Visualization of Agonist-induced Sequestration and Down-regulation of a Green Fluorescent Protein-tagged β2-Adrenergic Receptor*

To date, the visualization of β2-adrenergic receptor (β2AR) trafficking has been largely limited to immunocytochemical analyses of acute internalization events of epitope-tagged receptors in various transfection systems. The development of a β2AR conjugated with green fluorescent protein (β2AR-GFP) provides the opportunity for a more extensive optical analysis of β2AR sequestration, down-regulation, and recycling in cells. Here we demonstrate that stable expression of β2AR-GFP in HeLa cells enables a detailed temporal and spatial analysis of these events. Time-dependent colocalization of β2AR-GFP with rhodamine-labeled transferrin and rhodamine-labeled dextran following agonist exposure demonstrates receptor distribution to early endosomes (sequestration) and lysosomes (down-regulation), respectively. The observed temporal distribution of β2AR-GFP was consistent with measures of receptor sequestration and down-regulation generated by radioligand-receptor binding assays. Cells stimulated with different β-agonists revealed time courses of β2AR-GFP redistribution reflective of the intrinsic activity of each agonist.

To date, the visualization of ␤ 2 -adrenergic receptor (␤ 2 AR) trafficking has been largely limited to immunocytochemical analyses of acute internalization events of epitope-tagged receptors in various transfection systems. The development of a ␤ 2 AR conjugated with green fluorescent protein (␤ 2 AR-GFP) provides the opportunity for a more extensive optical analysis of ␤ 2 AR sequestration, down-regulation, and recycling in cells.
Here we demonstrate that stable expression of ␤ 2 AR-GFP in HeLa cells enables a detailed temporal and spatial analysis of these events. Time-dependent colocalization of ␤ 2 AR-GFP with rhodamine-labeled transferrin and rhodamine-labeled dextran following agonist exposure demonstrates receptor distribution to early endosomes (sequestration) and lysosomes (down-regulation), respectively. The observed temporal distribution of ␤ 2 AR-GFP was consistent with measures of receptor sequestration and down-regulation generated by radioligand-receptor binding assays. Cells stimulated with different ␤-agonists revealed time courses of ␤ 2 AR-GFP redistribution reflective of the intrinsic activity of each agonist.
Upon agonist stimulation, ␤ 2 -adrenergic receptors (␤ 2 ARs) 1 are rapidly desensitized by receptor phosphorylation (1). Receptor phosphorylation by G protein-coupled receptor kinases and subsequent binding of non-visual arrestins initiates the internalization of ␤ 2 ARs via clathrin-coated pits (2,3). Previous studies have demonstrated that internalized receptors have multiple potential fates. One such fate is the recycling of internalized receptors to the plasma membrane, presumably completing an ill defined "resensitization " process involving ␤ 2 AR dephosphorylation in an endosomal compartment (4,5). Depending upon the duration of agonist exposure, internalized ␤ 2 ARs may ultimately appear "lost" or destroyed (undetectable by radioligand binding), ostensibly trafficking to lysosomes where they are degraded. Previously, the analysis of these events has been heavily dependent on biochemical and pharmacological approaches.
Agonist-mediated subcellular redistribution of ␤ 2 ARs was initially inferred from ligand binding studies (6,7) and the coincident migration of ␤ 2 ARs with enzyme markers (8,9) or epidermal growth factor (10,11) into subcellular fractions resolved through centrifugation. However, direct visualization of ␤ 2 AR trafficking events remained lacking until von Zastrow and Kobilka (12) provided immunocytochemical evidence of rapid, agonist-induced redistribution of epitope-tagged ␤ 2 ARs into small, punctate accumulations within the cytoplasm. The time course of ␤ 2 AR redistribution assessed by confocal microscopy paralleled that of ␤ 2 AR sequestration measured by radioligand binding. Importantly, internalized ␤ 2 ARs colocalized with transferrin receptors, suggesting that sequestered ␤ 2 ARs undergo processing through endosomal compartments in a manner similar to that observed for constitutively internalized receptors.
