenase and β-actin. Aliquots of biotin-labeled cRNA fragments were then hybridized to rat genome U34A Arrays (Affymetrix) containing 7000 known genes and expressed sequence tags from build 74 of the UniGene database. The hybridization signal was amplified by the Antibody Amplification Protocol as described in the Affymetrix GeneChip Expression Analysis Manual.
For data analysis, matrix-based decisions concerning the hybridization of a cRNA to a particular probe set were executed using Affymetrix software. Briefly, each gene probe set was represented on the chip by 25 pairs of perfectly matched (PM) and mismatched (MM) oligonucleotides. Each of the 25 unique PM/MM pairs was different by only a single base-pair change in the MM oligonucleotide, and spanned various places along the gene represented by the probe set to include regions represented by the 5′-untranslated, coding, and 3′-untranslated regions. The MM probes acted as specificity controls that allowed the direct subtraction of both background and cross-hybridization signals. The number of instances in which the PM signal was greater than the MM signal was determined and the average of the logarithm of the PM/MM ratio was calculated. These values were used in a matrix-based algorithm that determined the absence or presence of a cRNA molecule in the experimental sample. To determine the abundance of each RNA, the average of the differences representing PM minus MM for each probe set was calculated after discarding the maximum, the minimum and any outliers beyond three standard deviations. All combinations of pair comparisons were then made between the control (nontransfected) Rat-1 fibroblasts and those transfected with each α1-AR subtype [i.e., α1A-AR (chip 1) versus control (chip 1); α1A-AR (chip 2) versus control (chip 1), etc.]. Therefore, for each subtype, a four-way comparison was performed. This results in four numerical values for each cell line. Additional comparisons were made between transfected cells (i.e., α1A-AR versus α1B-AR; α1A-AR versus α1D-AR, etc.). For each comparison, changes in the expression of a particular gene had to exhibit an average of 2.0-fold or greater and be present in each of the four-way analyses to be included in the tables. This decision was based upon previous reports that TaqMan verification of the microarray is valid in the 1.7- to 1.8-fold range (Tan et al., 2002). Tables reflect the mean values of the four-way comparisons plus or minus the S.E.M. Expressed sequence tags were excluded from the analysis. The use of two chips reduces the error rate of false positives and false negatives caused by chip-based errors to 1 in 10,000, according to manufacturer's specifications.
Northern Blot Analysis.
Two genes that were identified via microarray to either increase or remain unchanged by activation of α1-AR subtypes (c-fos and p21-cK-Ras) were examined by Northern analysis. Drug incubation protocols were kept identical to those in microarray experiments. Poly(A)+ RNA (10 μg) extracted with the Invitrogen FastTrack 2.0 kit were loaded onto 0.8% agarose-formaldehyde gels for subsequent transfer to nitrocellulose and hybridization. cDNA probes were derived from full-length plasmids obtained from American Type Culture Collection. Probes (c-fos, 5.0-kbEcoRI fragment; p21-cK-Ras, 4.4-kbEcoRI-HindIII fragment) were random-primed with [32P]dCTP (6000 Ci/mmol) using the random primed DNA labeling kit from Roche (Indianapolis, IN). Labeled probes (1 × 106 cpm/ml) were hybridized to the membranes overnight at 50°C. Membranes were each washed in 5× then 2× SSC containing 0.1% SDS for 10 min, and exposed to X-ray film for 3 (c-fos) or 48 h (p21-cK-Ras). Qualitative differences in band intensities were compared between α1-AR subtype transfected and nontransfected Rat-1 cells using the software program NIH Image (http://rsb.info.nih.gov/nih-image/).
Measurement of IL-6 Secretion.
Rat-1 fibroblasts stably expressing α1-AR subtypes were incubated for 24, 48, and 72 h with 10 μM epinephrine in the presence of 1 μM propranolol and 0.1 μM rauwolscine. Additionally, experiments were performed in α1-AR subtype expressing Rat-1 fibroblasts that were incubated for 48 h with epinephrine in the presence of β-AR and α2-AR antagonists and the α1-AR antagonist prazosin (1 μM). Antagonists and epinephrine were replenished in the culture media every 12 h. After incubation, the culture medium was removed and concentrated by vacuum centrifugation. The IL-6 levels in the concentrate were determined by enzyme-linked immunosorbent assay (ELISA) using the rat IL-6 module set from Alexis Biochemicals (San Diego, CA) following the manufacturer's instructions. Sample IL-6 concentrations were extrapolated by spectrophotometric (405 nm) conversion of standard curves ranging in sensitivity from 31 to 2000 pg/ml. To account for IL-6 in serum, background absorbance values (DMEM containing serum and antibiotics) were subtracted from each sample.
Immunoblotting-Immunoprecipitation Studies.
