Heterogeneity in β-Adrenergic Receptor Kinase Expression in the Lung Accounts for Cell-specific Desensitization of the β2-Adrenergic Receptor

The principal mechanism of homologous desensitization of the β-adrenergic receptor (β2AR) is phosphorylation of the receptor by the βAR kinase (βARK) or other closely related G protein-coupled receptor kinases (GRKs). However, within a single organ such as the lung where many cell types express the receptor, the presence or extent of β2AR desensitization in different cells has been noted to be highly variable. We hypothesized that such variability in desensitization is due to significant cell-type differences in βARK expression and/or function. To approach this, in situ hybridization was carried out in the lung and indeed revealed heterogeneity in βARK gene expression. Quantitative studies using ribonuclease protection assays with cell lines revealed that the level of βARK mRNA in airway smooth muscle cells was ∼20% of that in bronchial epithelial cells and ∼11% of that in mast cells (6.65 ± 0.96 versus 32.6 ± 4.0 and 60.7 ± 1.5 relative units, respectively, p < 0.001). βARK2 gene expression was not detected in any of these cells. At the protein level, βARK expression in airway smooth muscle cells was nearly undetectable, being ∼10-fold less than that expressed on mast cells. The activities of the GRKs in cell extracts were assessed in vitro by quantitating their ability to phosphorylate rhodopsin in the presence of light. Consistent with the gene and protein expression results, a marked discrepancy in activities was observed between extracts derived from mast cells (90.7 ± 0.5 relative units) as compared to airway smooth muscle cells (9.28 ± 0.6 relative units, p < 0.001). In contrast, the activities of protein kinase A (the other kinase that phosphorylates β2AR) in these extracts were not different. We predicted, then, that airway smooth muscle β2AR would undergo minimal short-term (5 min) agonist-promoted desensitization as compared to the β2AR expressed on mast cells. Mast cell cAMP reached maximal levels after 90 s and did not further increase over time, indicative of receptor desensitization in this cell. In contrast, cAMP levels of airway smooth muscle cells did not plateau, increasing at a rate of 103 ± 9% per min, consistent with little desensitization over the study period. We conclude that there is significant cell-type variation in expression of βARK and that such variation is directly related to the extent of short-term agonist-promoted desensitization of the β2AR.

The principal mechanism of homologous desensitization of the ␤-adrenergic receptor (␤ 2 AR) is phosphorylation of the receptor by the ␤AR kinase (␤ARK) or other closely related G protein-coupled receptor kinases (GRKs). However, within a single organ such as the lung where many cell types express the receptor, the presence or extent of ␤ 2 AR desensitization in different cells has been noted to be highly variable. We hypothesized that such variability in desensitization is due to significant cell-type differences in ␤ARK expression and/or function. To approach this, in situ hybridization was carried out in the lung and indeed revealed heterogeneity in ␤ARK gene expression. Quantitative studies using ribonuclease protection assays with cell lines revealed that the level of ␤ARK mRNA in airway smooth muscle cells was ϳ20% of that in bronchial epithelial cells and ϳ11% of that in mast cells (6.65 ؎ 0.96 versus 32.6 ؎ 4.0 and 60.7 ؎ 1.5 relative units, respectively, p < 0.001). ␤ARK2 gene expression was not detected in any of these cells. At the protein level, ␤ARK expression in airway smooth muscle cells was nearly undetectable, being ϳ10fold less than that expressed on mast cells. The activities of the GRKs in cell extracts were assessed in vitro by quantitating their ability to phosphorylate rhodopsin in the presence of light. Consistent with the gene and protein expression results, a marked discrepancy in activities was observed between extracts derived from mast cells (90.7 ؎ 0.5 relative units) as compared to airway smooth muscle cells (9.28 ؎ 0.6 relative units, p < 0.001). In contrast, the activities of protein kinase A (the other kinase that phosphorylates ␤ 2 AR) in these extracts were not different. We predicted, then, that airway smooth muscle ␤ 2 AR would undergo minimal short-term (5 min) agonist-promoted desensitization as compared to the ␤ 2 AR expressed on mast cells. Mast cell cAMP reached maximal levels after 90 s and did not further increase over time, indicative of receptor desensitization in this cell. In contrast, cAMP levels of airway smooth muscle cells did not plateau, increasing at a rate of 103 ؎ 9% per min, consistent with little desensitization over the study period. We conclude that there is significant cell-type variation in expression of ␤ARK and that such variation is directly related to the extent of short-term agonistpromoted desensitization of the ␤ 2 AR.