Advances in the development of proteins conjugated with green fluorescent protein (GFP) have since provided the opportunity for real time optical analysis of protein trafficking events in individual cells (13)(14)(15). Green fluorescent protein is a naturally occurring protein isolated from several different species of jellyfish (Aequoria) and sea cucumbers (Renilla). When expressed in cells, proteins conjugated with GFP may be visualized with routine fluorescent microscopy without the need to fix cells. Recently, Barak et al. (16) established the utility of a ␤ 2 AR-GFP fusion protein in visualizing agonistmediated ␤ 2 AR internalization in HEK293 cells transiently expressing the construct. This study also demonstrated that ␤ 2 AR-GFP is fully functional with regard to ligand binding and adenylyl cyclase stimulation and that it can undergo agonistdependent phosphorylation and sequestration. Here we demonstrate that stable expression of ␤ 2 AR-GFP in HeLa cells enables a detailed temporal and spatial analysis of ␤ 2 AR trafficking events associated with receptor sequestration, downregulation, and recycling. Colocalization of the ␤ 2 AR-GFP with rhodamine-labeled transferrin and rhodamine-labeled dextran during agonist treatment revealed sequential localization of receptor in cellular compartments containing these compounds. Moreover, this system appears capable of distinguishing the differing effects of various ␤-agonists (including the highly hydrophobic ligand salmeterol) on ␤ 2 AR trafficking, overcoming some of the limitations in previous analyses of these compounds.

EXPERIMENTAL PROCEDURES
Construction of a ␤ 2 AR-GFP Fusion Expression Construct-To create a Flag-tagged ␤ 2 AR-GFP fusion protein, PCR was used to amplify both the ␤ 2 AR and Aequoria victoria GFP-S65T. An amino-terminal ␤ 2 AR primer, TGCGCGCCATGGGGCAAC, was combined with a primer designed against the carboxyl terminus, GCTCTAGACAGCAGTGAGT-CATTTGT, in which the stop codon is replaced by an XbaI site that encodes two extra amino acids, serine and arginine. Similarly, primers were designed to amplify GFP from the phGFP-S65T template (CLON-TECH) with an XbaI site at the NH 2 terminus, GCTCTAGAATGGT-GAGCAAGGGCGAG, and a SalI site at the COOH terminus, AAGCT-TGTCGACTTACTTGTACAGCTCGTC. The two PCR products were cut with NcoI/XbaI for the ␤ 2 AR or XbaI/SalI for the GFP and subcloned into NcoI/SalI digested pBCflag␤ 2 AR (provided by B. Kobilka). An NcoI/NcoI fragment of the pBC backbone was recloned in, and to circumvent sequencing the entire open reading frame of the ␤ 2 AR, an NcoI/EcoRV region was replaced with that portion of the original pBCflag␤ 2 AR construct. The final ligated product was sequenced through the regions that were generated by PCR. This construct was further modified by exchanging a small HindIII/NcoI fragment containing the region encoding the flag epitope with an identical fragment generated by PCR that contains a Kozak consensus sequence for translation initiation (ACCATGG). For the present studies, the region encoding the entire Flag-tagged ␤ 2 AR-GFP fusion was removed with HindIII/SalI and cloned into the BamHI site of pcDNA3 (Invitrogen) using BamHI linkers. This final construct, pcDNA3-␤ 2 AR-GFP, was used in transient transfections and for making the stable HeLa cell line. A Flag-tagged ␤ 2 AR construct in pcDNA3 was created by digesting pBCflag␤ 2 AR with SalI followed by Klenow to blunt the 3Ј end. The insert was excised with HindIII and the ϳ2-kilobase fragment was ligated into HindIII/EcoRV digested pcDNA3.