To investigate whether changes in transcription for gp-130 andSTAT3 genes would be reflected in changes at the translational level, antibodies to gp-130 and STAT3 were employed in Western blotting experiments of α1-AR subtype transfected cells treated with 10 μM epinephrine for 1 or 24 h. Nontransfected Rat-1 fibroblasts were also included in these experiments to exclude any effect of epinephrine incubation on gp-130 and/or STAT3 total protein levels. The effect of α1-AR subtype stimulation on gp-130 total protein levels was studied after 15, 30, 60, 120, and 180 min of 10 μM epinephrine stimulation. After incubation, the monolayers were washed with ice-cold PBS, and the cells were scraped in M-PER lysis buffer (Pierce) containing protease inhibitors (1.0 mM phenylmethylsulfonyl fluoride, 1.0 mM EDTA, and 0.2 mM aprotinin). Total protein content in the lysates was determined using the Bio-Rad protein assay. Proteins were separated on 4 to 15% gradient gels by electrophoresis. The gels were transferred onto nitrocellulose membranes, blocked for 1 h, and incubated overnight with either STAT3 or gp-130 antibodies. Membranes were incubated with an anti-rabbit HRP-conjugated secondary antibody and visualized by enhanced chemiluminescence.
The epinephrine-induced transcriptional activity of STAT3 was determined in immunoprecipitation-immunoblotting experiments of subtype transfected Rat-1 fibroblasts that had been incubated with 10 μM epinephrine for various times (15, 30, 60, and 120 min). Total STAT3 protein was immunoprecipitated from cellular lysates (100 μg of total protein per time point) using the STAT3 antibody mentioned above. Immunoprecipitated proteins were separated in 4 to 15% gradient gels, which were transferred to nitrocellulose membranes. Membranes were blocked for 1 h and incubated overnight with an antibody designed to recognize the transcriptionally active form of STAT3 (phosphorylated at the Ser727 residue) or the dimerization-dependent phosphorylation of Tyr705. The level of epinephrine-stimulated phosphorylated STAT3 levels at each time point was compared with that of basal levels (time 0, no epinephrine). To compare protein loading, membranes were stripped and reprobed with the same antibody used to precipitate STAT3. To measure IL-6–mediated effects, IL-6 (40 ng/ml; R and D Systems) was added to the medium for 30 min for STAT3 activation or 3 h for gp-130 protein determination. Inhibition of IL-6–mediated biological function was achieved through a neutralizing antibody (R and D Systems), first given in a dose-response (0.05–5 μg/ml) and then used at its maximal dose (5 μg/ml). Data from immunoblotting-immunoprecipitation studies were analyzed by qualitative comparison of band intensities obtained at each time point versus that of time 0 (basal). Western blots were analyzed and compared in band intensities using the software program NIH Image.
Results
Saturation Binding.
The levels of α1-AR subtype expression in Rat-1 fibroblasts were compared in membrane saturation binding assays using the radioligand 125I-BE-2254. Rat-1 fibroblasts expressed similar densities and affinities of single-site binding curves. The Bmax values for α1A-, α1B-, and α1D-AR transfected cells were 1.3 ± 0.1, 1.4 ± 0.4, and 2.6 ± 0.5 pmol/mg, respectively. TheKd values for125I-BE-2254 were also similar among subtypes: α1A-AR, 19.8 ± 1.5 pM; α1B-AR, 30.5 ± 6.5 pM; α1D-AR, 58 ± 6.5 pM. In all binding assays, nonspecific binding at the Kdranged from 9.2 to 15% of the total radioligand bound, and no more than 10% of total radioactivity added was bound.
Quantitation of Epinephrine-Stimulated Total [ 3H]IP Accumulation.
As a measure of α1-AR subtype-Gq signaling, concentration responses to epinephrine stimulation of [3H]IP accumulation were performed in α1-AR subtype transfected Rat-1 fibroblasts (Fig. 1). No significant differences in potency (−log EC50, molar) were noted for epinephrine stimulation of total [3H]IP accumulation among α1-AR subtypes: α1A-AR, 7.16 ± 0.15; α1B-AR, 7.86 ± 0.24; α1D-AR, 7.72 ± 0.15. However, the maximal responses to epinephrine varied among subtypes, and were significantly higher in α1A-AR cells than those of α1B-AR (by 4.1-fold) and α1D-AR (by 2.8-fold) cells. The maximal responses to epinephrine in α1D-AR cells were modestly higher than those in α1B-AR cells (by 1.2-fold). There were no differences in basal or agonist-independent [3H]IP counts among α1-AR subtype transfected fibroblasts (data not shown).
Comparison of mRNA Quality between Samples.
The results of the “test chip” analysis between the samples used in the microarray study revealed that both hybridizations from separate RNA preparations had reproducible parameters of quality. All three parameters that determine the amount (percentage present) and quality of mRNA (3′/5′ ratios) for both glyceraldehyde-3-phosphate dehydrogenase and β-actin were comparable between samples and are considered excellent.