Many G protein-coupled receptors display a waning of signal transduction during continuous activation. This phenomenon, termed desensitization, is an important component in maintaining homeostasis under normal physiologic conditions, may contribute to or act to compensate in pathologic states, and may limit the effectiveness of therapeutic agonists (1)(2)(3). Of the G protein-coupled receptors, desensitization of the ␤ 2 -adrenergic receptor (␤ 2 AR) 1 has been one of the most extensively studied (2,3). Agonist-promoted desensitization of ␤ 2 AR has been demonstrated in in vitro reconstituted systems, a variety of naturally and recombinantly expressing cell lines, and in intact animals. The earliest component of agonist-promoted desensitization of the ␤ 2 AR is phosphorylation of the receptor by a cAMP independent kinase, termed the ␤AR kinase (␤ARK). 2 Such phosphorylation ultimately results in partial uncoupling of the agonist-occupied form of the receptor from the stimulatory guanine nucleotide-binding protein G s , thereby limiting receptor function. Over the past few years, it has become clear that ␤ARK is one of several related kinases that serve to phosphorylate the agonist-occupied forms of a number of G protein-coupled receptors (4,5). This family of kinases, termed G protein-coupled receptor kinases (GRKs) consist of the following mammalian isoforms: rhodopsin kinase (GRK1), ␤ARK (GRK2), ␤ARK2 (GRK3), GRK4 (initially termed IT-11), and two kinases denoted as GRK5 and GRK6 (6 -11). The potential for these other kinases to phosphorylate ␤ 2 AR has only been explored to a limited extent and their roles in agonist-promoted desensitization of the receptor at the cellular level are not well established. On the other hand, multiple lines of evidence (see "Discussion") have definitively shown that ␤ARK mediated phosphorylation represents a key process in homologous desensitization of the receptor.
While desensitization in the aforementioned model systems has been largely internally consistent, physiologic studies evaluating a variety of responses suggest that desensitization of ␤ 2 AR may not occur at all, or to the same extent, in different organs or tissues. For example, repetitive administration of ␤ 2 AR agonists to asthmatics appears to result in desensitization of responses thought to be mediated by the pulmonary mast cell ␤ 2 AR, but not the bronchodilatory response of ␤ 2 AR expressed on bronchial smooth muscle (Ref. 12 and reviewed in Ref. 13). One potential explanation for the apparent cell-type differences in ␤ 2 AR desensitization is a heterogeneity in the * This work was supported by National Institutes of Health Grants HL45967 and HL41496. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
expression of GRKs, such as ␤ARK, which play critical roles in the desensitization process. The possibility of cell-specific expression of ␤ARK within an organ has been largely unexplored, except for a single report in brain (14). That differences in ␤ARK expression can indeed alter ␤AR signal transduction has been demonstrated in recombinant cells overexpressing ␤ARK (15) and in transgenic mice overexpressing ␤ARK in the heart (16).
In the current study, we initially examined ␤ARK gene expression in the lung and found significant differences in expression among different cell types. This was then further explored by quantitating ␤ARK mRNA and protein expression in three human cell lines representing physiologically relevant functions in the lung: bronchial epithelial cells, mast cells, and airway smooth muscle cells. The BEAS-2B cell line was derived from normal human bronchial epithelial cells transformed by infection with Ad12-SV40 virus (17). BEAS-2B cells have the characteristics of epithelial cells by light and electron microscopy and stain positively for keratin (17). Furthermore, BEAS-2B cells have a physiologic number of ␤ 2 AR comparable to that of freshly dissociated human bronchial epithelial cells (18). The human mast cell line HMC-1 was derived from a patient with mast cell leukemia (19) and is the only established cell line that is phenotypically similar to normal human mast cells. On the basis of neutral protease contents (HMC-1 cells contain tryptase but not chymase) and other markers, HMC-1 cells resemble the MC T subset of human mast cells (20) which corresponds to the lung mast cell (21). The airway smooth muscle line is a primary culture of smooth muscle obtained at autopsy from an individual without lung disease. These cells maintain their morphologic characteristics over several passages (22). Marked differences in expression and activity of ␤ARK were indeed observed, particularly between mast cells and airway smooth muscle, which was correlated to the extent of agonist-promoted desensitization.