Transient and Stable Transfection of HeLa Cells-HeLa cells were maintained in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10 -15% fetal bovine serum, 100 units/ml penicillin G, and 100 g/ml streptomycin sulfate at 37°C in a humidified atmosphere of 95% air/5% CO 2 . Cells grown to 80 -90% confluence were transfected with 10 g of pcDNA3-␤ 2 AR-GFP using 65 l of Lipo-fectAMINE reagent (Life Technologies, Inc.) per T75 flask, according to the manufacturer's instructions. Briefly, HeLa cells were incubated with a DNA/LipofectAMINE mixture for 3-4 h, the media were replaced, and the cells were analyzed 48 h after transfection. For stable transfections, 10 g of PvuI linearized pcDNA3-␤ 2 AR-GFP was used to transfect HeLa cells. Three days after transfection, cells were trypsinized, diluted, and replated in media supplemented with 1 mg/ml Geneticin (Life Technologies, Inc.). Media were subsequently replaced every 3 days with complete media containing 0.5 mg/ml Geneticin. Stable transformants were isolated approximately 2 weeks after transfection, and clonal expression was confirmed by examining cells grown on coverslips by fluorescent microscopy.
cAMP Assays-cAMP assays were performed using CHW cells (provided by R. Lefkowitz), a hamster fibroblast line that does not express endogenous ␤ 2 ARs. CHW cells were transfected using a modification of an adenovirus-assisted transfection procedure (17). Briefly, harvested cells in Dulbecco's modified Eagle's medium containing 2% fetal bovine serum were mixed with 40 g/ml DEAE-dextran, 100 l of replicationdefective adenovirus (GPT-Ad5; a gift from P. Garcia), and 2 g of plasmid DNA. Transfected cells were passaged the next day onto 12well plates, and experiments examining agonist-mediated cAMP production were performed 5 days after transfection. Cells were washed with phosphate-buffered saline (PBS) and stimulated at 37°C for 10 min with 500 l of PBS containing 300 M ascorbic acid, 1 mM isobutylmethylxanthine, and no addition (basal), 10 Ϫ11 -10 Ϫ4 M (Ϫ)-isoproterenol, or 100 M forskolin. Reactions were stopped by placing the plates on ice, aspirating the media, and adding 500 l of ice-cold ethanol. The contents of each well were collected, lyophilized, resuspended, and assayed for cAMP content by radioimmunoassay using [ 125 I]cAMP (NEN Life Science Products) and anti-cAMP antibody (a gift from M. Ascoli) as described previously (18). ␤ 2 AR Binding and Ligand Competition Assays-For determination of ␤ 2 AR density in control and transfected cells, whole cells were harvested with trypsin/EDTA, or membranes were prepared (for competition binding assays). Whole cells were incubated in PBS containing 200 pM [ 125 I]iodopindolol (NEN Life Science Products, 2200 Ci/mmol) Ϯ 10 M (Ϫ)-alprenolol at 37°C for 1 h as described previously (18). Competition binding studies on cell membranes were performed as described previously (18). Briefly, cells were collected into cold homogenization buffer (25 mM Tris, pH 7.5, 5 mM EDTA, 1 mM EGTA, 0.02 mg/ml leupeptin, 0.2 mg/ml benzamidine, 0.5 mM phenylmethylsulfonyl fluoride) and homogenized by Polytron disruption. Homogenates were centrifuged, washed, resuspended, and assayed. Approximately 5 g (transfected cells) or 50 g (untransfected cells) of membrane protein were incubated at 37°C for 1 h with 30 pM [ 125 I]iodopindolol in the presence of 10 Ϫ11 -10 Ϫ4 M (Ϫ)-isoproterenol. For down-regulation studies, cells were treated with isoproterenol for various times, washed, and homogenized in lysis buffer (20 mM Tris, pH 8, 5 mM EDTA, 2 mM EGTA, 5 g/ml leupeptin, 0.2 mg/ml benzamidine) using a Polytron (2 ϫ 30 s at 25,000 rpm), and approximately 50 g of lysate were incubated at 37°C for 1 h with 1 nM [ 125 I]iodopindolol. All binding reactions were terminated by the addition of 5 ϫ 4 ml of ice-cold 25 mM Tris, pH 7.5, 2 mM MgCl 2 followed by filtration through Whatman GF/C filters using a Brandel cell harvester.