Genes commonly modified by α1-AR subtypes.
The gene expression profiles of Rat-1 fibroblasts stably expressing α1-AR subtypes is shown in Table1. Common to all three subtypes, gene expression changed for 29 genes by at least two-fold relative to nontransfected Rat-1 fibroblasts. These common gene expressions have been divided into four clusters, which include genes that code for: 1) secreted endogenous peptide ligands (cytokines and growth factors); 2) DNA binding proteins (transcription factors and immediate early genes); 3) signaling and/or catabolic enzymes; and 4) extracellular matrix proteins. In each cluster, genes have been arranged in decreasing order of magnitude of expression changes, starting with the highest fold-values of α1A-AR fibroblasts as reference. Thus, the strongest stimulations for each cluster were for the gp-130 related cytokine, IL-6, the immediate-early gene c-fos, the tyrosine phosphatase, CL100, and fibronectin, respectively. Figure 1 shows that even though the epinephrine-stimulated accumulation of total IPs was most efficacious in α1A-AR cells, all three subtypes changed the gene expressions of most of the genes in Table 1by similar folds. In addition, some genes (GD-VEGF, CELF, rNFIL-6, Arc, RNR-1) were more strongly changed in their expression in α1B-AR and/or α1D-AR cells than α1A-AR cells.
Table 1 also shows that within each cluster there were genes belonging to particular families: VEGF3 and GD-VEGF (heparin binding FGF); IL-6 and LIF (gp-130 binding cytokines) in cluster 1; c-fos and c-jun (activator protein-1); Egr-1 and Egr-2 (Zinc-finger DNA binding) in cluster 2; and Tyrosine phosphatases CL100 and BAD2 (dual purpose, threonine/tyrosine phosphatase) in cluster 3.
Analysis of the microarray data also revealed that α1-AR subtypes commonly inhibited the gene expression of 9 genes by at least 2-fold versus nontransfected cells (Table 2). The fold-decreases in the gene expressions for each of the 9 “commonly-inhibited” genes were of similar magnitude among α1-AR subtypes.
α1-AR subtype-unique changes in gene expression.
Table 3 shows that compared with nontransfected Rat-1 fibroblasts, multiple changes in gene expression that were specific to each α1-AR subtype existed. These included, for each subtype, both positive and negative changes, which varied in magnitude and number of genes modified. The α1B-AR modified the greatest number of genes (17), followed by the α1D-AR (12) and α1A-AR (6). We noted that in α1B-AR cells, a relatively large number of gene expressions modified have been associated with neurodegeneration and apoptosis, including tau and synuclein (Golbe, 2002), transforming growth factor β3 (Dunker et al., 2001), and caspase-6 (Mac-Lachlan and El-Deiry, 2002).
Differential regulation of IL-6 signaling genes by α1-AR subtypes.
In addition to changes in gene expression that were common to all subtypes, and to subtype-specific changes, the expression of certain genes changed in a different way: stimulated by two, but not all three subtypes. Table4 shows that this was true of genes that code for members of the IL-6 signaling pathway (gp-130, STAT3 and Ras). Table 4 shows that although all subtypes increased the gene expression of the cytokine, IL-6, only α1A-AR and α1D-AR subtypes were linked to increases in gene expression of gp-130 (the IL-6 high affinity receptor and signal transducer), and STAT3 (prototype IL-6–stimulated transcription factor). p21-Ras has been included in this pathway becauseit has been shown to relay IL-6/gp-130 signals in some cells (Taga and Kishimoto, 1997).
Northern blots.
To validate some of the data obtained with the microarray technique, Northern blotting analyses were carried out for two different genes (c-fos and p21-cK-Ras) present in the microarray. Drug incubation protocols for obtaining northern and microarray data were kept identical. Figure 2 shows that in nontransfected cells epinephrine had no effect in the transcription of c-fos (NT lane). On the other hand, epinephrine robustly stimulated the transcription of c-fos in α1-AR subtype transfected cells. The increase in c-fos transcription by northerns followed a similar pattern to those obtained by microarrays: α1A-AR (by 63.8 ± 23.4-fold), α1B-AR (43.5 ± 12.5-fold), and α1D-AR (61 ± 13-fold) cells relative to nontransfected controls. However, these changes were significantly higher in the microarray analysis, which is consistent with previous reports that the array is a more sensitive assay for detecting changes in mRNA levels (Yun et al., 2003).
Northern blots also showed increased transcript levels of p21-cK-Ras in α1A-AR and α1D-AR fibroblasts, whereas these levels remained unchanged in α1B-AR cells, relative to nontransfected controls. Qualitatively, these transcriptional changes also mimic those obtained by microarrays: α1A-AR (increased by 5.3 ± 0.4 fold), α1B-AR (no change), and α1D-AR (8.4 ± 0.6) compared with nontransfected cells.