RNA Probes-A template for the synthesis of ␤ARK riboprobes was prepared by subcloning a 295-base pair EcoRI-SacI restriction fragment from the human ␤ARK cDNA (23) into pGEM 4Z (Promega). For ␤ARK2, an AvaI-HindIII restriction fragment from the bovine cDNA (8) was subcloned into the same sites of pGEM 4Z. Orientation of subcloned fragments was confirmed by restriction analysis and dideoxy sequencing. Plasmids were linearized with the appropriate restriction enzyme, and in vitro transcription reactions were carried out with either T7 or SP6 RNA polymerase to generate antisense or sense RNA probes. The probes were labeled with 35 S-UTP for in situ hybridization experiments and [ 32 P]UTP for ribonuclease protection assays.
In Situ Hybridization-Frozen sections from cryoprotected, paraformaldehyde-fixed monkey lungs were subjected to in situ hybridization (24). Cryosections were rehydrated, treated with proteinase K, fixed with 4% paraformaldehyde, acetylated with acetic anhydride, and dehydrated. Sections were then incubated overnight at 52°C with hybridization solution that contained 5-7 ϫ 10 4 cpm/l of sense or antisense ␤ARK riboprobe labeled with 35 S-UTP. The sections were washed under high stringency conditions, treated with RNase A, subjected to a second high stringency wash, and dehydrated. Dried slides were dipped in Kodak NTB-2 autoradiographic emulsion and exposed at 4°C for 8 weeks.
RNA Analysis-Expression of ␤ARK and ␤ARK2 mRNA was measured by ribonuclease protection assays (25). Total cellular RNA was isolated from BEAS-2B, HASM, and HMC-1 cells by the rapid acid guanidinium thiocyanate-phenol-chloroform method (26). Ribonuclease protection assays were performed as previously reported (27) using 20 g of total RNA and the riboprobes described above. The reaction products were separated by electrophoresis on 6% polyacrylamide gels containing 8 M urea. Radiographic bands were visualized with a Phos-phorImager (Molecular Dynamics) and conventional autoradiography. Quantitation of the phosphorimage was performed with the Image-Quant software package (Molecular Dynamics).
Western Analysis-The detection of ␤ARK protein in whole cell lysates was carried out by standard SDS-PAGE and immunoblotting techniques as described previously (28). The cells were washed with phosphate-buffered saline and homogenized in hypotonic lysis buffer (5 mM Tris, pH 7.4, 2 mM EDTA, containing the protease inhibitors aprotinin (10 g/ml), benzamidine (5 g/ml), and soybean trypsin inhibitor (5 g/ml)). The homogenates were diluted with 2 ϫ Laemmli sample buffer, subjected to 10% SDS-PAGE, and electroblotted onto a nitrocellulose membrane (Protran, Schleicher & Schull). The membranes were blocked with 5% nonfat dry milk in Tris saline buffer (50 mM Tris, pH 7.4, 200 mM NaCl) containing 0.1% Tween, and then incubated with a monoclonal antibody raised against purified ␤ARK diluted 1:200 in the blocking buffer. After washing with Tris saline/Tween buffer, the filters were incubated with an anti-mouse horseradish peroxidase-conjugated second antibody and developed using enhanced chemiluminescence (DuPont NEN). The radiographic film was scanned and quantitated using Scan Analysis (Biosoft).
Bovine Rod Outer Segments (ROS) Phosphorylation Assay-Cytosolic ␤ARK activity was measured by phosphorylation of rhodopsin derived from rod outer segments (7). Urea-treated ROS were prepared from dark-adapted calf retinas by stepwise sucrose gradient centrifugation (29). The ROS consisted of ϳ90% rhodopsin as assessed by Coomassie Blue staining and had no significant endogenous kinase activity. To prepare cell lysates, HASM and HMC-1 cells were homogenized in 10 mM Tris, pH 7.4, 5 mM EDTA buffer containing the aforementioned protease inhibitors. The homogenates were centrifuged at 100,000 ϫ g for 30 min, and the supernatants were concentrated with a Centricon 10 cartridge (Amicon). For each phosphorylation reaction, 15 g of cytosolic proteins were incubated with 85 pmol of rhodopsin in buffer containing 20 mM Tris, pH 7.4, 2 mM EDTA, 6 mM MgCl 2 , and 100 M [␥-32 P]ATP (ϳ2,000 cpm/pmol). Incubations were carried out in the presence of light for 30 min at 30°C. Reactions were stopped by the addition of 1 ml of cold 100 mM sodium phosphate, pH 7.0, 5 mM EDTA buffer and centrifuged at 100,000 ϫ g for 30 min. The pellets were resuspended in Lammeli sample buffer and subjected to 10% SDS-PAGE. Dried gels were exposed on a PhosphorImager for quantitation. Light dependence of the phosphorylation reaction was confirmed using purified ␤ARK as a control (data not shown). One micromolar of the ␤ARK inhibitor heparin (30), 1 M of a protein kinase A inhibitor peptide (31), or 1 M of a protein kinase C inhibitor peptide (32) were added to some reactions to confirm that rhodopsin phosphorylation was ␤ARK-dependent.