Fluorescence Microscopy and Single Cell Time Courses-Fluorescence microscopy was performed on a Bio-Rad MRC-Zeiss Axiovert 100 confocal microscope (Hemmelholsteadt, UK), using a Zeiss Plan-Apo 63 ϫ 1.40 NA oil immersion objective. Cells were grown on glass coverslips and mounted on an imaging chamber (Warner Instrument Corp) with an inlet port through which media and drugs could be perfused. The media used for microscopy did not contain phenol red or antibiotics. For time course studies, the temperature was maintained at 37 Ϯ 1°C using an adjustable warm air flow and digital temperature probe (Yellow Springs Instrument Co., Inc.). For labeling of the lysosomal compartments, cells were incubated overnight with 1 mg/ml rhodamine-labeled dextran (Molecular Probes, Eugene, OR) on glass coverslips. The dextran was washed out of cells by rinsing once with media and then placing fresh media on cells 1.5 h before imaging. Agonists were added to the media and/or rhodamine-dextran mixes, depending on the treatment. For the transferrin experiments, the cells were incubated in media lacking serum for 30 min, followed by incubation with 20 or 200 g/ml rhodamine-labeled transferrin for 30 -60 min at 37°C before imaging. The rhodamine-transferrin was only briefly rinsed off of cells three times with PBS before imaging. Incubations were timed so that agonist exposure end points coincided with the rhodamine-dextran or -transferrin loading and washing procedure end points. Cells for the rhodamine colocalization studies were rinsed quickly three times with PBS, incubated for 10 min at room temperature in 3.7% formaldehyde to fix, rinsed again in PBS, and mounted on a microscope slide with Slowfade mounting medium (Molecular Probes) before imaging by confocal microscopy.

Pharmacological and Functional Properties of ␤ 2 AR-GFP-
The complete GFP-S65T open reading frame was directly fused to the carboxyl terminus of a flag-tagged ␤ 2 AR. Preliminary experiments were then performed to establish the suitability of ␤ 2 AR-GFP as a model. Radioligand binding experiments using HeLa cells transiently expressing the receptor constructs demonstrated pharmacological properties of ␤ 2 AR-GFP that were similar to those observed for the wild type ␤ 2 AR. Competition of [ 125 I]iodopindolol binding with isoproterenol revealed comparable IC 50 values (ϳ50 nM) measured using cells expressing either ␤ 2 AR-GFP or wild type ␤ 2 AR (Fig. 1A). Similar results were also obtained by Barak et al. (16), although calculated K i values for both ␤ 2 AR-GFP and wild type ␤ 2 AR expressed in HEK293 cells were considerably higher, perhaps reflecting differences in receptor-G protein coupling between the respective model cells.
The capacity of ␤ 2 AR-GFP to promote agonist-mediated cAMP production was also examined (Fig. 1B). CHW cells, which lack endogenous ␤ 2 ARs, were transiently transfected with either wild type ␤ 2 AR or ␤ 2 AR-GFP, and the dose-dependent response to isoproterenol was examined. Isoproterenol was both efficacious (maximal cAMP reached approximately 2.5fold basal) and potent (EC 50 ϭ ϳ1-2 nM) in stimulating cAMP production in cells expressing ␤ 2 AR-GFP. Moreover, ␤ 2 AR-GFP responsiveness was similar to that observed for the wild type ␤ 2 AR. Collectively, these data suggest that fusion of the 238-amino acid GFP protein to the carboxyl terminus of the ␤ 2 AR does not alter the salient pharmacological or functional properties of the receptor.