Epinephrine-Stimulated Secretion of IL-6 by α1-AR Subtypes.
Because IL-6 has been associated with cardiac hypertrophy and heart failure and was one of the most changed in gene expression, we decided to explore the IL-6 pathway in more detail. The concentration of secreted IL-6 in culture medium was measured in ELISA experiments. The first set of experiments showed that in Rat-1 fibroblasts, α1-AR subtypes stimulated an increase in the concentration of IL-6 in culture medium after epinephrine incubation. The effect was time-dependent. No IL-6 could be detected in culture medium during the first 24 h of epinephrine incubation; however, longer incubation times (48 and 72 h) resulted in increased IL-6 levels (α1A-AR, 7.3- and 24-fold; α1B-AR, 5.5- and 12.5-fold; α1D-AR, 8.5 and 12.6-fold, respectively). This is probably a result of the sensitivity of the ELISA assay and is similar to previous reports that examined IL-6 secretion (Norris and Benveniste, 1993). Separate experiments summarized in Fig.3B showed that at 48 h of epinephrine incubation, the accumulated IL-6 concentration in medium from α1-AR subtype expressing cells was completely abolished by prazosin (1 μM, 48 h) in α1A- and α1B-AR expressing cells and mostly blocked in α1D-AR cells. No IL-6 protein accumulation was detected in nontransfected Rat-1 fibroblasts treated with identical epinephrine conditions (data not shown).
Epinephrine-Stimulated Changes in STAT3 Protein and Ser727-Phosphorylation by α1-AR Subtypes.
Fig.4A shows the effect of epinephrine incubation on STAT3 total protein levels in Rat-1 fibroblasts transfected with α1-AR subtypes. After 1 h of incubation with 10 μM epinephrine, we observed no differences in STAT3 total protein levels between epinephrine-stimulated and nonstimulated subtype-transfected cells. Longer epinephrine exposure (24 h) resulted in substantially higher levels of STAT3 total protein, particularly for α1D-AR cells and, to a lesser extent, for α1A-AR expressing cells. We observed no differences in total STAT3 protein levels between epinephrine-stimulated and nonstimulated α1B-AR fibroblasts at either incubation time.
Fig. 4B shows the effect of epinephrine on stimulation of the Ser727 phosphorylated form of STAT3 in Rat-1 fibroblasts expressing individual α1-AR subtypes. We observed that activation of both the α1A- and α1D-AR subtypes increased the phosphorylation of Ser727 STAT3 in a time-dependent manner. On the other hand, epinephrine incubation of α1B-AR transfected cells consistently reduced the phosphorylation status of Ser727 STAT3 below basal levels (time 0, no agonist). In addition, Fig. 4B shows that epinephrine had no effect on total STAT3 protein levels of α1-AR subtype transfected Rat-1 fibroblasts at any time point.
α1-AR Subtype Regulation of gp-130 Protein Levels in Rat-1 Fibroblasts.
The effect of α1-AR subtype activation on total gp-130 protein levels was studied by Western blotting experiments with α1-AR subtype-transfected and nontransfected fibroblasts that were incubated with epinephrine (10 μM) for 1 and 24 h (Fig. 6A). At 1 h incubation, we observed no differences in gp-130 protein levels between epinephrine-stimulated and nonstimulated fibroblasts (applies to both nontransfected and α1-AR subtype-transfected fibroblasts). However, we observed that the α1B-AR reduced the levels of gp-130 total protein in an agonist-independent manner compared with nontransfected and α1A- and α1D- cells. This agonist-independent reduction in total gp-130 protein in α1B-AR cells was also apparent at 24 h of epinephrine incubation. Although epinephrine had little or no effect in α1A- and α1D-AR cells after 1 h of treatment, incubation for 24 h almost completely abolished the levels of gp-130 total protein relative to those of nonstimulated cells.
The reduction in levels of total gp-130 protein in epinephrine-stimulated α1A- and α1D-AR cells was characterized in a time-course experiment (Fig. 5B). By 2 and 3 h of epinephrine incubation, a reduction in total gp-130 protein levels was apparent in α1A- and α1D-AR cells, whereas no changes were observed in α1B-AR cells (even when using higher protein loading to compensate for the loss of gp-130 expression by α1B-AR transfection alone). Figure 5C indicates that at 3 h of epinephrine incubation, the observed reduction in total gp-130 protein content in α1A-AR and α1D-AR cells can be blocked by the α1-AR antagonist prazosin (1 μM).
IL-6 Regulation of STAT3 Activation and gp-130 Levels.
To ensure that a correct dose of the neutralizing antibody was used, the IL-6–mediated phosphorylation of Tyr705 was first used in a dose-response to increasing concentrations of the neutralizing antibody (Fig. 6). Using a range of concentrations from 0.05 μg/ml to 5 μg/ml, only the 5 μg/ml dose provided sufficient blockage of the IL-6 response.