PKA Activity-An assay that measures the phosphorylation of Kemptide (33), a peptide substrate for PKA, was used to determine the phosphotransferase activity of PKA in extracts from HASM and HMC-1 cells. Cell lysates, prepared as described above, were incubated in a reaction mixture that contained 20 mM MOPS, pH 7.2, 25 mM ␤-glycerol phosphate, 5 mM EGTA, 1 mM sodium vanadate, 15 mM MgCl 2 , 125 M [␥-32 P]ATP (ϳ4,000 cpm/pmol), and 125 M Kemptide for 10 min at 30°C. The reactions were stopped by spotting the assay mixture onto P81 phosphocellulose paper. The filters were washed three times with 0.75% phosphoric acid and once with acetone. Bound radioactivity was measured by liquid scintillation counting. PKA activity was defined as the amount of phosphate incorporated in the presence of 5 M PKC inhibitor peptide (32).
Radioligand Binding-Membranes from HASM and HMC-1 cells were prepared and radioligand binding carried out as described previously (22,34). Briefly, cells were washed three times with cold phosphate-buffered saline, resuspended in hypotonic lysis buffer (5 mM Tris, 2 mM EDTA, pH 7.4) containing the aforementioned protease inhibitors, disrupted with a Polytron (Brinkman) homogenizer, and centrifuged at 40,000 ϫ g for 10 min at 4°C. The resulting pellets were resuspended in 10 volumes of lysis buffer, centrifuged again, and resuspended in assay buffer (75 mM Tris, 12.5 mM MgCl 2 , 2 mM EDTA, pH 7.4). To determine total receptor density, membranes were incubated in a total volume of 250 l at room temperature for 120 min with concentrations of 125 I-CYP ranging from 3.125 to 400 pM. Nonspecific binding was determined in the presence of 1 M propranolol. Assays were stopped by dilution with cold wash buffer (10 mM Tris, pH 7.4) and vacuum filtra-tion through Whatmann GF/C glass fiber filters. The bound radioactivity was measured with a ␥-counter. Competition experiments with 40 pM 125 I-CYP and varying concentrations of ICI 118,551 or CGP 20712A were also performed using the conditions described above to determine the proportion ␤ 2 AR as compared to other ␤AR subtypes expressed on the membranes. Protein concentration was determined by the copperbicinchoninic acid method (35) with bovine serum albumin used as the standard. Data from the saturation binding and competition experiments were analyzed by nonlinear least squares techniques using the Prism software program (GraphPad). Curves were modeled to a one-site fit unless the two-site fit was significantly better (p Ͻ 0.5 by F test).
cAMP Assays-cAMP content of HASM and HMC-1 cells was measured by an acetylated radioimmunoassay method. All cells were treated with 0.1 mM isobutylmethylxanthine for 30 min at 37°C, washed, and incubated in 500 l of serum-free media at 37°C in the presence or absence of 1 M isoproterenol or 100 M forskolin for the indicated times. Reactions were stopped every 30 s by the addition of 50 l of 1.0 M HCl. The cAMP formed was acetylated and measured by radioimmunoassay using a polyclonal anti-cAMP antibody as described (36). Results were expressed as a percentage of the nonstimulated values.
Materials-Tissue culture supplies were purchased from JRH Biosciences. Radioisotopes were from DuPont NEN. ICI 118,551 was purchased from Research Biochemicals Int. CGP 20712A was a gift from Ciba Geigy. Heparin was from Sigma. PKA and PKC inhibitor peptides and Kemptide were from Upstate Biotechnology Inc. HMC-1 cells were obtained from J. Butterfield, Mayo Clinic, Rochester, MN. BEAS-2B cells were provided by C. Harris, National Institutes of Health, Bethesda, MD. GRK cDNAs, purified ␤ARK, and the ␤ARK antibody were provided by J. Benovic, Thomas Jefferson University, Philadelphia, PA.