Assessment of agonist-mediated internalization of ␤ 2 AR-GFP was performed in HeLa cells stably expressing ␤ 2 AR-GFP at ϳ200 fmol/mg protein (Fig. 1C). Cells were pretreated with 10 M isoproterenol for 0 -60 min, and cell surface ␤ 2 AR density was subsequently assessed using the hydrophilic ␤-antagonist [ 3 H]CGP-12177. The results demonstrate a classical timedependent sequestration of ␤ 2 ARs. Within 15 min of agonist treatment an ϳ30% loss of cell surface receptors was observed, reaching ϳ50% by 60 min. These results are also comparable with those observed by Barak et al., in which flow cytometry was used to measure agonist-mediated sequestration (16). In addition, these results imply that mechanisms involving the acute trafficking events of the wild type ␤ 2 AR (3) appear applicable to ␤ 2 AR-GFP. Indeed when transiently expressed in COS-1 cells, agonist-mediated ␤ 2 AR-GFP internalization was enhanced by co-expression of ␤-arrestin (data not shown). These results suggest that ␤-arrestin is capable of binding to the ␤ 2 AR-GFP and mediating its internalization, as has been observed with the wild type ␤ 2 AR (2, 3).
Alterations in cellular ␤ 2 AR content following chronic exposure to ␤-agonist were subsequently determined in HeLa cells (Fig. 1D). Cells were treated with 10 M isoproterenol for 0 -24 h, and total ␤ 2 AR density was measured in crude cell lysates by radioligand binding using the hydrophobic ␤-antagonist [ 125 I]iodopindolol. As with the sequestration data, the temporal profile of ␤ 2 AR-GFP down-regulation was typical of that observed for wild type ␤ 2 AR expressed in various cell systems (19,20).
Visualization of Time-dependent Trafficking of ␤ 2 AR-GFP-Having established ␤ 2 AR-GFP as an appropriate model of ␤ 2 AR function and trafficking, we subsequently performed experiments designed to visualize the subcellular distribution of ␤ 2 AR-GFP following acute and chronic exposure to ␤-agonist. These experiments were performed using HeLa cells in which stable transfection of ␤ 2 AR-GFP resulted in expression levels of 200 -700 fmol/mg protein. Examination of numerous cells suggests that the ␤ 2 AR-GFP is diffusely distributed on the cell surface before agonist treatment (seen in Fig. 2, left panels in top and bottom rows). However, significant perinuclear staining was also visible in some cells (e.g. Fig. 2, bottom row). Cyclohexamide treatment reveals that although some of the perinuclear staining is likely due to Golgi localization of receptors, some receptors are also in a presently unknown cellular compartment.
Initial studies examined the rapid internalization of ␤ 2 AR- FIG. 1. A, competition binding in transiently transfected HeLa cells. Wild type ␤ 2 AR and ␤ 2 AR-GFP were expressed transiently in HeLa cells using LipofectAMINE to levels of ϳ2 and 3.5 pmol/mg membrane protein, respectively (25-40-fold above endogenous ␤ 2 AR levels in HeLa cells). Binding experiments contained 30 pM [ 125 I]iodopindolol, which was competed with the indicated concentrations of isoproterenol. Results represent duplicate binding assays. B, cyclic AMP generation in CHW cells. Receptors were expressed transiently in CHW cells as described under "Experimental Procedures," and the cells were then treated with various concentrations of isoproterenol for 10 min at 37°C. cAMP was assayed by radioimmunoassay and was compared with receptor-independent cAMP levels generated by incubation with 10 M forskolin. Results represent the means Ϯ S.E. of two experiments performed in duplicate. The wild type ␤ 2 AR and ␤ 2 AR-GFP were expressed at ϳ600 and 500 fmol/mg protein, respectively, likely accounting for the slightly higher response of the wild type receptor. C, agonist-promoted sequestration of ␤ 2 AR-GFP. HeLa cells stably expressing ␤ 2 AR-GFP (ϳ200 fmol/mg protein) were treated for the indicated times with 10 M isoproterenol, and binding to the hydrophilic ligand [ 3 H]CGP-12177 was assessed. Binding at various time points was compared with the unstimulated cells. Results represent the the means Ϯ S.E. of four experiments performed in triplicate. D, down-regulation in HeLa cells stably expressing ␤ 2 AR-GFP. HeLa cells stably expressing ␤ 2 AR-GFP (ϳ200 fmol/mg protein) were treated for the indicated times with 10 M isoproterenol, the cells were washed and lysed in a hypotonic buffer, and binding to the hydrophobic antagonist [ 125 I]iodopindolol was assessed. Results represent the the means Ϯ S.E. of three to six experiments performed in triplicate.