Using an antibody that recognizes the dimerization-dependent Tyr705 phosphorylation state of STAT3, the exogenous addition of IL-6 to the culture medium resulted in increased levels of the phospho-protein in all three α1-AR subtype cell lines (Fig.7). This activation could be blocked with the addition of a neutralizing antibody to IL-6 at 5 μg/ml. Interestingly, all three α1-AR subtypes also caused increases in Tyr705 phosphorylation, including the α1B-AR, which previously did not activate Ser727 phosphorylation. Epinephrine-mediated Tyr705 phosphorylation could not be blocked by the IL-6 neutralizing antibody, suggesting direct effects. This is also supported by the fact that epinephrine and IL-6 together caused a synergistic phosphorylation of Tyr705 but was weak for the α1B-AR.
To determine the effects of IL-6 on the protein levels of gp-130, IL-6 was added to the culture medium for 3 h, a condition in which α1-ARs cause a down-regulation of the protein (Fig. 8). Whereas α1A- and α1D-AR activation cause the down-regulation of gp-130, the addition of IL-6 caused an increase in gp-130 levels. The α1B-AR cell line, as before, shows no changes in gp-130 levels and is persistently down-regulated because five times the protein needed to be loaded to observe a band. Interestingly, the IL-6–mediated effects on gp-130 were abolished in the α1B-AR cell line. With α1A- or α1D-AR activation, the addition of both epinephrine and IL-6 down-regulated gp-130 levels, but neither effect could be reversed by the addition of the IL-6 neutralizing antibody.
The combination of epinephrine and IL-6 together was also probed with the anti-phosphoSer727 antibody (Fig. 9). Although both α1A- and α1D-AR activation increased the phosphorylation of Ser727, α1B-AR activation did not, as shown in Fig. 4. The addition of IL-6 by itself also increased Ser727 phosphorylation as expected, but the combination of the two systems did not produce the synergism seen with Tyr705 phosphorylation.
Discussion
Although much effort has focused on the regulation of α1-AR subtype structure-function relationships and activation of common signaling pathways, no study has directly explored the comprehensive regulation of gene expression downstream of each subtype in the same cellular environment. This study represents the first microarray analysis of a GPCR in a transfected cell system. A common question raised in the GPCR field is whether each subtype can couple to different functions, or are they merely degenerate receptors? With the use of microarrays, we obtained the gene expression profiles generated by the α1-AR subtypes (α1A-, α1B-, and α1D-) (epinephrine-treated) stably expressed in Rat-1 fibroblasts, and compared them with those of control cells (epinephrine-treated, nontransfected Rat-1 fibroblasts). We found that of 7000 genes in the microarray, epinephrine modified the expression of a total of 47, 55, and 53 genes by at least 2-fold relative to nontransfected controls for the α1A-, α1B-, and α1D-ARs, respectively. Of these, a profile was established that was highly similar to the three α1-AR subtypes (38 genes; Tables 1 and 2). However, minor profiles that were unique to each α1-AR subtype (Table 3) and a profile (IL-6 signaling genes) activated by two but not all three subtypes (Table 4) were also produced. The changes in gene expression, which were invoked by a 1-h activation protocol, are likely to be different from profiles generated from sustained activation. We chose the 1-h stimulation time point because many second-messenger studies use this same condition.
However, a limitation of our system is the use of a separate parental cell line that lacks the α1-AR as the control instead of using each stably-transfected cell line as its own internal control of stimulated versus nonstimulated transcription. We did this for practical reasons. Therefore, with our experimental system, we cannot rule out gene expression changes that are caused by differences in cell line propagation. However, these should be rare events. Verification of the protein changes, however, used stimulated versus nonstimulated conditions within the same cell line. Along these same lines, we cannot rule out effects that are caused by subtype-specific but promiscuous couplings, because expression of the receptors is above physiological levels.
In these signaling studies, we used the endogenous α1-AR ligand epinephrine and Rat-1 fibroblasts, a cell model commonly used in α1-AR signaling studies (Garcia-Sainz et al., 1998; Chen et al., 1999), which we found to express similar densities of α1-AR subtypes. Besides gene expression, another measure of α1-AR function in Rat-1 fibroblasts included the activation of total IP accumulation by epinephrine. As expected, epinephrine stimulated the accumulation of IPs with higher efficacy in α1A-AR fibroblasts than the other two subtypes, consistent with previous studies of α1-AR subtype-stimulated second messenger activation (reviewed in Zhong and Minneman, 1999). Interestingly, the gene expression changes in common between the three α1-AR subtypes had similar fold changes, suggesting that the efficacy of the IP response does not influence the degree of expression changes. This could be because the IP effectors were saturated for all three subtypes, and the excess IP response may not be physiologically relevant. On the other hand, it may also suggest that a non-IP signaling molecule(s) may be responsible for the gene expression changes.