RESULTS
To examine the possibility that ␤ARK is differentially expressed among different cell types within the lung, we performed in situ hybridization experiments using cryosections of monkey lung (Fig. 1). Panel A, a phase-contrast micrograph depicts the cellular architecture in a region of a moderate sized bronchus of the lung with some adjacent alveoli. Hybridization with the antisense ␤ARK probe gave a specific signal that was greatest over bronchial epithelial cells and cells lining the alveolar space (Fig. 1B). A minimal signal was also present over bronchial smooth muscle. No specific signal was detected when sections were hybridized with the sense probe (data not shown).
To assess ␤ARK expression in a more quantitative fashion, we performed ribonuclease protection assays using RNA from three human cell types that are physiologically relevant to lung function and are targets for therapeutic ␤-agonists: airway epithelial cells (BEAS-2B), airway smooth muscle cells (HASM), and mast cells (HMC-1). Although a band corresponding to the 295-base pair fragment expected for ␤ARK mRNA could be detected in all three cell types, there was a substantial difference in the level of expression among the different cell types ( Fig. 2A). Analysis of the gels showed that ␤ARK mRNA content in HASM cells (6.65 Ϯ 0.96 relative units) was only ϳ11% of that in HMC-1 cells (60.7 Ϯ 1.5 relative units, n ϭ 4, p Ͻ 0.001) and ϳ17% of that in epithelial cells (32.6 Ϯ 4.1 relative units, n ϭ 4, p Ͻ 0.001). We also used ribonuclease protection assays to assess ␤ARK2 gene expression in these cell lines. Although a strong band was observed in the positive control, no significant signal for ␤ARK2 was detected in BEAS-2B, HASM, or HMC-1 cells (Fig. 2B). We subsequently focused our experiments on the role of ␤ARK in regulating ␤ 2 AR desensitization in these cell types.
We next measured ␤ARK protein content by Western blotting. A distinct band with a molecular mass of ϳ80 kDa corresponding to ␤ARK was detected in BEAS-2B, HASM, and HMC-1 cell extracts (Fig. 3). ␤ARK levels were lowest in HASM cells (6.2 Ϯ 1.3 relative units), as compared to that of HMC-1 cells (56.5 Ϯ 3.5 relative units, n ϭ 3, p Ͻ 0.001) and that of BEAS-2B cells (37.3 Ϯ 5.0 relative units, n ϭ 3, p Ͻ 0.001). Thus ␤ARK protein expression closely paralled mRNA levels in the three subtypes, with HASM cells expressing about onetenth the level of ␤ARK as compared to HMC-1 cells and one-fifth of that of BEAS-2B cells.
To confirm the above observations and to determine if increased ␤ARK content resulted in increased kinase activity in vitro, we measured the ability of extracts from HASM and HMC-1 cells to phosphorylate ROS (Fig. 4). Initial studies were carried out to assure the specificity of the reactions for assessing GRK-mediated phosphorylation by the use of purified bovine ␤ARK and various inhibitors. Using purified ␤ARK, light dependence of ROS phosphorylation was demonstrated (data not shown). ␤ARK-dependent phosphorylation was then assessed by carrying out reactions in the presence of heparin (1 M), a PKA inhibitor peptide (1 M), or a PKC inhibitor peptide (1 M). ROS phosphorylation was blocked by the ␤ARK inhibitor heparin in both cell types, whereas inhibitors of PKC and PKA had no effect (Fig. 4A). Using this in vitro assay, we determined the activities of extracts from both cell types (Fig.  4B). As can be seen, there were marked differences in activities, with the kinase activity of HMC-1 cells being nearly 10fold greater (90.7 Ϯ 0.6 relative units) than that of HASM cells (9.2 Ϯ 0.6 relative units, n ϭ 4, p Ͻ 0.001). We also assessed protein kinase A activities in these extracts since this kinase also phosphorylates ␤ 2 AR in response to elevated cAMP, and any cell-type differences might confound interpretation of functional desensitization studies. In contrast to the marked differences in ␤ARK activities between the two cell types, PKA activities from extracts of HASM and HMC-1 cells were not different (Fig. 5).