GFP following exposure of HeLa cells to 10 M isoproterenol (Fig. 2). Fluorescent images obtained from single cells using confocal microscopy demonstrate the rapid appearance of a punctate staining pattern with accumulations that become progressively larger and more numerous over a 20-min course of agonist exposure (Fig. 2, top row). These images are consistent with the time course of sequestration suggested by our radioligand binding data (Fig. 1C) as well as with those studies utilizing immunocytochemistry of fixed and permeabilized cells to visualize internalization of epitope-tagged ␤ 2 ARs (3,21).
We then observed cells in which a 10-min agonist treatment was followed by media washout and exposure to the ␤ 2 AR antagonist alprenolol (Fig. 2, lower panels). Antagonist was included in the wash to prevent agonist rebinding during the washes. The punctate localization pattern of the receptor was shown to revert to a more diffuse pattern within 20 min of agonist removal, suggesting that receptors are recycling back to the plasma membrane. Radioligand binding experiments in transiently transfected HeLa cells indicate that the return of the ␤ 2 AR-GFP to the cell surface after agonist removal is temporally and quantitatively similar to that of the wild type ␤ 2 AR (Fig. 3). However, complete recycling of internalized receptor was not observed until 60 min after agonist washout. Our findings agree with other pharmacological studies, suggesting that the receptor relocalizes to the plasma membrane upon removal of agonist (7), and with results from immunocytochemistry experiments using fixed cells (21). The ability to observe this process in live cells should provide unique insight into receptor recycling, and what signals, if any, affect this dynamic process.
We next examined the localization patterns of ␤ 2 AR-GFP induced by exposure to ␤-agonists of differing pharmacological properties (Fig. 4). We hypothesized that the rate of observable punctate pattern formation would correlate with the intrinsic activity and/or onset of action of the various ␤-agonists tested. The effect of formoterol, a long acting ␤-agonist of high intrinsic activity and rapid onset of action, on ␤ 2 AR-GFP redistribution is depicted in Fig. 4A. The rate and pattern of vesicular formation were comparable with that induced by isoproterenol. Conversely, albuterol, a short acting agonist with moderate intrinsic activity, exhibited a slightly slower rate of receptor relocalization than that observed with either isoproterenol or formoterol (Fig. 4B). Multiple experiments suggest that cells exposed to albuterol required approximately 25-30 min to reach a pattern of ␤ 2 AR-GFP distribution caused by a 20-min exposure to isoproterenol (Figs. 2 and 4 and data not shown).
Most striking, however, were the data obtained using salmeterol, a long acting ␤-agonist with low instrinsic activity and a very slow rate of onset of action, characteristics determined in part by the very hydrophobic nature of the compound (22,23). Cells exposed to salmeterol (Fig. 4C) required ϳ70 min of treatment to exhibit a level of receptor internalization comparable with that observed at the 20-min isoproterenol time point. Yet it was possible to show that salmeterol clearly caused relocalization of receptors to intracellular vesicles, providing evidence of salmeterol-induced sequestration not obtainable in previous studies (23,24). Because treatment of cells with salmeterol causes stable activation of adenylyl cyclase that survives extensive wash procedures and sucrose-gradient purification of plasma membrane fractions (23), radioligand binding is rendered an unreliable tool for the analysis of salmeterol effects on ␤ 2 AR trafficking. Here we demonstrate that direct visualization of ␤ 2 AR-GFP distribution circumvents the limitations conferred by salmeterol retention and represents an important tool in future analyses of this compound.