The largest profile generated by the α1-AR subtypes was characterized by gene expressions commonly activated or repressed by all three α1-AR subtypes (Tables 1and 2), suggesting that a certain level of redundancy among α1-AR subtypes exists. Several of these genes have been previously associated with α1-AR activation in primary cell lines and tissues. For example, increases in mRNA levels for c-fos, an immediate-early gene that was highly up-regulated in the present study, have been reported after α1-AR activation of cardiomyocytes (Deng et al., 1998), vascular smooth muscle cells (Okazaki et al., 1994), cerebral cortex (Shen and Gundlach, 2000), aortic rings (Carcillo and Hough, 1995), and hepatocytes (Im et al., 1998). α1-AR activation in cardiomyocytes also results in activation of c-jun and egr-1 (Iwaki et al., 1990; Jin et al., 2000), both up-regulated in the present study. α1-AR–mediated signaling in hepatocytes, renal tubular cells, and cardiomyocytes invoke protein tyrosine phosphatases and calcineurin (Aperia et al., 1992; Nguyen and Gao, 1999; Sugden, 2001). Furthermore, NF-1-X, a transcription factor up-regulated in this study, has been shown to bind to the α1B-AR promoter and regulate the expression of the α1B-AR gene (Gao and Kunos, 1998). Overall, evidence from the literature provides a certain measure of confirmation of the observed changes in the microarray and suggests that α1-AR signals in Rat-1 fibroblasts can extend to other cell types and tissues. However, as an independent test of confirmation, we measured by Northern blotting the expression levels of two genes that showed significant changes in expression in the microarray. We found that the expression profiles of both c-fos and p21-cK-Ras were similar to actual mRNA levels (Fig. 2), but the increased sensitivity of the microarray analysis does lead to higher fold-changes.
A few additional genes that were commonly changed in expression by α1-AR subtypes have never been associated with α1-AR signaling (Table 1). These code for proteins that are primary regulators in cAMP-mediated responses (cAMP response element modulator protein and C/EBP-like factor), angiogenesis (VEGF and glioma-derived VEGF), fatty acid metabolism (stearoyl-CoA-desaturase 2), and formation of extracellular matrix (fibronectin, collagen III, and collagenase).
The existence of gene expression changes that were unique to each α1-AR subtype suggests that nonredundant functions may exist (Table 3). One of the gene expressions that is the most robustly repressed in our study was synuclein (unique to the α1B-AR), which we have recently shown to be abnormally modified in transgenic mouse brains overexpressing this subtype (Papay et al., 2002). In addition, the modification of proapoptotic genes and those associated with neurodegeneration, including tau, synuclein, transforming growth factor β3, and caspase-6 by the α1B-AR subtype, is also supported by a neurodegenerative and apoptotic phenotype in the same transgenic mouse model (Zuscik et al., 2000), as well as a microarray analysis in transgenic brains (J. Yun, R. J. Gaivin, A. Boongird, Z. Ying, P. J. Gonzalez-Cabrera, R. S. Papay, I. Najm, and D. M. Perez, submitted), which contain caspase-3–mediated apoptosis, or in the heart, which contained terminal deoxynucleotidyl transferase dUTP nick-end labeling-positive myocytes (Yun et al., 2003).
Gene expressions can be grouped into a third profile in Rat-1 fibroblasts through the activation of two but not three subtypes (Table4) of the prototypic members of the IL-6 signaling pathway (STAT3 and gp-130). IL-6 is an endocrine cytokine that is secreted from fibroblasts, myocytes, and other cell types after inflammatory and stressful stimuli; its receptor mechanism includes the signal-transducing receptor gp-130. IL-6 signals through the JAK/STAT pathway induced by gp-130 (Taga and Kishimoto, 1997). Our results indicate that protein levels for gp-130 and STAT3 were differentially regulated by the three α1-AR subtypes. Recent evidence suggests that increased protein levels of STAT1 by itself, which is not phosphorylated, is sufficient to drive the transcription of a subset of genes (Chatterjee-Kishore et al., 2000a,b), suggesting that α1-AR–mediated increases in STAT3 protein may be physiologically relevant. Although the microarray studies successfully predicted changes in the protein levels for IL-6, gp130, and STAT3, the activational status for these proteins was more complicated. Our data showed that although all three subtypes stimulated the synthesis and secretion of IL-6 in Rat-1 fibroblasts (Fig. 3), only α1A- and α1D-AR subtypes increased both the protein level and the Ser727 phosphorylated form of STAT3 (Fig. 4), whereas α1B-AR did not. Figure 6 confirms that α1-AR activation of STAT3 could be direct; at the very least, however, it occurs through non–IL-6 effects because a neutralizing antibody to IL-6 could not block it. Direct effects are also suggested by the synergistic stimulation of Tyr705 phosphorylation by both epinephrine and IL-6. However, this synergistic effect with IL-6 was weak or nonexistent with α1B-AR activation. The synergistic activation of STAT3 may have implications in heart failure, where there are elevations in both circulating catecholamines and IL-6. However, the most interesting effect was the α1B-AR activation of Tyr705 phosphorylation (Fig. 7) but not of Ser727 phosphorylation (Figs. 4 and 9). There is precedent for the differential phosphorylation of STATs. IL-1 has been shown to induce the phosphorylation of STAT1 on Ser727 but not Tyr 701 (H. Nguyen, M. Chatterjee-Kishore, Z. Jiang, Y. Qing, C. V. Ramana, J. Bayes, M. Commane, X. Li, and G. R. Stark, submitted).