The results of the above experiments clearly showed that ␤ARK was differentially expressed among lung cells, and that increased ␤ARK content was associated with increased activity in vitro. We speculated that ␤ 2 AR from cells with the highest levels of ␤ARK activity might be most subject to desensitization. We thus compared short-term desensitization in HASM cells, which had low levels of ␤ARK, to HMC-1 cells which we found to have substantially higher levels of ␤ARK. We have previously shown that the ␤AR of the HASM cells consists entirely of the ␤ 2 AR subtype (22). The ␤AR of HMC-1 cells has not been previously characterized. Briefly, we found that the ␤AR radioligand 125 I-CYP bound to a population with high affinity (22.7 Ϯ 1.8 pM, n ϭ 3). In competition studies with subtype-specific antagonists, these cells were found to express a single population of ␤AR with a high affinity for ICI 118,551, consistent with this receptor being of the ␤ 2 AR subtype. Expression of the ␤ 2 AR in these HMC-1 cells as measured in saturation binding experiments was 8.5 Ϯ 1.3 fmol/mg of protein (n ϭ 3). We then assessed agonist-promoted desensitization in HASM and HMC-1 cells using a previously described intact cell paradigm, where the kinetics of cAMP accumulation were determined following exposure to 1 M isoproterenol (36). cAMP levels were determined every 30 s for 5 min. In HMC-1 cells, cAMP levels initially increased but reached a maximum within 90 s, and the levels remained relatively constant (rate ϭ Ϫ7.9 Ϯ 4.3% per min) for the remaining 210 s, indicative of agonist promoted desensitization of the ␤ 2 AR on this cell (Fig.  6). In marked contrast, HASM cell cAMP continued to increase at a rate of 103 Ϯ 9.7%/min throughout the course of the study, reflective of substantially less desensitization of these receptors as compared to those of HMC-1 cells (Fig. 6). Over a similar time period, the cAMP responses to forskolin were found to be linear for both HMC-1 cells (r 2 ϭ 0.93 Ϯ 0.02, n ϭ 3) and bronchial smooth muscle cells (r 2 ϭ 0.97 Ϯ 0.01, n ϭ 3), pointing toward the desensitization of the isoproterenol response observed in the former cells to be receptor specific. DISCUSSION During continuous exposure of ␤ 2 AR to agonist, a number of regulatory events occur which act to limit the cellular responsiveness. The most rapid process (seconds to minutes) is phosphorylation of the receptor by ␤ARK (and potentially other GRKs) leading to the binding of an arrestin-like moiety (termed ␤-arrestin) which results in depressed coupling to G s (2,3). Phosphorylation of the receptor also occurs via protein kinase A, whenever intracellular cAMP is increased due to receptor activation by agonist, or by other means. After more prolonged agonist exposure, an internalization of receptors occurs which results in a loss of some proportion of cell surface receptors. This process, termed sequestration, has also been considered to be another mechanism of desensitization, but recent studies have suggested that its major role in short-term regulation of the receptor may be in resensitization (37,38), since it appears that the sequestered pool is the site of dephosphorylation of the receptor. After hours of agonist exposure, a net loss of cellular receptors occurs (denoted down-regulation) via several mechanisms that are independent of receptor phosphorylation.
The role of ␤ARK-mediated phosphorylation of ␤ 2 AR in short-term agonist-promoted desensitization has been elucidated using multiple approaches (15, 16, 36, 39 -42). In recombinant cell lines, mutated ␤ 2 AR lacking ␤ARK phosphorylation sites display attenuated agonist-promoted desensitization as assessed in intact cell (36) and membrane (39) based assays. In addition, treatment of permeabilzed cells with the ␤ARK inhibitor heparin results in a loss of receptor desensitization and Ribonuclease protection assays were performed using 20 g of total RNA and the described ␤ARK or ␤ARK2 antisense probe. A, a representative autoradiogram using the ␤ARK antisense probe. Full-length probe is on the left. ␤ARK mRNA was detected in all three cell lines but the signal intensity from HASM cells was significantly less than that of HMC-1 and BEAS-2B cells. The data from three independent experiments are summarized in the bar graph (mean Ϯ S.E.). Relative units represent the pixel density of each sample from the phosphorimage expressed as a percentage of the total pixel density for all samples on the gel. *, p Ͻ 0.001 compared to HASM cells. B, a representative autoradiogram using the ␤ARK2 antisense probe. Full-length, undigested probe is on the left. A strong signal was noted from a sample of mRNA from COS-7 cells transfected with an expression plasmid encoding bovine ␤ARK2, which was used as a positive control. However, no significant signal was observed in BEAS-2B, HASM, or HMC-1 cells.
phosphorylation (40,41), and expression of a dominant-negative ␤ARK in cells that natively express ␤ 2 AR inhibits agonistpromoted desensitization (42). Overexpression of ␤ARK has been found to enhance agonist-promoted desensitization and phosphorylation in Chinese hamster ovary cells overexpressing ␤ 2 AR (15). While in vivo agonist-promoted desensitization was not assessed per se, ␤AR of cardiac membranes from transgenic mice expressing a dominant-negative ␤ARK display increased coupling, while receptors in transgenic mice overexpressing ␤ARK display decreased coupling (16). Although some in vitro studies have been carried out with the ␤ 2 AR and other known GRKs, little is known regarding their potential for mediating desensitization of the receptor as assessed in studies such as discussed above.