Colocalization of ␤ 2 AR-GFP with Rhodamine-labeled Transferrin and Dextran-To assess differences in subcellular local-ization of ␤ 2 AR-GFP following acute (sequestration) versus chronic (associated with down-regulation) exposure to ␤-agonists, we examined the time-dependent colocalization of ␤ 2 AR-GFP with additional fluorescent compounds (rhodamine-labeled transferrin and dextran) known to accumulate in distinct subcellular compartments. Transferrin primarily internalizes with transferrin receptors and constitutively recycles with the receptors through early endosomes to a recycling compartment and then back to the cell surface (25). Conversely, dextran has been shown to specifically accumulate in late endosomes and lysosomes (26). Previous studies have demonstrated co-localization of transferrin receptors and an epitope-tagged ␤ 2 AR following ␤-agonist treatment (21), whereas dextran has recently been used to demonstrate lysosomal localization of the thyrotropin-releasing hormone receptor (27). Fig. 5A demonstrates that in unstimulated cells, the ␤ 2 AR-GFP distribution displays a relatively diffuse membrane localization pattern, whereas 20-min incubation of these cells with transferrin results in accumulation of transferrin in small vesicles. 30 min after treatment with isoproterenol, a significant portion of ␤ 2 AR-GFP is seen to distribute into early endosomal compartments coincident with the presence of transferrin (Fig. 5B,  colocalization shown in yellow). In contrast, when cells are exposed to isoproterenol for 3.5 h followed by a 1-h treatment with transferrin in the absence of ␤-agonist, minimal co-local- ization of ␤ 2 AR-GFP and transferrin is evident (Fig. 5C). These results demonstrate that under such conditions a large fraction of the ␤ 2 AR-GFPs remain in internal vesicles that lack transferrin receptors, possibly in late endosomes and/or lysosomes.
The agonist-and time-dependent localization of ␤ 2 AR-GFP with lysosomes is revealed in cells loaded with rhodaminelabeled dextran. Cells were incubated with 1 mg/ml rhodaminelabeled dextran for 24 h and then washed for 1.5 h in the absence of dextran to remove any accumulation in early endosomes. During the latter portion of these incubations the cells were also incubated with 10 M isoproterenol for 30 min, 1 h, 3.5 h, or 24 h prior to fixing the cells. Unstimulated cells display dextran localized to large vesicles typical of lysosomes, many of which are centrally located in the cells (Fig. 6, upper  left). Following 30 min of isoproterenol exposure, the ␤ 2 AR-GFP has moved into early endosomes and shows minimal colocalization with dextran. After 1 h of isoproterenol treatment, the ␤ 2 AR-GFP exhibits significant co-localization with dextran, with this colocalization increasing somewhat at the 3.5-and 24-h time points. Cells treated in a time-dependent manner with salmeterol displayed a similar pattern, although the progression of colocalization of ␤ 2 AR-GFP with dextran was slower. Although salmeterol treatment for 24 h results in images similar to those obtained with isoproterenol treatment, colocalization of dextran with ␤ 2 AR-GFP was not evident until after 3 h of treatment with salmeterol (data not shown), suggesting that the relatively slow kinetics of salmeterol binding/ activation of the ␤ 2 AR translate into attenuated rates of ␤ 2 AR sequestration and down-regulation.
In conclusion, we have utilized ␤ 2 AR-GFP to visualize real time cellular redistribution of ␤ 2 ARs in live cells responding to agonists. We have demonstrated the rapid colocalization of the ␤ 2 AR in transferrin-containing endosomes following acute ␤-agonist exposure. Following prolonged (but not acute) exposure to ␤-agonists, ␤ 2 AR-GFP is shown to colocalize with dextran in lysosomes. Experiments examining the effects of various ␤-agonists suggest that ␤ 2 AR distribution, sequestration, and down-regulation are regulated by the intrinsic activity and onset of action of a ligand. In this regard, the ␤ 2 AR-GFP was especially advantageous for the examination of the lipophilic compound salmeterol. Our results suggest that GFP conjugated G protein-coupled receptors are powerful tools for visualizing the dynamics of receptor trafficking in living cells.