STAT3 is activated by tyrosine phosphorylation at Tyr705, which induces dimerization, nuclear translocation, and DNA binding (Darnell et al., 1994). Transcriptional activation seems to be stimulated by serine phosphorylation at Ser727, apparently through MAPK or mammalian target of rapamycin (mTOR) pathways (Wen et al., 1995; Yokogami et al., 2000). Both phosphorylation events are not coupled but are independent of one another, although both are thought to be involved in transcriptional processes to achieve maximal effects (Wen et al., 1995). In the same Rat-1 fibroblasts that we used here, it has been shown previously that the α1-AR subtypes can activate the MAPK pathways differentially with α1B-AR activation, resulting in increased p38 activity but not JNK activity, whereas the α1D-AR activated JNK but not p38 (Waldrop et al., 2002). In addition, α1B-AR–mediated inhibition of STAT3 in hepatocytes was found to be determined by p42/44 MAPK (Nguyen and Gao, 1999). These differences in MAPK activation could be the determining factor why the α1B-AR does not phosphorylate Ser727. Because the combination of epinephrine and IL-6 is not synergistic with Ser727 phosphorylation (Fig. 9), this also suggests that the two pathways converge at Ser727 phosphorylation, unlike Tyr705 phosphorylation. Nevertheless, the ability of the α1B-AR to suppress Ser727 phosphorylation is likely to produce differences in the transcriptional activity and eventual biological activity of STAT3.
Interestingly, whereas gp-130 protein levels were inhibited by long-term α1A- or α1D-AR stimulation (Fig. 5), they were constitutively down-regulated by the α1B-AR. Stimulation with IL-6 produced the opposite effect by increasing gp-130 protein levels (Fig. 8). The IL-6 neutralizing antibody could inhibit neither the epinephrine response nor the IL-6 response, suggesting that the IL-6–mediated changes in gp-130 protein levels are not a result of its biological function or signaling. However, the α1-AR–mediated effects on gp-130 could be blocked with prazosin (Fig. 5C). The combination of both epinephrine and IL-6 were again synergistic, displaying even greater changes in gp-130 protein levels than either alone, suggesting independent events. The nonsignaling mechanism of the IL-6–mediated changes in gp-130 protein levels and the lack of α1-AR–mediated IL-6 coupling, suggests that α1-AR- and IL-6–mediated effects on gp-130 protein are different. α1-AR–mediated changes are probably caused primarily by direct transcriptional events, as supported by the microarrays, and are not caused by activational paradigms with IL-6. However, the IL-6–mediated changes in gp-130 are probably caused by protein stability through complex formation (i.e., with IL-6/IL-6R/gp-130 trimer), which does not require IL-6 activation.
IL-6 is reported to increase in astrocytes and hepatocytes upon α1-AR stimulation (Norris and Benveniste, 1993;Jung et al., 2000), confirming that our microarray data can be translated to endogenous systems. In addition, the α1B-AR transgenic mouse model also constitutively down-regulated gp-130 protein levels in the heart (Yun et al., 2003), again confirming the fibroblast results. Confirmatory for the α1A-AR–mediated increases in STAT3, Ser727 phosphorylation is also seen in the normal rat myocyte (P. J. Gonzalez-Cabrera, J. Yun, B. R. Rorabaugh, D. F. McCune, and D. M. Perez, in preparation), which is composed of both the α1A- and α1B-AR subtypes, but heart function is thought to be mainly driven by the α1A-AR (Lin et al., 2001). α1-ARs as well as the Angiotension AT1 receptor have been previously linked to JAK/STAT activation (McWhinney et al., 1997; Zhong et al., 2000). Although there is controversy about whether this is caused by a direct coupling of the receptors to JAK/STAT, our data suggest at least a non–IL-6–mediated mechanism, because the IL-6–neutralizing antibody could not block STAT3 activation.