While a loss of signaling via G protein-coupled receptors during continuous exposure to agonist has been observed with many members of the superfamily, some receptors do not demonstrate the phenomenon, including those that share the same endogenous agonist and signal transduction pathways. For example, the ␤ 2 AR undergoes rapid agonist-promoted desensitization, but the ␤ 3 AR appears to be relatively resistant to such regulation (43). Similarly, the human ␣ 2A AR displays desensitization after brief exposure to agonist, while the human ␣ 2C subtype does not (44,45). In both of the above instances, the lack of desensitization is paralleled by a lack of phosphorylation by GRKs. Mutagenesis studies have delineated the structural determinants within the intracellular regions that define GRK phosphorylation sites within these, and other, receptors (39,43,46,47). Thus one way in which cell-specific desensitization of agonist responsiveness occurs is by selective expression of certain receptor subtypes. Another potential component which may dictate the presence or absence of short-term desensitization by agonist is the level of GRKs expressed in a given cell. We considered that if such heterogeneity of GRK expression was indeed present, the responsiveness of cell-specific signaling to agonist might differ markedly in an organ populated by multiple cell types even though the same receptor subtype is present on the cells. Whether differences in GRK gene expression occur in cell types of a given organ, and whether such correlate with differences in protein expression, kinase activity, and agonist-promoted desensitization of the ␤ 2 AR has not been explored.
We approached this issue by examining ␤ARK gene and protein expression, kinase activity, and ␤ 2 AR signal transduction in lung cells. We utilized lung for several reasons. First, this organ has a large number of different cell types, many expressing exclusively the ␤ 2 AR as compared to other ␤AR subtypes. Second, in vivo the ␤ 2 AR of these different cell types have clearly defined physiologic functions: bronchial epithelial cell receptors regulate ciliary beat frequency, airway smooth muscle cell receptors regulate relaxation of bronchial smooth muscle, and mast cell receptors regulate inflammatory mediator release (13). These functions have provided for distinct signals in the assessment of the relevance of ␤ 2 AR desensiti- FIG. 3. Western analysis of ␤ARK expression in BEAS-2B, HASM, and HMC-1 cells. 20 g of protein isolated from whole cells were subjected to Western analysis using a monoclonal antibody directed against ␤ARK. A shows a representative blot. The signal in HASM cells is faint compared to that for BEAS-2B and HMC-1 cells. The data from three independent experiments (mean Ϯ S.E.) are summarized in B. Relative pixel density was derived from the analysis of scanned autoradiograms as described for ribonuclease protection assays. *, p Ͻ 0.001 compared to HASM cells.
FIG. 4. In vitro assessment of ␤ARK activity in HASM and HMC-1 cells. ROS phosphorylation assays were performed using cytosolic proteins (15 g) from HASM and HMC-1 cells. A shows that phosphorylation of ROS by both cell types was inhibited by heparin, an inhibitor of ␤ARK, whereas inhibitors of PKC and PKA had no effect. B, a representative autoradiogram illustrating the marked difference in activity between HASM and HMC-1 cells. Purified ␤ARK (10 ng) was used as a positive control. C, summary of data from four experiments comparing activities between HASM and HMC-1 cells (mean Ϯ S.E.). Quantitative analysis of gels was performed with a PhosphorImager as described under "Experimental Procedures." *, p Ͻ 0.001 compared to HASM cells. zation (or lack thereof) in these cell types in vivo. Finally, ␤ 2 AR agonists delivered by inhalation are used for the treatment of bronchospasm, which has afforded others the opportunity for studying desensitization in a physiologically relevant setting.
Our initial approach was to screen for gene expression of ␤ARK in the lung using in situ hybridization. These results indeed showed a paucity of expression in smooth muscle as compared to epithelial cells. This difference is not simply a reflection of differences in ␤ 2 AR expression, since the receptor densities are similar between the two cell types (22,48). We then explored quantitatively the expression of ␤ARK, and the closely related isoform ␤ARK2, in three cell lines. For ␤ARK, airway smooth muscle mRNA was clearly less than that of mast cells and bronchial epithelial cells. (␤ARK2 transcripts could not be detected in any of the three lines.) Consistent with these results, expression of ␤ARK as assessed using Western blots indicated a significant difference in expression in the cell lines, with mast cell Ͼ epithelial cell Ͼ ϾϾ smooth muscle cell. We then further studied the two cell types with the greatest difference in ␤ARK expression, the mast cell and the smooth muscle cell. To assess the activity of ␤ARK (and potentially other GRKs) in the two cell lines, extracts were used to phosphorylate rhodopsin in ROS. These studies revealed ϳ10% activity in smooth muscle cells as compared to mast cells. These results were remarkably similar to those assessing ␤ARK expression, where smooth muscle cells were found to express ␤ARK at only ϳ11% of that in mast cells. Based on our hypothesis, then, we expected that the two cell types would exhibit differences in the extent of agonist-promoted desensitization. To approach this, cells were exposed to a saturating concentration (1 M) of isoproterenol, and cAMP accumulation determined every 30 s for 5 min. Such an approach has been verified with mutant ␤ 2 AR lacking phosphorylation sites (36) and with a dominant-negative ␤ARK (42). As shown in Fig. 6, the smooth muscle cells continued to accumulate cAMP over time during exposure to agonist, consistent with little desensitization. In contrast, cAMP accumulation in mast cells had a rapid onset, but the accumulation rate markedly decreased after ϳ90 s. As described previously (36), this plateau in cAMP levels represents receptor desensitization, which is clearly evident in these cells but not smooth muscle cells. The difference in susceptibility to ␤ 2 AR desensitization did not appear to be due to differences in PKA, as the activity of this kinase was similar in both cell types.
Thus, we have demonstrated that there is significant variation in the expression of ␤ARK between cell types of the lung, and that this variation correlates with the extent of agonistpromoted desensitization. This implies, that unlike some components involved in ␤ 2 AR regulation in which their levels may be in excess, physiologic variation in the level of ␤ARK may have a distinct effect on signal transduction. These findings also support functional studies in man examining in vivo desensitization of pulmonary ␤ 2 AR during treatment with agonists in asthma. Most studies have observed desensitization of the so-called bronchoprotective effects of agonists (implying desensitization of mast cell receptors) while desensitization of the bronchodilatory effects (mediated by smooth muscle receptors) appears to be substantially less, if at all, present (12). Our findings also imply that pathologic conditions that result in changes in ␤ARK levels may indeed result in alterations of receptor function. For example, in chronic heart failure, ␤ARK mRNA levels are elevated (49), which is consistent with the impaired agonist-mediated receptor function in the heart that is observed in this disease. In addition, opiod-dependence is associated with increases in ␤ARK in the locus ceruleus (50). The extents of the changes in ␤ARK that have been reported in these types of studies appear to be of sufficient magnitude, based on our current results, to be potentially relevant.
Taken together, then, these studies indicate that the basis of cell-type specificity of agonist-promoted desensitization of the ␤ 2 AR in natively expressing, physiologically relevant cells, can in part be ascribed to the level of expression of ␤ARK. Furthermore, these studies reveal that ␤ARK expression is indeed highly variable among different cell-types, and supports the concept that dynamic regulation of the kinase can significantly alter ␤AR signal transduction. Finally, given that ␤ARK phosphorylates other G protein-coupled receptors, it is likely that cellular variation in its expression is important to desensitization of other receptors as well. Following treatment with 1 mM isobutylmethylxanthine for 30 min, cells were incubated in 0.5 ml of serum-free media containing 0.1 mM ascorbic acid, 1 mM isobutylmethylxanthine, and 1 M isoproterenol. Reactions were stopped at the indicated times by addition of HCl, and cAMP was measured by radioimmunoassay. As is shown, after 90 s of agonist exposure cAMP levels of HMC-1 cells plateaued (rate ϭ Ϫ7.9 Ϯ 4.3% per min from 90 s to 300 s), indicative of receptor desensitization. In contrast, cAMP levels continued to increase in HASM cells (rate ϭ 103 Ϯ 9.7% per min) consistent with little desensitization over the time period studied. Data shown are the mean Ϯ S.E. from four independent experiments each performed in duplicate.