One of the most striking features of gene expression by α1-AR subtypes in the present study was the appearance of numerous genes with known roles in the processes relevant to cellular growth or hypertrophy. A major effect of α1-AR stimulation in the heart is the activation of growth-promoting pathways that lead to cardiac hypertrophy or vascular cell proliferation (Varma and Deng, 2000). Activation of all three α1-AR subtypes increased the transcription of three potent inducers of myocyte hypertrophy (IL-6, LIF, and calcineurin), whereas the α1B-AR seems to inhibit STAT3 Ser727 phosphorylation and possibly gp-130–mediated pathways. The inhibition of both STAT3 and gp-130 protein levels has been implicated in wall thickening, heart failure, and apoptosis (Hirota et al., 1999). This suggests that the α1B-AR may not be as potent an inducer of hypertrophy as the other two α1-AR subtypes. This was also suggested in Yun et al. (2003), in which many genes/proteins associated with hypertrophy were down-regulated in the heart by the α1B-AR in a transgenic mouse model. Although pharmacological studies have implicated the α1A-AR subtype in mediating the growth of rat myocytes (Rokosh et al., 1996), α1A-AR transgenic mice models have discarded this subtype as the sole mediator of hypertrophy (Lin et al., 2001;Rokosh and Simpson, 2002). Other studies have shown that overexpressed α1B-AR constitutively active mutants manifest a mild hypertrophic phenotype (Milano et al., 1994; Zuscik et al., 2001), whereas no evidence exists for the α1D-AR subtype in mediating any heart function (Tanoue et al., 2002). One study has even suggested that both the α1A- and α1B-AR subtypes need to be coactivated to achieve the previously reported robust hypertrophy phenotype seen in cultured cell models (McWhinney et al., 2000). Thus, although the subtypes may be differentially coupled to blood pressure homeostasis and induction of cardiac hypertrophy, it still seems that more than one α1-AR subtype might be needed to mediate the same function.
Recent evidence indicates that vascular fibroblasts that make up the adventitial layer in blood vessels express functional α1-AR subtypes (Faber et al., 2001; Zhang et al., 2002). These fibroblasts respond to catecholamines by activating growth and migratory pathways (vascular remodeling) as seen in vascular injury models. Here, too, gene expression modified by α1-ARs in Rat-1 fibroblasts may be part of a coordinated regulation of groups of genes (i.e., IL-6, collagenase, VEGF, and fibronectin) whose products act in the remodeling (apoptosis, migration, extracellular matrix changes) process of vessels. Another gene cluster modified by the α1-AR subtypes not reported here includes cell-cycle regulatory genes (i.e., cyclins, cyclin-dependent kinases, cyclin-dependent kinase inhibitors, etc.) that may be involved in differential α1-AR subtype coupling to proliferative responses. In this cluster, we observe that the α1B-AR subtype signals differently from α1A- and α1D-AR cells, which display identical profiles with respect to this cluster (P. J. Gonzalez-Cabrera, J. Yun, B. R. Rorabaugh, D. F. McCune, and D. M. Perez, in preparation).
In conclusion, we show an in-depth comprehensive comparison of the gene expression changes caused by the α1-AR subtypes. Although most of the gene expression changes were similar among the subtypes, some subtype-specific profiles were generated. This was verified at the protein and/or activational level for IL-6, gp-130, and STAT3 and suggests α1-AR subtype-dependent differential coupling to pathways that can mediate hypertrophy and vascular remodeling.
Acknowledgments
We thank Carley Gwinn of the Gene Expression Core of the CCF for help in the Affymetrix software analysis. We thank GlaxoSmithKline, Inc., for the generous gift of the Rat-1 fibroblasts that express the α1-AR subtypes. We also thank George Stark, Ph.D., for discussions on our data and insights into STAT3.
Footnotes
-
↵1 Present address: GlaxoSmithKline Inc., 5 Moore Drive, PO Box 13398, Research Triangle Park, NC 27709.
-
This work was funded by R01-HL61438 (to D.M.P.), a local American Heart Association fellowship to (to S.A.R. and P.J.G.-C.), an NRSA (to D.F.M.), and a T32-HL07914 training grant in Vascular Cell Biology (to B.R. and J.Y.).
- Abbreviations:
- AR
- adrenergic receptor
- GPCR
- G protein-coupled receptor
- IL
- interleukin
- IP
- inositol phosphate
- MAPK
- mitogen-activated protein kinase
- STAT
- signal transducer and activator of transcription
- 125I-BE-2254
- 2-[β-(4-hydroxy-3-[125I]iodophenyl)ethylaminomethyl]tetralone
- DMEM
- Dulbecco's modified essential medium
- PBS
- phosphate-buffered saline
- CCF
- Cleveland Clinic Foundation
- PM
- perfectly matched
- MM
- mismatched
- ELISA
- enzyme-linked immunosorbent assay
- JAK
- Janus tyrosine kinase
- LIF
- leukemia inhibitory factor
- VEGF
- vascular endothelial growth factor
- gp
- glycoprotein
- Received August 29, 2002.
- Accepted February 4, 2003.
- The American Society for Pharmacology and Experimental Therapeutics